Citation
Aversive learning, individual differences, and psychophysiological response

Material Information

Title:
Aversive learning, individual differences, and psychophysiological response
Creator:
Greenwald, Mark Kenneth, 1960-
Publication Date:
Language:
English
Physical Description:
vii, 150 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Acceleration ( jstor )
Associative learning ( jstor )
Galvanic skin response ( jstor )
Habituation ( jstor )
Learning ( jstor )
Mental stimulation ( jstor )
Sensitization ( jstor )
Serenity ( jstor )
Startle reflex ( jstor )
Temperament ( jstor )
Aversion therapy ( lcsh )
Psychophysiology ( lcsh )
Startle reaction ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1992.
Bibliography:
Includes bibliographical references (leaves 138-149).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Mark Kenneth Greenwald.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
001806273 ( ALEPH )
AJN0106 ( NOTIS )
27804064 ( OCLC )

Downloads

This item has the following downloads:


Full Text










AVERSIVE LEARNING, INDIVIDUAL DIFFERENCES, AND
PSYCHOPHYSIOLOGICAL RESPONSE




By

MARK KENNETH GREENWALD


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1992












ACKNOWLEDGEMENTS

Although my graduate training has taken me through several distinct

phases, several people have been my constant supporters, mentors and

friends. Foremost among these is Peter Lang--whose role as my supervisor
and model researcher have made an indelible mark on the way I perceive

the broad field of psychology, and how I will conduct my future research.
I thank Peter profusely for helping me to do my best.
Two equally important and outrageously talented people, Margaret

Bradley and Bruce Cuthbert, provided daily inspiration and guidance, and
served as constant reminders of my own humility. For their patience and
willingness to listen to my problems, and for the myriad times they came to

my rescue over the years, I am deeply indebted to them.
I have also been very fortunate to have two friends whose collaboration

on various projects made me appreciate their growing talents. To Alfons
Hamm, whose early teamwork eventually resulted in this dissertation, and

to Christopher Patrick, with whom I have shared countless stimulating

discussions--I am very thankful. Other scientists whose ideas have also

excited my imagination, and with whom the Lang laboratory has been a great

place to work, are Ed Cook, Mark McManis, Margaret Petry, Ellen Spence,

Scott Vrana, and David York.
During my training, my committee members have been most helpful

and supportive of my work. Keith Berg is a marvelous role model for his

keen methodological eye and technical abilities; he also taught me many
fundamentals of psychophysiology. Ira Fischler and Michael Levy were








instrumental in helping me to negotiate the ever-complicated terrain of

contemporary cognitive psychology. Jim Cauraugh, always pleasant and

enthusiastic, coached me in my emerging interest in the cognition of motor

control.

Finally, the hectic pace of doctoral work was actually enjoyable because

three incredibly dear people kept me sane and happy. My loving wife and

companion, Margaret, and my devoted parents, David and Rosemarie, have

been my cheerleaders and saviors throughout my education. The golden

opportunities they have given me help put all this work in the proper

perspective.













TABLE OF CONTENTS


ACKNOWLEDGEMENTS ........................................ ii

ABSTRACT ......... ...... ..................................... vi

INTRODUCTION. ............................................. 1

The Startle Probe in the Assessment of Emotion ................. 1
Aversive Learning........................................... 3
Sensitization .......... ............... .................. 4
Associative Learning ..................................... 5
Individual Differences .................................. 10
Statement of the Experimental Problem ........................ 13

METHODS ..................................................... 20

Subjects ............... ........ ......... ................. 20
Stimulus Materials ................... ..................... ... 20
Apparatus and Psychophysiological Recording ................... 22
Scoring and Definitions of Physiological Response Measures ...... 23
Procedure ................................................. 24
Design and Data Analysis ...................................... 31

RESULTS ..................... .............................. 36

Sensitization .................................................. 36
Startle Probe Response.................................. 36
Visceral Reactions ........................................ 40
Summary .......................................... ..... 42
Associative Learning................................... ... 42
Affective Ratings ..................... ................. 43
Startle Probe Response ................................... 47
Visceral Reactions ...................................... 54
Summary ........................... ................... 61
Individual Differences ......................................... 61
Temperament Characteristics .............................. 61
Rated Shock Aversiveness ............................. 68
Sum m ary ............................ ................... 75








D ISCU SSIO N .................................. .................. 76

Sensitization .................................................. 76
Startle M agnitude ......................................... 76
Methodological Considerations ............................ 78
Associative Learning .......................................... 80
Habituation and Sensitization in Associative Learning ....... 81
Affective Associative Change .............................. 85
Temporal Specificity and Preparatory Processes .............. 87
Brain-Behavior Relationships in Associative Learning ....... 90
Individual Differences ....................................... 91
Aversive Reactivity and Sensitization ...................... 91
Associative Learning ...................................... 93
Summ ary ................................................... 96

APPENDIX A STIMULUS SELECTION ............................... 98

APPENDIX B PROCEDURAL INSTRUCTIONS. .................. 103

APPENDIX C QUESTIONNAIRES ............................... 109

APPENDIX D AUXILIARY DATA TABLES....................... 113

REFERENCES .................................................... 138

BIOGRAPHICAL SKETCH ...................................... 150












Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

AVERSIVE LEARNING, INDIVIDUAL DIFFERENCES, AND
PSYCHOPHYSIOLOGICAL RESPONSE

By

Mark Kenneth Greenwald

August, 1992


Chairman: Peter J. Lang, Ph.D.
Major Department: Psychology

In this experiment, a psychophysiological analysis of human aversive

learning was conducted. There were three primary goals in this study. First,

sensitization of the eyeblink reflex to an acoustic startle probe (95 dBA noise

burst) and visceral activation (skin conductance, heart rate) following electric

shock exposure were examined. Subjects were randomly assigned to two

groups, which were exposed to shock either before an adaptation series of

two pictures or midway through this series. To control for unrelated effects,

each group completed questionnaires while the other was exposed to shock.
Shock exposure significantly increased startle magnitude relative to control

conditions, whereas visceral activity was not influenced.

Second, the degree to which the startle reflex, visceral activity, and

affective ratings index associative learning was tested, using differential

aversive conditioning. The series of two pictures continued and one slide

was paired with shock. Associative learning potentiated startle reflex and








electrodermal responses, and generated more cardiac acceleration and rated
aversiveness of the shocked slide (relative to the nonshocked slide). This
experiment also examined whether pairing shock with different affective
slides would yield variable degrees of associative change. Stimulus selection
was based on each subject's affective ratings, not specific content. Slides

varied in pleasantness and arousal across four groups. The results, however,

did not support the hypothesis that the affective foreground constrains the

degree of associative modulation.

Third, this study explored whether individual differences modulate
measures of aversive learning or reactivity to the prepotent stimuli used

here (i.e., shock, startle probe). Subject temperament dimensions, as well as

tolerance and rated aversiveness of the shock, were examined. Although
temperament characteristics were not generally related to learning, subjects

who were more fearful and inhibited exhibited greater reactivity to aversive

stimulation (i.e., larger startle responses, less shock tolerance, and more
reported fearfulness). In addition, subjects who rated shock as more aversive

demonstrated greater learning in startle reflex, electrodermal, and verbal

report measures. A cognitive learning model, in which representations of

the shock and conditioned stimulus are connected to the brain's aversive
motivational system, is used to explain the present results.













INTRODUCTION

The startle probe reflex is reliably augmented when ongoing processing
involves aversive information. This affect-startle effect, repeatedly found in
aversive learning experiments with animal subjects (Davis, 1989a), now
forms the centerpiece for a broader theory of human emotion (Lang, Bradley,
& Cuthbert, 1990, 1992). In this dissertation experiment, a psychophysio-
logical analysis of human aversive learning is conducted. First, sensitization
of the startle reflex and measures of visceral activation (skin conductance,
heart rate) during electric shock exposure (i.e., nonassociative learning) are
examined. Second, the degree to which the startle probe reflex, visceral
activity, and affective judgments index associative learning, through pairing
photographic slides with shock, is tested. Third, the present experiment will
explore the extent to which individual differences modulate psychophysio-
logical indicators of these two forms of aversive learning.
The Startle Probe in the Assessment of Emotion

The startle reflex is an obligatory motoric response to a rapid-onset,
relatively intense, exteroceptive (e.g., acoustic) probe. Although the startle
response can be a generalized bodily reaction (Hunt & Landis, 1936), the
magnitude and latency of the eyeblink to the probe are the characteristics
most typically measured in experiments using human subjects (Anthony,
1985). This reflex occurs in many species and has been most widely
investigated in animals (e.g., the whole body response in rats), for whom the
neurobiological foundations of the primary startle circuit are now well
understood (Davis, Gendelman, Tischler, & Gendelman, 1982).








There is an additional and highly valuable use for this reflex behavior:

The blink-eliciting probe may be introduced unpredictably during

information processing (e.g., perception, mentation, learning), and changes

in the magnitude or latency of the probe reflex in these contexts (relative to

control conditions) permit one to infer the disposition of the organism. In

fact, the affective value of environmental stimuli has been repeatedly found

to modulate the magnitude of the startle probe reflex. Specifically, the

magnitude of the startle probe reflex is largest when perceiving aversive

pictures, relatively smaller during neutral pictures, and smallest during
pleasant pictures (Bradley, Cuthbert, & Lang, 1990a, 1990b, 1991; Greenwald,

Bradley, Cuthbert, & Lang, 1990; Vrana, Spence, & Lang, 1988). Similarly,

startle probe response is greater when imagining aversive scenes than

neutral or pleasant events (Cook, Hawk, Davis, & Stevenson, 1991; Cuthbert,

Bradley, York, & Lang, 1990; Vrana & Lang, 1990). Since this valence-reflex

relationship does not depend on sensory processing, it suggests that startle

modulation indexes a central motivational state.

A recent theoretical model offers a framework for interpreting this

affect-startle effect (Lang et al., 1990). In this view, emotions are conceived

as tendencies to act, represented in memory as an associative network of

context-specific perceptual and response information. A postulate of this

model is that all affective representations are fundamentally appetitive or

aversive, reflecting action dispositions of approach or avoidance (see also

Dickinson & Dearing, 1979; Konorski, 1967). Unpleasantness is associated

with the bias to avoid/withdraw, whereas pleasantness is associated with

approach/seeking behavior. Both extrinsic (environmental) and intrinsic

(subject) variables contribute to the formation of this centrally organized

response set. The efferent system, which includes verbal, overt motor (e.g.,








reflex), and visceral outputs, is motivationally tuned by this central set to
respond in accordance with ongoing processing. Physiological measures

can, therefore, be used to index this response organization in real time.
Using an affective matching principle, Lang et al. (1990) argue that
defensive behaviors (e.g., the startle response to an aversive probe) are
facilitated when the processing task also occasions an aversive state (in

which case a "match" occurs) and inhibited when the situation is pleasant
(a "mismatch"). The affect-startle effect thus reflects the biphasic conceptual
organization of emotional valence. Startle augmentation during aversive
processing thus provides a compelling link to previous dimensional
accounts of emotion, all of which emphasize the direction (valence) and

intensity (arousal) parameters of behavior (Duffy, 1972; Hebb, 1949;
Mehrabian & Russell, 1974; Osgood, Suci, & Tannenbaum, 1957; Schlosberg,

1952; Schneirla, 1959; Tellegen, 1985).
Aversive Learning

Since this study focuses on learning, a clarification of terms is first

presented. Here, learning is defined as a hypothetical process inferred from
measurable behavioral changes which result from specific experiences

(Mackintosh, 1983). Learning processes have been subdivided as either

nonassociative, e.g., within-session habituation and sensitization, or

associative, e.g., classical and operant conditioning. In this experiment,

sensitization and associative learning (via classical conditioning) are the

main subjects of interest, although habituation is also addressed as a

secondary issue. Both types of learning studied here are further
characterized as aversive by the type of stimulus which will be used to

modify behavior: electric shock. Sensitization will be inferred when a

response increment occurs to shock exposure in the absence of an








association, whereas associative learning will be inferred from response
increments which result from explicit pairings of a stimulus with shock.

Sensitization
It is well established that relatively intense stimulation can produce
response increments which reflect sensitization. Studies of the hindlimb

flexion reflex in the spinal cat (Thompson & Spencer, 1966) and siphon-

withdrawal reflex in Aplysia (Hawkins & Kandel, 1984) have shown that

repeated tactile stimulation can result in an initial increase in reflex

magnitude (sensitization), followed by decreased response (habituation)--

especially if stimulation is more intense. This finding led to the develop-

ment of "dual-process theory" (Groves & Thompson, 1970; Petrinovich,

1984), which proposes that sensitization and habituation are independent
neural processes, the net activation of which determines response output
at any moment. Dual process theory--which relates neurophysiological
mechanisms of habituation and sensitization to molar theories of behavior--
presumes that habituation occurs in the stimulus-response pathway,

whereas sensitization occurs because of tonic changes in the state of the
organism, which then modulate the stimulus-response pathway.

Startle modulation. Sensitization of the acoustic startle response has
been observed in several contexts. First, early habituation experiments
showed that an intense startle probe can, by itself, elicit transient reflex

sensitization prior to habituation (Davis, 1972; Groves & Thompson, 1970;
Szabo & Kolta, 1967). A second type of sensitization has been identified, in
which increasing intensity of constant background noise enhances the probe

reflex response (Davis, 1974a, 1974b; Hoffman & Searle, 1965; Ison &

Hammond, 1971; but see Putnam, 1975, for a qualification of this effect).




5


Third, and most recently, Davis and his colleagues have convincingly

shown that the startle reflex is sensitized following exposure to electric shock
(Boulis & Davis, 1989; Davis, 1989b; Hitchcock, Sananes, & Davis, 1989).

Davis (1989b) showed that sensitization (1) is attainable with a single shock,

but the effect increases with 5 or 10 shocks, (2) develops quickly, and (3)

cannot be attributed to rapid association with the experimental context. It is

proposed here, as Davis suggested, that this effect represents the action of

unconditioned properties of shock which can support aversive conditioning.

Therefore, sensitization appears to be a building block of conditional

learning (Hawkins & Kandel, 1984). Sensitization could also be interpreted

as a type of state transition in which response codes activated by stimulus

exposure become eligible for associative learning (cf. Sutton & Barto, 1981).

Konorski (1967) developed a similar cognitive model of learning,

proposing that exposure to an unconditioned stimulus (US) leads to an
internal sensory representation of the US, which then activates a central

motivational state related to the affective quality of the US (Figure 1). This

motivational state produces preparatory behavior which reflects the affective

character of the stimulus (e.g., withdrawal from aversive stimulation). This

account suggests why the startle reflex--a protective reaction--should be

sensitized after shock exposure. According to the affective matching

principle (Lang et al., 1990), shock exposure activates the aversive system (via

the US representation) and should prime related defensive behaviors.

Associative Learning

The mechanism of classical conditioning, Konorski (1967) argued,

involves a simple extension of the cognitive events described for

sensitization. Following repeated presentation and pairing of the

conditioned stimulus (CS) with the US, the internal representation of the CS














US -


unconditioned
responses,
e.g., reflexes


aversive preparatory responses,
e.g., freezing, sweating, cardiac
acceleration, hyperventilation,
facial expression, reports of fear


Figure 1. Cognitive model of aversive learning (adapted from Dickinson &
Dearing, 1979). Presentations of the CS and aversive US lead to internal
representations of each stimulus (shaded circles). The US activates (via
pre-existing connections) the aversive motivational system which, in turn,
generates associated behaviors. Reinforcing the CS with the US generates
learned connections with the US representation and aversive motivational
system. Thus, direct US exposure (as in sensitization), and CS-US pairings
(as in associative learning) can both potentiate aversive behavior.


Aversive
Motivational
System


preexisting
connection
_ learned
connection








develops independent excitatory connections with both the US
representation and the relevant motivational system (Figure 1). Due to its

association with the motivational system, the CS representation comes to

activate preparatory behavior related to the affective quality of the US.

Konorski's theory not only deals with connections and representation

but, unlike many extant cognitive models of associative learning (e.g., Estes,
1973; Frey & Sears, 1978; Kamin, 1969; Kehoe, 1988; Pearce & Hall, 1980;

Rescorla & Wagner, 1972; Wagner, 1978), also explicitly models excitatory

and inhibitory interactions between hedonic states. In this framework (see

also Dickinson & Dearing, 1979; Wagner & Brandon, 1989), aversive and

appetitive motivational systems--which comprise the directive forces of
adaptive behavior--are assumed to interact via reciprocal inhibitory

connections. Since conditioning procedures modulate affective stimulus

associations, they can be used to transform the affective meaning of a CS

(e.g., by pairing an appetitive stimulus with shock) and, thus, demonstrate
the operation of these opponent systems. Traditionally, these aversive-
appetitive interactions have been measured in animals using conditioned

suppression (e.g., Dickinson, 1977; Rescorla, 1971). However, use of startle

reflex modulation affords a number of advantages in evaluating affective

associative learning, including its independence from operant behavior.

Startle modulation. Experiments using animal subjects have shown

that the startle reflex can be augmented by aversive classical conditioning,

i.e., the "fear-potentiated startle" effect. During training, a previously

neutral light CS is paired with an aversive electric shock. In subsequent test

trials (in which shock ceases), an acoustic probe is presented during

presentation of the light cue, and in the absence of the light. In these studies,

fear-potentiated startle is defined as the within-subject difference in startle








amplitude during the cue versus its absence (Davis, 1986). Numerous
investigators have demonstrated this effect (Berg & Davis, 1984, 1985; Brown,
Kalish, & Farber, 1951; Cassella & Davis, 1986; Davis & Astrachan, 1978; Kurtz
& Siegel, 1966; Leaton & Borszcz, 1985).
However, few experiments have been conducted using human subjects.
The first studies which examined potentiated startle in humans were
originally completed about 30 years ago, and were primarily concerned with
the temporal delay of CS/probe presentation on reflex potentiation. One
conclusion from this early research (Ross, 1961; Spence & Runquist, 1958),

as well as some recent related work (Vrana et al., 1988), is that probe blink
response to an aversive cue is less reliably potentiated when the probe
follows stimulus onset by relatively short intervals (e.g., 500 msec). The

selective absence of reflex potentiation at short time lags suggests that
prepulse inhibition--an attentional mechanism--may obscure measurement
of fear induction (Davis, Schlesinger, & Sorenson, 1989).

Recently, Hamm, Greenwald, Bradley, and Lang (1991) reintroduced the
fear-potentiated startle paradigm with human subjects, suggesting that the
hiatus in experimentation (despite Konorski's enlightened views) may have

been due to a disinterest in conditioning during the "cognitive revolution."
However, the basis for understanding the relationship between learned fear
and startle potentiation was also complicated by Hull's (1943) explanatory use
of the motivational construct drive, which was presumed to have a broad
energizing effect on behavior (and is therefore analogous to arousal rather
than valence). According to this position, pairing a light with aversive
shock increased organismic drive, which then led to the increase in startle

magnitude. However, early animal studies failed to show augmented startle
effects when appetitive reinforcement was used: Probe startle responses









during a stimulus paired with food were unchanged (Trapold, 1962) or even
diminished (Armus, Carlson, Guinan, & Crowell, 1964; Armus &
Sniadowski-Dolinsky, 1966), relative to control conditions. Therefore, these
results argue against a drive theory interpretation of startle potentiation.
Hamm et al. (1991) thus adopted the Konorskian model to study

emotional learning. Unlike the single cue paradigm used with animals,
Hamm et al. used a differential learning paradigm, in which each subject
viewed two alternating slides during the experiment. For training, one slide
was always followed by shock, whereas the second slide was never followed
by shock. This enables within-subject control over the associative process.
Before and after training, the startle probe was presented during the slide
associated with shock (CS+), the slide never followed by shock (CS-), and
during interslide intervals. Visceral (heart rate, skin conductance) responses
and affective (valence, arousal) ratings of the stimuli were also measured.
Results indicated that blink reflexes were augmented when the startle probe
was administered during the slide that was paired with shock. Reflexes
evoked in the presence of this cue were significantly larger than those
elicited during the slide never paired with shock, or during the interval
between slide presentations. These data replicated and extended findings
from the animal literature.
A second objective of the Hamm et al. (1991) experiment was to explore
whether a priori stimulus affective properties influenced the degree of
emotional learning. Thus, different groups of subjects were assigned to view
slide contents which varied in affective valence and arousal throughout
learning. Two alternative hypotheses were evaluated. One hypothesis,
based on belongingness theory (Garcia & Koelling, 1966), predicted that better
conditioning should be obtained for slides with affective values which were








initially similar to shock. The second hypothesis, based on the "delta rule"

(Rumelhart, Hinton, & McClelland, 1986), predicted that better conditioning

should be obtained for slides with affective values which were initially

dissimilar to shock. Results indicated that the degree of startle potentiation

varied linearly with the a priori affective valence of the slide contents: Slides

of unpleasant content (i.e., mutilated bodies, threatening animals) yielded

significantly less startle potentiation than slides depicting pleasant content

(i.e., nature scenes, attractive nudes). Conditioned startle potentiation thus

reflected changes in hedonic value (i.e., pleasure). In contrast, the degree of

skin conductance discrimination in extinction varied with the a priori level

of arousal of the conditioned stimulus: Slides depicting arousing content

(i.e., attractive nudes, mutilated bodies) yielded significantly less conductance

discrimination than slides of calm content (i.e., household objects). These

data suggest that electrodermal differentiation reflects arousal (rather than

hedonic) change. Therefore, the results supported the delta rule, rather than

the belongingness hypothesis.
Individual Differences

In the myriad contemporary conditioning studies which have exposed

human subjects to electric shock USs (e.g., Cook, Hodes, & Lang, 1986;

Hammond, Baer, & Fuhrer, 1980; Hugdahl & Ohman, 1977; Lanzetta & Orr,

1986; McNally & Reiss, 1984), the subject is asked--for obvious ethical

reasons--to set the shock level to be moderately to highly annoying

(uncomfortable), but not painful. In other words, the final shock level is

correlated with the subject's pain tolerance. However, none of these studies

has addressed whether a subject's tolerance for shock relates to personality

characteristics, nor whether temperament or shock tolerance differences can








predict the degree of learning. Even so, there is an older set of studies and
theory which bears on this issue.
Extensive, but inconclusive, research has been conducted on the

relationship between personality and conditioning (see Levey & Martin,
1981, for a review). As noted above, much of early conditioning research was
influenced by Hull's concept of drive; this approach was wedded with the

study of individual differences in anxiety by Spence's group at Iowa in the
1950s and early 1960s. This work was not related to a theory of personality
but, rather, tested the influence of a single trait on learning. Spence (1958,
1964) demonstrated that anxiety level (which was equated with drive)
positively related to the rapidity of conditioning, especially when the subject
reported acute anxiety. Although many learning (particularly eyelid
conditioning) experiments were conducted within this framework, aversive
shock reinforcement was not employed.
In contrast to the work above, Eysenck (1967) developed a two-factor
theory of personality emphasizing extraversion-introversion and
neuroticism as central dimensions. This scheme evolved from Pavlov's

(1957) observational typology of nervous system "strength" into modern
psychophysiological variants. The basic idea is that introverts possess
weaker (i.e., less strong) nervous systems than extraverts, rendering them

more susceptible to intense, especially aversive, stimulation. Gray (1967) and
Nebylitsyn (1972) generated research testing the hypothesis that introverts
should form conditioned responses more readily than extraverts. Gray
noted that the interaction of neuroticism--which involves sensitivity to both
rewarding and punishing stimulation--and extraversion personality factors
implies that neurotic introverts should respond more than all others to








aversive stimulation. This quadrant of the two-factor space has been
associated with high trait anxiety (Gray, 1973, 1982).
Unfortunately, there is some lack of clarity in two-factor theory and

research. First, it is not clear whether anxious (neurotic introverted) subjects

should be more reactive to aversive stimulation (e.g., shocks, loud noises),
whether they should exhibit better associative learning with aversive

reinforcers (e.g., better discrimination between a stimulus paired with shock
and one not paired with shock), or both. Second, Eysenck's early research

showed that there was a positive correlation between introversion and

neuroticism, which suggests (cf. Gray, 1982) that the Eysenckian structure is

impure. Third, impulsivity (or, conversely, inhibition)--considered by these

researchers to be a secondary factor, but given primary importance by others

(e.g., Kagan, Reznick, & Snidman, 1987)--loads positively on the extraversion
dimension and, thus, may account for variance in conditioned responding
(Barratt, 1971; Eysenck & Levey, 1972).
To minimize these problems, the present experiment uses a related

alternative framework to explore individual differences in aversive
reactivity and learning. The EASI (Emotionality [fear, anger], Activity,

Sociability and Impulsivity) temperament theory of Buss and Plomin (1975)
was chosen because of its basis in heritability studies, and because the factor
structure of the EASI assessment instrument relates to two-factor theory. For
example, Eysenck's extraversion could be roughly translated into relatively
high sociability and activity in the Buss and Plomin scheme. Thus, if the

sociability and activity dimensions were to control variance in aversive

reactivity or learning, this would argue against Eysenckian theory. In fact,

the most logical expectation is that fearful and/or inhibited temperament--

analogous to neurotic introversion--would facilitate aversive responding.








Conditioning researchers have also not addressed whether perceived
aversiveness of the shock US, regardless of personality factors or shock
tolerance, relates to the amount of affective associative learning. Within the

cognitive model of learning proposed above, this perceptual variable would
appear to be rather important. To reiterate, the US representation is
presumed to develop an excitatory connection with the CS representation
after repeated pairings, which activates a central motivational state and
produces preparatory behavior which reflects the affective character of the
stimulus. Since each representation is assumed to be influenced by
perceptual factors, then a more aversive US should enhance learning.
Statement of the Experimental Problem

The flow of experimental events generally follows the procedure of
Hamm et al. (1991) described earlier. First, prior to learning, all subjects

viewed and rated a series of colored photographic slides which varied widely
in affective valence and arousal. Based on these ratings, picture stimuli for
conditioning were selected.
Second, subjects were exposed to an aversive electric shock to evaluate
its effect on response sensitization (Figure 2). To control for passage of time,

dishabituation (from the experimenter leaving/re-entering the room), and
other factors not related to sensitization, there were two groups that served
as each other's control. For one group, shock exposure occurred prior to an 8
trial adaptation series of two slides, whereas the second group was exposed to
shock midway through the adaptation series. During each group's exposure
to shock, the other group completed questionnaires to serve as a control for

shock sensitization. For both groups, startle probes were presented before
shock exposure, and during and between slide presentations throughout the
adaptation period.












































00
- '
O ^
" I,,,
....


0)















>>
in






-o





'.0

















. 4---
T3

v0




CI~
U3!

a)





in
?33 ---












a)
o ^








U)
-4-








ei
a) ^


i 4-



a)
T3
;l -4
{3 )


,uZ
_^-


3: 0

CGdu


01









t l
Ur,






-]-












)





















m j
>'









I -







0O
I-^ -


-4-


LI.0












r .
+e r 4 -









c Wo 01



092















Mo t
c In '







Ch0
0r 0
















- 00
fn CU *




aa







*0 o n





o 0 nC -





-o w vu -5


3 0 C r:














2S c- Q-
10 i c' 1-
3u -
. Qj C >
SM% {=>CT
(Ug m. "^
5..a^
.&^ 3
[!-- [- cr-















++ u
U U




( J

u "Wo 7 -N 1
- ?i g^go, t








U U

+ + ~d
(1 (A dJ






S^ -m .A



w j


(u ue

____ -i ?-






,u u S
za r. 6




+ U Cl .L; n+
cu n u cU( -








Third, the subject continued to view this unpredictable series of two
slides, which now served as the CS+ and CS- in a discriminative learning
procedure (Figure 3). In acquisition, one of the two habituated slides (CS+)
was always reinforced with shock at slide offset, whereas the other (CS-) was
not. Each slide was presented 8 times, but startle probes were not presented

during associative learning. This acquisition sequence was followed by a 16
trial extinction series, during which probes were again presented during and
between slides (8 presentations each slide). Finally, after the learning
procedure, a sequence of 12 slides varying in affective valence and arousal
(including each subject's CS+ and CS- pictures) was presented. Subjects rated

the slides as well as the perceived aversiveness of the shock US, CS+ slide,
CS- slide, and the startle probe.
There were three primary goals of this experiment. The first goal, based
upon demonstrations in the animal literature (Boulis & Davis, 1989; Davis,

1989; Hitchcock et al., 1989), was to assess the sensitizing effects of simple,
noncontingent, shock exposure on the startle probe reflex, as well as

activation in the cardiac and electrodermal systems. Using human subjects,

Greenwald et al. (1990) recently demonstrated that exposure to electric shock

significantly increased startle reflex magnitude, but that skin conductance

was not similarly augmented. However, their experiment was not explicitly

designed to test this hypothesis, e.g., the design did not control for response

increases which might have resulted from the passage of time.

In the present study, subjects were randomly assigned to two groups,
varying only in the timing of shock exposure, so that each group served as

the other's control in testing sensitization learning. It was expected that

shock exposure would increase the startle probe response in humans, much
as it does in animals. In contrast, studies with animal subjects have not








recorded autonomic activity before and after shock exposure; thus, although
electrodermal sensitization was not obtained by Greenwald et al. (1990), only
a weak hypothesis of "no effect" can be advanced. From the standpoint of

dimensional emotion theory (Lang et al., 1990), startle reflex sensitization

should reflect changes primarily in the hedonic (valence) state of the
individual, whereas sensitization of visceral response should indicate a
change in arousal level.
The second goal of this study, based upon early human aversive

conditioning experiments (Ross, 1961; Spence & Runquist, 1959) and recent

follow-up studies (Greenwald et al., 1990; Hamm et al., 1991), was to conduct
a psychophysiological analysis of shock associative learning using startle
reflex, visceral response, and affective judgment measures. The objective
here was to examine the sensitivity of each of these measures to simple

association (i.e., measured by potentiation of response to the slide paired
with shock [CS+]), and to associative discrimination (i.e., measured by
differences in response between CS+, the slide not paired with shock [CS-],
and in the absence of slides). The data generated by Hamm et al. (1991)

suggest that both startle potentiation and discrimination should occur. The
results of previous autonomic learning experiments (e.g., Hodes, Cook, &
Lang, 1985) suggest that cardiac response discrimination should primarily

occur in acquisition, and that skin conductance differentiation should be
present during both acquisition and extinction of the association. It is not

clear, however, whether affective judgments should better reflect the simple

association (i.e., CS+ increases in displeasure and arousal), or the

discriminative aspects (i.e., CS+/CS- difference) of the learning situation.

In addition, this experiment examined whether pairing aversive shock

with different emotional slides would yield variable degrees of affective








associative change. Whereas Hamm et al. (1991) assigned subjects to view
slide pairs which differed in affective valence and arousal, slide content was

confounded with this manipulation. In this study, there were four groups

who viewed either unpleasant-calm, unpleasant-arousing, pleasant-calm, or

pleasant-arousing slide foregrounds during conditioning. (Unlike Hamm et

al., 1991, neutral contents were not used as foregrounds in this experiment.)

Each subject viewed two pictures which he or she rated as most extreme

(regardless of content) in terms of his or her stimulus assignment to one

quadrant of the two-dimensional (valence X arousal) affective space.

The present approach is a deliberate, extreme-case test of the hypothesis

that slide affective evaluation should influence classical conditioning.

Although pairing any stimulus with shock should generally increase its

unpleasantness (and arousal), when the CS itself is aversive initially, the

amount of affective change due to association with shock may be minimal.

Conversely, a stimulus which is initially pleasant (or calm) might accrue

more displeasure (or arousal). If startle blink potentiation indexes the

amount of learned aversion, then greater changes in rated displeasure and

the startle reflex should occur for pleasant foregrounds, whereas smaller

changes are predicted for slides initially high in unpleasantness. Similarly, if

skin conductance response potentiation is an index of learned arousal, then

larger changes in rated arousal and electrodermal reactions should occur for

calm slide foregrounds, while smaller changes are predicted for slides

initially high in arousal. Such results, if obtained, would confirm the

findings of Hamm et al. (1991).

The third goal, clearly exploratory at present, was to identify individual

differences which mediate emotional learning. Since it is possible that a

subject's shock tolerance level is determined by personality characteristics,








and that temperament or shock sensitivity may relate to aversive learning,
these variables were examined. The EASI Temperament Survey (Buss &
Plomin, 1975) was used here to assess differences in aversive reactivity and
learning. The EASI subscales measure temperament dimensions of
fearfulness and impulsivity, as well as anger, activity, and sociability. If
research relating to Eysenck's personality theory is correct in suggesting that
anxious and/or inhibited subjects should either respond more to aversive
stimulation or show better conditioning, then subjects who score relatively
higher on the EASI fear subscale and lower on the impulsivity subscale
should demonstrate these effects maximally.
Finally, as the cognitive model of conditioning adopted here implies,
perceptual factors may relate to the degree of emotional learning. Therefore,
it was predicted that psychophysiological measures of learning would be
enhanced in subjects who report greater aversiveness of the shock US.












METHOD

Subjects

Sixty-four introductory psychology subjects (32 males, 32 females)

volunteered to participate in this experiment and received experimental

credits toward fulfilling their course requirements. Subjects were informed

before arriving at the laboratory that the study involved "cognitive and

physiological responses to stimuli in different sensory modalities." Total

participation time for each subject was two hours.
Stimulus Materials

Affective pictures. Thirty-nine 35 mm colored photographic slides,

selected from catalogue of the International Affective Picture System (lAPS;

Lang, Ohman, & Vaitl, 1988), were used in this experiment. These slides

depict a wide variety of content, including wild and domestic animals,

natural scenes, objects, human faces and bodies, social interactions, and

conditions of life. The IAPS slides have relatively simple figure-ground

relationships which facilitate perceptual resolution and evoke a broad range

of affective responses within the limits of this stimulus format.

The composition of the initial slide set was based on an inspection of

normative ratings for all 240 IAPS slides (Lang, Greenwald, & Bradley, 1988).

Appendix A lists all slide materials used in this study and their normative

affective values on each dimension. The slides were distributed across the

four quadrants in the coordinate space defined by the intersection of affective

valence and arousal axes. Eight slides from each of these four quadrants

(unpleasant-arousing, unpleasant-calm, pleasant-arousing, pleasant-calm)








and four neutral-calm slides comprised the slide set. Additionally, three
practice slides (unpleasant, neutral, and pleasant) preceded presentation of

the above 36 slides during the pre-experimental viewing/judgment series, to
anchor subsequent ratings of these materials.

Acoustic startle probe. The startle-inducing stimulus was a 50 msec
burst of 95 dBA white noise (20 20,000 Hz), with instantaneous rise time

(< .05 msec). The white noise was generated by a Coulbourn S81-02 Noise

Generator, gated through a Coulbourn S82-24 amplifier, and delivered

binaurally via matched Telephonics TDH-49P earphones. To calibrate noise

intensity, the earphone was clamped to a Bruel and Kjaer Type 4153 artificial

ear (1/2" Type 4134 condenser), thus directing the sound wave perpendicular

to the diaphragm (Putnam, Graham, & Sigafus, 1975). Sound pressure level

of the continuous noise was calibrated using the fast-filtered A scale of a

Bruel and Kjaer Type 2203 sound level meter.

Aversive unconditioned stimulus (US). The electric shock US was a

continuous 500 msec train of rectangular pulses generated by a Farrall

Instruments Mark 300C Aversive Conditioner. The tactile stimulus was

delivered through a concentric Tursky electrode. The electrode was attached

to the volar surface of the left upper arm, which was first rubbed lightly with

Ultraphonic conductive gel. Intensity of the US was set at a level which the

subject described as "highly annoying, but not painful" (see Procedure).
Shock intensity was calibrated using a Micronta DC analog meter (85 ohm

internal resistance), a bridge circuit (to rectify the shock pulse), a resistance

load of 27 Kohms, and a 100 iF capacitor across the output terminals of the

Farrall shocker. For each step on the shock current knob, the value on the

Micronta meter was multiplied by the duty cycle (32.05) to obtain peak

amperage levels.







Apparatus and Psychophysiological Recording
Stimulus presentation and data acquisition were coordinated by
distributing the processing tasks (via digital input/output connections)
between PDP 11/23 (Data Translation 12 bit analog-to-digital [A-D] converter
board) and Apple IIe microcomputers. The Apple IIe triggered a Gerbrands
electronic shutter (for slide onset/offset; rise/fall time < 5 msec) and the slide
carousel advance/reverse switches of a Kodak Ektagraphic II AM projector.
The projector lens was located behind a clear plexiglass sound baffle in the
experimenter's room; thus, the subject had no audible cues as to which slide
(CS+, CS-) would be presented next during conditioning. The PDP computer
recorded all physiological data and controlled the timing of slide, startle
probe, and shock presentations. Physiological signals (except heart rate) were
sampled at 10 Hz, except during startle probes, during which the sampling
rate was temporarily accelerated to 1000 Hz from -50 to +250 msec relative to

probe onset.
The eyeblink component of the startle response was measured by
recording the electromyographic activity over the left orbicularis oculi site,
using Sensor Medics miniature silver/silver chloride electrodes filled with
Teca electrolyte gel. The raw signal was bandpass-filtered (from 90 to 250 Hz)
and amplified (20,000 gain factor for all subjects) by a Coulbourn S75-01
Bioamplifier, and integrated on-line using a Coulbourn S76-01 Contour
Following Integrator (at a time constant, calibrated off-line, of 120 msec).

The analog heart rate signal was measured using collarbone and arm
leads with Sensor Medics standard size silver/silver chloride electrodes,
using Teca electrolyte gel. Following input to a Coulbourn S75-05 Isolated
Bioamplifier, the signal was filtered from 8 to 40 Hz. A Schmitt trigger
detected R-wave peaks and interrupted the computer to record interbeat








intervals to the nearest msec from 3 sec prior to slide onset to 2 sec after slide
offset.
Skin conductance activity was recorded with Sensor Medics standard

size silver/silver chloride electrodes attached adjacently to the hypothenar
eminence of the left palmar surface. The skin was first cleaned with distilled
water and sensors were filled with K-Y jelly electrolyte. A Coulbourn S71-22
skin conductance coupler provided a constant 0.5 volt across electrodes. The
signal was calibrated prior to each session to detect activity in the range from

0 to 40 gSiemens (pS). The calibration value was used to convert the
digitized raw signal off-line to analog skin conductance values.
Scoring and Definitions of Physiological Response Measures

Startle reflexes were scored using a computer program that scored the

blink waveform for onset latency and amplitude. Eyeblinks were scored by

the algorithm if (1) minimum response amplitude was 20 A-D units and if

the response had a (2) minimum onset latency of 20 msec and (3) maximum

peak latency of 150 msec relative to probe onset. Null response trials (due to

violations of criteria 1-3) were assigned an amplitude of zero A-D units.
Startle reflex magnitude (including zero amplitude responses) was the

principal measure of interest. For startle onset latency, null response trials

were excluded from analyses.
Heart rate was converted from digitized interbeat intervals to analog

rate data in half-second bins, with intervals weighted in proportion to the
time occupied (Graham, 1980). Initial and late deceleratory (Dl and D2) and
midinterval acceleratory (Al) components of the cardiac waveform were

scored (Gatchel & Lang, 1973), since these peaks and troughs in the

waveform reflect psychologically significant events (Bohlin & Kjellberg,








1979). These component scores were expressed in beat/minute change
scores, deviated from a 1 sec prestimulus baseline.
Skin conductance responses were scored using an adaptation of the
Strayer and Williams (1981) SCORIT algorithm. The First Interval Response
(FIR)--which presumably reflects orienting to stimulus onset--was defined as
the largest response during the window from 0.9 to 4.0 sec after stimulus

onset (Prokasy & Raskin, 1973). The Second Interval Response (SIR)--which
reflects anticipation of the shock US during acquisition and extinction--was
scored as the largest response on each trial between 4.1 and 6.8 sec after slide
onset (i.e., ending 0.8 sec after slide offset). Third Interval Responses (TIRs)--
which index the UCR during acquisition and orienting to shock omission
during extinction--were scored as the largest responses between 0.9 and 4.0
sec after slide offset. Skin conductance response data were later transformed

(log [SCR+1]) to normalize the distribution of responses.
Procedure

Equipment preparation and instructions. After arriving at the

laboratory, each subject completed a consent form to participate in a brief (ca.
30 min) slide viewing/rating experiment. The subject was informed about
the different emotional pictures, a "brief noise click" and recording of
physiological responses. The subject was also told that that there would be a

second optional experiment, which the experimenter would describe later.
No mention of shock was made at this time. The subject was seated in a

comfortable recliner in a sound-attenuated, dimly lit room adjacent to the
equipment room. Physiological sensors were attached at this time, but
(unknown to subjects) physiological responses to slides were not actually
recorded during the first experiment.








Subjects were then instructed in using the Self-Assessment Manikin

(SAM; Lang, 1980). Subjects used the pencil-and-paper version of SAM (see
Appendix B, Figure 19) to make affective valence and arousal judgments of

experimental stimuli. To represent changes on the valence dimension, the

SAM figure ranges from smiling and happy to frowning and unhappy. To

represent changes in arousal, SAM ranges from active and wide-eyed to a

static, eyes-closed figure. There are 5 cartoon-like figures for each dimension.
Appendix B also provides the experimental instructions for using SAM.

Briefly, subjects were instructed to place an "X" on the figure, or between

figures, which best represented their affective reaction to the slide on each

dimension. This scheme formed a 1 9 rating scale for each affective

continuum. "Low" (1) on the valence dimension was unpleasant, whereas
"high" (9) was pleasant. "Low" (1) on the arousal scale was calm, and "high"

(9) was excited. During initial practice with SAM, the experimenter

instructed subjects with verbal labels for the endpoints and middle of each

scale. These labels correspond to the anchor adjectives of the semantic

differential (Mehrabian & Russell, 1974).

Prior to leaving the room, the experimenter instructed the subject that

the noise would occur "at various times during the session" and then placed

on the headphones.

Pre-experimental slide series and selection of conditioning slides.

Subjects paced themselves through a series of 39 slides, the first three of
which (unknown to subjects) were practice slides and did not contribute to

their ratings distribution. The subject was instructed to press a button which

initiated slide onset, watch the slide closely while it was on the screen, and

rate each slide for valence and arousal immediately after slide offset. After

the subject initiated slide exposure, it remained on for 2 sec. The subject was







only told that the slide exposure would be "relatively brief," so that after
pressing the button (1 sec delay), he or she should be ready to view the slide.
After each slide trial, the subject repeated this sequence. Noise probes were
not delivered during this slide sequence.
The subject was instructed to relax after the final slide trial (which was
evident from the numbered rating form). At that time, the experimenter
determined from a video monitor that the subject was finished writing and
resting comfortably. When this was confirmed, two acoustic startle probes
were presented (16 sec apart) to assess prelearning probe response.
The subject's rating form was then collected, and the experimenter
examined the two-dimensional (valence, arousal) distribution of affective
judgments. Before arriving at the laboratory, the subject had been randomly
assigned to view one of four slide types (unpleasant/arousing, unpleasant/
calm, pleasant/arousing, pleasant/calm) during learning. Accordingly, the
experimenter first selected the two conditioning slides which the subject had
rated most extremely in his or her assigned category. The experimenter then
selected ten adjunct slides--two in each of five affective categories (same four
types as above, plus neutral/calm)--which were not used in learning but
were rated again after the experiment (see below). Appendix A explains the
decision rules for choosing these conditioning and adjunct slides.
Shock exposure. The subject was instructed that a new slide sequence
would begin, and to "watch each slide for the duration of exposure and
ignore the noise." While subjects were not informed of the actual slide
exposure duration (6 sec) or when the noise would occur, probes were
presented equally often at each of 4 times during slide exposure--either 3.00,
3.75, 4.50 or 5.25 sec--as well as between slides (i.e., interslide intervals).








Subjects were instructed to relax quietly during interslide intervals, which
ranged from 10 to 20 sec in 5 sec increments.
Next, the procedure diverged for two groups (see Figure 2). Forty-eight
subjects were told, "Experiment 1 is now over." These subjects--designated
as the Immediate shock exposure group--were thanked and invited to
continue in the next experiment, which provided additional credits toward
their class research requirements. This group was told that there would be a
new consent procedure for this second experiment and that there was no
obligation to participate. If the subject agreed to continue, he or she was
given the next informed consent to read. Three subjects (in addition to the
sample of 48) declined to continue; they were thanked for their participation,
given credit, and debriefed.
The second consent procedure informed the subject about electric shock.
Otherwise, the description of slide viewing and noise probes was the same as
before. After written consent was obtained, the experimenter attached the
shock electrode and initiated shock exposure (Appendix B). In this standard
procedure, the intensity of shock was adjusted for each subject to a level
which was reported as "highly annoying but not painful." The shock level
began just above perceptual threshold and was increased gradually in fixed
steps to the subjective criterion. Increments in shock level were calibrated to
approximate a power function (Appendix B, Figure 20). All subjects reached
at least the second shock intensity.
Once the shock level was determined, the subject was told that, at
various times during the remainder of the experiment, the shock would

occur at the intensity and duration agreed upon. No mention was made of
"conditioning," association, nor timing of stimulus presentation. The








experimenter answered any questions from the subject, informed the subject
that the next phase would begin, and exited the room.
The adaptation sequence of to-be-conditioned slides began. Each slide
(CS+, CS-) was presented twice and probed twice; two startle probes were
presented randomly between slides. This was the first block of adaptation.
Then the experimenter re-entered the room to break the procedure for an
unanticipated questionnaire session. After questionnaires, the experimenter
left the room for the last time. The second block of adaptation (identical to
the first block) began, and acquisition and extinction slide series continued
uninterrupted from this point.
The remaining 16 subjects--designated as the Delayed exposure group--
completed questionnaires first instead of being exposed to shock. Following
questionnaires, they were told that a new slide sequence would be presented
and were given the instructions for slide viewing above. Adaptation block 1
slide and probe presentations were then administered in the same manner
as for the Immediate exposure group. The experimenter then re-entered the
room to inform subjects in this group that experiment 1 was over. The
experimenter then obtained the second informed consent, conducted the

shock exposure protocol, and left the room a final time. After this point, the
second adaptation block, acquisition, and extinction slide series proceeded
without interruption--identical to that for the Immediate exposure group.

To summarize, when the Immediate group was first exposed to shock,
the Delayed group completed questionnaires. This process was reversed, so
that when the Delayed group was exposed to shock, the Immediate group

completed questionnaires. (Both shock exposure and questionnaires lasted
about 10 mins.) This interlude enabled each group to serve as the other's
control group at their time of shock exposure, as well as an opportunity to








collect individual difference data. All subjects completed an abridged

version of the Fear Survey Schedule (FSS; Wolpe & Lang, 1964), consisting of
40 items describing different types of fears depicted by IAPS slide contents. In

addition, the 25 item EASI Temperament Survey (Buss & Plomin, 1975) was
administered. The EASI subscales tap constructs of emotionality (fear and

anger), activity, sociability, and impulsivity. Items on both questionnaires

were rated on a 1 5 scale, yielding a 40 200 range for this FSS version and a
5 25 range for each EASI subscale. These questionnaires (see Appendix C)

were used to address the relationship of temperament dimensions and

situational fears to aversive reactivity and learning.

Shock conditioning. The remainder of the experiment involved the

discriminative learning procedure. Subjects were randomly assigned to

view an unpredictable sequence of either two unpleasant/calm, unpleasant/
arousing, pleasant/calm, or pleasant/arousing slides. These two slides

served as CS+ and CS- (see Figure 3). During acquisition, the CS+ slide was

always reinforced with a 500 msec shock US which coincided with slide

offset. The CS- slide was never reinforced. Each slide was presented 8 times

(16 total trials; 8 different CS+/CS- orders, counterbalanced across subjects).1



1 For all subjects, the last trial of adaptation was a CS+ trial, followed by the
shock US. Since subjects could not anticipate this contingency, preceding
events were functionally part of the adaptation phase. Thus, the first CS+
presentation in acquisition accompanied a shift in the subject's experience of
event-relations and is a pure acquisition trial. By similar logic, the last CS+
trial of acquisition was not reinforced with shock. As before, subjects could
not anticipate this new contingency and, therefore, all preceding events were
functionally part of acquisition. In addition, to minimize the tendency for
subjects to expect a reversal of the contingency, an unreinforced CS- buffer
trial--which was not analyzed--was introduced between acquisition and
extinction for all subjects. The intent of these latter two manipulations was
to purify interpretation of the extinction phase.








Since the startle probe is prepotent aversive US, probes were not presented
during acquisition so as not to be associated with learning.

During prior adaptation and subsequent extinction phases only, probes

were presented during and between slides. As noted earlier, each of the 2 CS

slides was presented and probed 4 times during the adaptation series. These
8 trials were intermixed with 4 no slide/probe trials (12 total trials). Because

of the between-group shock exposure manipulation, adaptation was divided

into two equal parts. In each of four extinction blocks, each slide was seen

and probed twice, and 1 no slide/probe was presented. Thus, there were a

total of 8 CS+, 8 CS-, and 4 no slide/probe trials.

Throughout conditioning, slide exposure was 6 sec. Variable interslide
intervals ranged from 10 to 20 sec in 5 sec increments.

Postlearning: Affective ratings and debriefing. After the last extinction
trial, the experimenter removed the recording sensors, shock electrode, and
headphones. The experimenter informed the subject that the noise and
shock would no longer be presented. The experimenter then instructed the
subject that he or she would view a final series of slides and rate each slide

again using SAM. The procedure was identical to prelearning ratings, except

that there were only 12 slides. Two of these slides were the subject's CS+ and
CS-. The remainder were the 10 adjunct slides falling into the five affective

categories mentioned above. After completing the slide ratings, the subject
also rated (from memory) the noise and shock stimuli with SAM.

After affective judgments were completed, subjects were asked in a

debriefing session to state (1) their estimate the number of shocks received;

(2) whether they recognized any rule during the procedure; (3) whether some

stimuli were followed by shock more often than others; (4) whether they

heard the noise over the headphones; (5) if they knew which physiological








measures were being recorded; and (6) whether they knew or could guess the
purpose of the experiment. In addition, they were also asked to rate each of
the following items for their aversiveness (i.e., "the extent to which you
would avoid the stimulus") on a 1 (not at all aversive) to 7 (extremely
aversive) scale: "noise over the headphones," "shock," "shocked slide," and
"nonshocked slide." After these debriefing questions, the subject was
informed about the purpose of the experiment, thanked, given class credit,
and dismissed.
Design and Data Analysis
In previous human aversive learning studies, subjects have received
shock exposure prior to learning. In this experiment, however, designed
variations in the timing of shock exposure across groups complicate
interpretation of the adaptation phase data. Thus, to facilitate interpretation
of the conditioning data within this context, the effects of shock exposure
timing on startle reflex, tonic visceral levels, slide visceral responses, and
habituation rates for these measures are reported first.
Otherwise, this study largely repeats (with minor improvements) the
discriminative conditioning procedure used by Hamm et al. (1991). To
facilitate comparisons with those data, univariate analyses of physiological
measures (startle magnitude and onset latency, electrodermal response, and
heart rate response) were conducted separately for each phase of
conditioning (adaptation, acquisition, and extinction). Affective ratings
(valence and arousal) were assessed before and after associative learning.
The experiment was divided, both from the standpoint of design and
analysis, into several parts, hereafter referred to as (1) Prelearning, or initial
slide viewing/rating; (2) Sensitization; (3) Associative learning, comprised of
adaptation, acquisition, and extinction phases; and (4) Postlearning. It is







important to note that the adaptation slide sequence is common to both the
sensitization and associative learning segments.
Prelearning. The first objective was to assess the ratings distribution for
each subject and--based on the subject's a priori group assignment--use these
ratings to select two matched slides as foregrounds for conditioning. Thus,
analyses of these ratings served as a manipulation check that the four
conditioning groups significantly differed in valence and arousal, and that
conditioning slides within each group did not differ prior to learning. Slide
Valence (2: Unpleasant, Pleasant; between-subject) X Slide Arousal (2: Calm,
Aroused; between-subject) X Stimulus (2: CS+, CS-; within-subject)
univariate analyses of variance (ANOVAs) were conducted on both valence
and arousal ratings. Similar analyses were conducted on adjunct (i.e.,
nonconditioned) slides to determine their distribution in the affective space.
Sensitization. To assess shock sensitization of the startle reflex without
the confound of different affective slide foregrounds, analyses focused on
startle probe responses during interslide intervals. Missing data (2% of all
prelearning and adaptation trials) were estimated by regression, separately
for each exposure group. One subject in the Immediate exposure group was
excluded because more than 50% of his responses had been rejected, leading
to insufficient data for estimation. Changes in the probe startle response as a
consequence of shock exposure were analyzed in an ANOVA for Exposure
Group (2: Immediate, Delayed; between-subject) X Interslide Probes (2) X
Probe Blocks (3: prelearning, first half adaptation, second half adaptation;
within-subject).
To examine sensitization of tonic visceral levels during shock exposure,

average heart rate and skin conductance for the 1 sec prior to each adaptation
trial (in the absence of slide stimulation) were computed for each subject.








Since there were 12 adaptation "trials"--interslide intervals were treated as
trials here--the change in pretrial baseline levels during shock exposure
could be examined. Although tonic visceral level data were not available
during the prelearning rest period, the effect of shock exposure on visceral
levels could be tested for the Delayed exposure group since, for these subjects,
adaptation block 1 preceded exposure, and adaptation block 2 followed

exposure. These data were explored using Exposure Group (2: Immediate,
Delayed; between-subject) X Trials (6: consecutive, regardless of CS+, CS- or
interslide) within Trial Blocks (2: first half adaptation, second half

adaptation; within-subject) ANOVAs.
To examine effects of shock exposure on slide visceral reactivity, slide-
by-slide heart rate leg scores and skin conductance responses (consecutive

trials, without regard to CS+ or CS-, within each half of adaptation, as well
as acquisition and extinction) were analyzed, separately for each exposure
group. Slide visceral responses were tested using Exposure Group (2:
Immediate, Delayed; between-subject) X Slide Trials (4: consecutive,
regardless of stimulus) X Blocks (2: first half adaptation, second half
adaptation; within-subject) ANOVAs.
Associative learning. All physiological data were analyzed in blocks of
two trials separately for each phase of conditioning (adaptation series [2 trial
blocks], acquisition [4 trial blocks], and extinction [4 trial blocks]) with
univariate ANOVAs. Subjects were randomly assigned to receive one of
four slide affective foregrounds during conditioning. Slide Valence and
Arousal (2 levels each) constituted the between-group factors in the design.

To evaluate conditioning within groups over trials, Stimulus (2: CS+, CS-)
and Trial Block (2 or 4 within phase) factors formed the remaining factors of

the design. For ANOVAS involving the trial block repeated measure term,








unadjusted degrees of freedom and Greenhouse-Geisser epsilon (e)

corrections are reported.

In analyses of associative learning, "discrimination" was defined as the

extent to which CS+ response exceeded CS- response during acquisition or
extinction. Startle potentiationn" was defined as the degree to which CS+

response increased from adaptation block 2 to extinction block 1. Since

visceral responses rapidly habituate, potentiation of visceral activity was

tested as the change in CS+ response from adaptation block 2 to acquisition

block 1.
Postlearning. For affective judgments, changes in affective ratings of

each subject's CS+ and CS- before and after conditioning were tested in

univariate Phase (2: Pre, Post) X Stimulus (2: CS+, CS-) X Slide Valence (2:

Unpleasant, Pleasant) X Slide Arousal (2: Calm, Aroused) analyses of

variance for valence and arousal ratings. Statistical tests evaluated the
hypothesis that CS+ ratings should become (or remain) more unpleasant

and arousing than CS- ratings. Similar analyses were conducted on adjunct

slides to assess the degree of affective change.

Individual differences. Relationships of the temperament dimensions

to both aversive reactivity and learning were initially examined with zero-
order correlations and stepwise multiple regression. The EASI subscale

scores were entered as independent variables in the regression models to

predict measures of aversive reactivity (e.g., shock tolerance, prelearning

startle response, shock aversiveness, situational fearfulness) and the extent

of aversive learning (e.g., startle sensitization and potentiation, cardiac and

electrodermal conditioning, affective ratings change). Follow-up ANOVAs

were conducted to describe group mean differences.





35


Relationships of shock tolerance and rated shock aversion to learning

measures were also examined. Stepwise multiple regression was used to

enter both variables to predict the extent of learning in the psychophysio-

logical response measures. Again, ANOVAs were conducted to describe

mean differences among groups.

For all statistical tests, the size of the rejection region was set at .05.












RESULTS

The strategy for data analysis and presentation proceeds as follows.
First, the effect of shock exposure on response sensitization during the
initial slide adaptation series is examined. Second, the influence of aversive
classical conditioning on psychophysiological responses is evaluated,
separately for initial adaptation, acquisition, and extinction slide series.
Third, effects of individual differences (temperament characteristics,

tolerance of shock, and judgments of the aversiveness of stimuli) on
psychophysiological reactivity and associative learning are assessed.

Sensitization
Startle Probe Response
Reflex magnitude. It was predicted that after exposure to shock, each

group would show a marked rise in startle magnitude, and that this would

not occur under control conditions. As mentioned earlier, this was tested
using interslide interval data to avoid the confounding influence of slide

valence or arousal on sensitization. As predicted, startle magnitude was

clearly sensitized by shock exposure (see Table 1). The omnibus test of

sensitization learning was significant, Exposure Group X Trial Block
F(2,124)= 7.11, e= .91, p< .002. Planned comparisons testing the sources of this
interaction used blocked averages for each subject, since interslide responses

within prelearning and adaptation series probe blocks did not differ.

There was no group difference in startle magnitude during prelearning
(i.e., before shock exposure), F(1,62)= 1.77, p= .19. For the Immediate

exposure group, the increment in startle magnitude following shock







Table 1

Startle magnitude means (ADUs + standard errors)
during prelearning and interslide intervals of adaptation.


Exposure Group Prelearning Adapt. block 1 Adapt. block 2


Immediate 364 (shock 502 424
(51) exp.) (59) (56)

Delayed 236 290 (shock 415
(65) (72) exp.) (81)


Table 2

Startle onset latency means (msec + standard errors)
during prelearning and interslide intervals of adaptation.

Exposure Group Prelearning Adapt. block 1 Adapt. block 2


Immediate 38.3 (shock 35.1 36.1
(1.0) exp.) (1.0) (1.1)

Delayed 45.9 39.2 (shock 36.7
(2.1) (1.2) exp.) (1.6)





























* "
/ I- ia
S ,

S _
I. -


-I,
0r



B
U





I b


(siiun a-V) m!j4 4luuealajd Isuj
wuoi apnlrUeuw -uilq ui afueqD


- Ct


Ta

..
.a- 0
M


~1
000
0-4





*D bO-~
(U* C1



CU
(a cU
8" o)

0
&'.- 0
'0 r-; 0




baa) E j
a0
"SS3






bO
5; s.
lii
5 4
~c fi a
OC~K^
^s-1
rn ~ <"
d aE
ca^
1||i ~







exposure (A= +138 ADUs) was significant, F(1,47)= 18.57, p< .005. After the
sensitization peak, startle response habituated, F(1,47)= 12.30, p< .005.
For the Delayed exposure group, the increase from prelearning to
adaptation series block 1 (control period; A= +54 ADUs) was not significant,
F(1,15)= 3.22, p> .09. As predicted, the subsequent increase in startle reflex
magnitude directly attributable to shock exposure (A= +125 ADUs) was
statistically supported, F(1,15)= 9.42, p< .01.
As seen in Figure 4, startle magnitude increased markedly at the first
probe after shock exposure for each group. The size of the sensitization effect
was equal across groups (about 40%), and absolute startle magnitude in block
2 of the adaptation series--after both groups had been exposed to shock--was
nearly identical (see Table 1). This is noteworthy, since the end of the initial
adaptation series was intended to serve as the general baseline for the test of
startle potentiation. Also, startle reflex sensitization was independent of
orbicularis oculi tension (Appendix D, Table 7) and responses during slides
were comparable with responses during interslide intervals.
Startle onset latency. Although latency differences were obtained, the
findings were not consistent with a shock sensitization effect (Table 2).

Shock exposure was expected to speed onset latency (first for the Immediate

group, later for the Delayed group), but latencies were generally speeded
from prelearning to the first block of the adaptation series for both groups,
F(2,122)= 15.84, e= .88, p< .0000. A Group X Block interaction did emerge
(which would be anticipated for sensitization), F(2,122)= 5.06, e= .88, p< .02,
but this was due to overall slower responses in the Delayed group, F(1,61)=
6.66, p< .02, which became significantly quicker in onset by the second block
of the adaptation series. Thus, the present data do not support the
conclusion that startle onset latency is reliably related to shock exposure.








Visceral Reactions
Skin conductance. Electrodermal tonic levels habituated within blocks
of adaptation trials for both groups. Although Table 3 suggests a group
difference in levels in block 1 of the adaptation series, this effect was weak
(p= .15). There was a large increase in conductance level for the Immediate
group from adaptation block 1 to 2 (i.e., after questionnaires), but not for the
Delayed group (i.e., after shock exposure), Group X Trial Block F(1,60)= 6.42,
p< .02. The source of this interaction was the significant increase in the
Immediate group. Importantly, conductance levels of the Delayed group
were not sensitized by shock exposure (p= .64). Tonic levels of the two
groups did not significantly differ in acquisition or extinction.
Slide conductance responses of both groups increased from block 1 to 2
of the adaptation series; the slightly greater increase for the Delayed shock
exposure group was not significant, Group X Trial Block F(1,62)< 1 (Table 3).
Tested alone, the increase in the Delayed group following their exposure to
shock was not reliable (p= .14). Since slide electrodermal response increased
in both groups, dishabituation may have contributed here. Incidentally, the
Immediate group had larger mean skin conductance responses to slides
throughout conditioning, but this overall group effect was only significant in
acquisition, F(1,62)= 3.85, p= .05. The Immediate group also had larger
Second Interval anticipatory responses than the Delayed group in acquisition

(Ms= .14 and .09 pS), F(1,62)= 5.33, p< .03, but the groups did not differ in
Third Interval Response magnitude. Shock exposure groups did not differ
on Second Interval or Third Interval response measures in extinction.
Heart rate. Heart rate levels of the Immediate group generally decreased
from block 1 to block 2 of the adaptation series, whereas levels of the Delayed
group increased slightly (see Table 3). There was a Group X Trial Block







Table 3

Mean visceral levels and slide responses for
Immediate and Delayed shock exposure groups.


Visceral Adaptation Acquisition Extinction
Measure block 1 block 2


Skin conductance tonic level (pS)

Immediate 6.42 7.76 7.05 6.79

Delayed 7.75 7.92 8.03 7.69

Slide skin conductance response (log [FIR+11 pS)

Immediate 0.23 0.24 0.22 0.13

Delayed 0.15 0.19 0.13 0.08


Heart rate tonic level (bpm)

Immediate 71.39

Delayed 69.52

Early-interval slide deceleration

Immediate -3.33

Delayed -3.90

Mid-interval slide acceleration

Immediate 2.69

Delayed 4.96

Late-interval slide deceleration

Immediate -3.43

Delayed -3.85


69.14

70.47

(Dl; Abpm)

-3.47

-3.99

(Al; Abpm)

3.30

4.35

(D2; Abpm)

-2.89

-3.45


68.68

70.31


-2.83

-4.71


3.30

3.70


-5.00

-6.28


69.35

70.08


-2.81

-4.00


3.51

3.62


-4.20

-6.15








interaction, F(1,61)= 4.33, p< .05, which was due solely to habituation of
levels across blocks in the Immediate group, F(1,46)=9.23, p< .005; that is,
tonic heart rate levels did not increase (i.e., subjects were not sensitized) in
the Delayed exposure group (p= .53). Heart rate levels did not differ across
groups during conditioning.
Shock exposure also did not sensitize cardiac response to the slides
during the adaptation series, although group differences emerged later in
conditioning (Table 3). For initial deceleration (D1 leg), the Delayed group
decelerated more than the Immediate group during acquisition, F(1,61)= 8.02,
p<.01, and extinction, F(1,61)= 4.41, p< .05. The Delayed group showed
greater mean deceleration late in the slide interval (D2 leg) in acquisition,
but this was not significant (p= .13). In extinction, the group difference in
late deceleration was significant, F(1,61)= 4.76, p< .04. There were no group
differences in Al leg response during any part of conditioning.
Summary
Only startle magnitude was significantly influenced by shock exposure:
Startle responses were 40% larger immediately following shock, relative to
control conditions, suggesting that this increase cannot be attributed to
dishabituation or simple passage of time. In contrast, neither startle onset
latency nor visceral reactions were sensitized by exposure to shock.

Associative Learning
Prior to learning, the CS+ slide, CS- slide, and 10 adjunct slides were
selected from each subject's ratings distribution (see Appendix A).1 Mean

1 The terms "Negative," "Positive," "Calm" and "Arousing" and their
combination (e.g., "Negative/Calm") refer to the different conditioning
groups viewing slides of different affective valence or arousal. To minimize
confusion, these group references will be distinguished from the lower-case








valence and arousal ratings for conditioning and two control slides--which
were drawn from the same-affect category as each subject's CS slides--are
presented in Appendix D (valence, Tables 8-9; arousal, Tables 10-11). Adjunct
slide ratings (Appendix D, Tables 12-13) were not significantly affected by
conditioning.
Affective Ratings
Valence judgments. Figure 5 illustrates each group's affective valence
and arousal ratings of their foreground slides both before and after learning.
Valence ratings of CS+ and CS- slides did not differ prior to learning. Not
surprisingly, affective valence ratings by subjects assigned to view pleasant
CSs were higher than ratings by subjects assigned to view unpleasant CSs
(regardless of arousal), F(1,60)= 1910.38, p< .0000.
Regardless of content, CS+ was rated more unpleasant than CS- after
learning (Ms= 3.95 vs. 4.55), F(1,60)= 9.78, p< .003; i.e., valence discrimination
did not significantly vary by affective slide condition.
Analysis of valence change scores also indicated that, overall, CS+

became significantly more unpleasant than CS- (As= -.81 vs. -.25 units),

F(1,60)= 7.87, p< .01. Figure 5 shows that the direction of ratings change was
opposite for subjects viewing unpleasant vs. pleasant foregrounds, F(1,62)=
54.86, p< .0000. Follow up tests found that for Positive slide groups, CS+
became significantly more unpleasant (A= -2.19) whereas for Negative
groups, CS+ did not significantly change (A= +.56).
Depending on the slide foreground, conditioning and control slides
(from the same affective category as the subject's CSs, but not conditioned)


terms "unpleasant," "pleasant," "calm," and "arousing," which denote the
affective qualities of slide pictures or subjects' ratings of the pictures.














0


Positive/Calm
group


S 2 3 4 5


Negative/Calm
group


Positive/
Arousing
group


6 7









g


Negative/
Rousing
group


Arousal Ratings



Figure 5. Affective valence and arousal ratings of conditioned
stimuli before learning (origin of lines) and after learning (terminal
point of arrows), separately for each slide foreground group.


J I I























.n b 4


bo < ne



4-.4 .

0) 0

0 0 b I
bu o
> (L
s s s









00 V
u54 -V .




u n


U)
(U U-
n > >






< U


(s4unm Supei faid-4sod) aueq)ip aualeA


+ o
au

UUU6 l


a oa 9








showed different relative change, Slide Valence X Stimulus F(2,124)= 9.93,
p< .0001 (see Figure 6). Follow up tests found that, for Negative groups, CS-
became significantly more pleasant than both CS+ and control slides. In
contrast, a gradient resulted for Positive slide groups, in that CS+ became
marginally more unpleasant than CS- (p< .07), whereas both CSs became
significantly more unpleasant than control stimuli.
The shock US and startle probe were rated as equally unpleasant after
learning (Ms= 1.97 and 2.16, respectively).
Arousal judgments. Arousal ratings of the CS+ and CS- slides did not
differ before learning. Due to the experimental selection criteria, arousal

judgments by subjects assigned to view more exciting CSs were significantly
higher than ratings by subjects assigned to view calm CSs (regardless of
valence), F(1,60)= 62.31, p< .0000.
Unlike valence ratings, classical conditioning did not produce arousal
judgment discrimination. The overall difference between CS+ and CS- was
not significant, F(1,60)< 1, and did not vary by slide foreground.

Arousal change scores also failed to provide evidence of learning.

Although CS+ maintained slightly more arousal than CS-, the overall

difference between CS+ and CS- change was not significant, F(1,60)< 1.

However, the direction of ratings change was opposite for subjects viewing

calm vs. arousing foregrounds (see Figure 5), F(1,62)= 15.96, p< .0005. For

Arousing groups, CS+ became significantly more calm (A= -1.59), whereas
for Calm groups, CS+ did not significantly change (A= +.60). There were no

other significant findings.

Depending on the slide foreground group, conditioning and control

slides showed different relative change, Slide Arousal X Stimulus F(2,124)=
5.16, p< .01 (see Appendix D, Table 11). For Arousing slide groups, both CS+








and CS- became significantly more calm than control slides (ps< .02). Calm
slide groups failed to show this effect; differences in arousal change between
control slides and both CS+ and CS- slides were not significant.
Arousal ratings of the shock and startle probe were similarly high (Ms=
7.19 vs. 6.77); even so, this small difference was reliable, F(1,60)= 4.08, p< .05.

Startle Probe Response
Reflex magnitude. As expected, CS+ and CS- startle magnitude did not
differ before learning (see Appendix D, Tables 14-15 for reflex magnitude
means for all groups during conditioning). There were no other significant
effects on startle magnitude during the adaptation slide series.2
It was predicted that startle magnitude evoked during the CS+ would

increase immediately after learning. Clearly, however, the ability to detect

change at extinction relies on the initial level from which change is assessed.

Recall that delaying shock exposure for some subjects in this sample was a

departure from other conditioning studies (in which subjects have routinely
been exposed to shock prior to the adaptation series). Although mean startle

responses of the two groups did not differ in the second block of adaptation,

responses of the Delayed group were relatively high within that group due to

their recent shock exposure; conversely, responses in the Immediate group

had decreased from their earlier postshock exposure (see Table 1). Thus, the

larger responses for Delayed subjects may not have fully habituated, making
it more difficult to detect potentiation.

2 Subjects viewing unpleasant slides had larger mean startle responses
(across CS+ and CS- pictures) than subjects viewing pleasant slides during the
adaptation series (Ms= 518 vs. 360 ADUs), F(1,60)= 3.18, p< .08. Covarying for
prelearning startle magnitude levels, however, eliminated this marginal
effect. These apparent slide foreground response differences could therefore
be attributed to initial subject response differences.








Overall analysis indicated that the CS+/shock pairing produced a mean

increase in startle magnitude from block 2 of the adaptation series to block 1

of extinction (A= +41 ADUs, from 443 to 484 ADUs), but this change was not

significant, F(1,62)= 1.85, p< .18. A closer look at the trials immediately
surrounding the acquisition phase (i.e., the last trial of adaptation and first

trial of extinction) showed that startle was significantly potentiated across all

subjects (A= +139 ADUs, from 414 to 553 ADUs), F(1,61)= 5.20, p< .03.

There was, however, a dramatic difference in startle reflex potentiation

between the two groups differing in the timing of shock exposure (Figure 7).

The Immediate exposure group showed startle potentiation (A= +83 ADUs),

whereas the Delayed group showed a commensurate decrease (A= -86 ADUs),

F(1,62)= 6.62, p< .02. Tested alone, the increase in the Immediate group was
significant, F(1,47)= 6.77, p< .02. Thus, in a statistical test directly comparable

with previous studies, fear-potentiated startle was replicated.

Overall mean CS+ response was significantly larger than CS- response

throughout extinction, F(1,62)= 5.84, p< .02, and this discrimination did not

vary over slide trial blocks (see Figure 8). Clearly, startle magnitude declined

overall during extinction, F(3,186)= 35.94, p< .0000. There were no Slide

Arousal or Shock Exposure group effects on extinction startle discrimination.

In two recent experiments, startle discrimination tended to be greater

among subjects viewing pleasant slides than unpleasant slides. In this study,

this effect failed to replicate, Slide Valence X Stimulus F(1,62)< 1 (Figure 9).

An effect of Slide Valence on CS+ startle potentiation (cf. Hamm et al., 1991)

was also not replicated, F(1,62)< 1. Consistent with these recent studies, the

Negative/Arousing group showed sma.llir mean startle potentiation than

other groups (-7 vs. 122, 15, and 32 ADUs for Negative/Calm, Positive/Calm,

and Positive/Arousing groups, rcspc'il\'l.v'). I however, statistical analyses of

























--O-- Immediate
-U- Delayed


Adaptation
block 2


Extinction
block 1


Blocks of 2 CS+ Trials




Figure 7. Extent of CS+ startle potentiation in groups exposed to
shock immediately before the adaptation series (open circles) or
delayed exposure to shock midway through the adaptation series
(dark squares).


520-


500-


480-


460-


440-


420-


400















475-


425-


375 -


325-


275-


-0- CS+
-CS-


I I
Adapl Adap2


Ext
Extl


Ext2
Ext2


Ext3
Ext3


Ext4
Ext4


Figure 8. Startle magnitude differentiation in the overall
sample (n=64). Mean responses are presented in blocks of
2 trials during the adaptation and extinction slide series.
Note that probe reflexes decrease in extinction to a point
which is below prelearning levels (dashed line), and that
CS- response reaches this level before CS+ response.


>19x:O














Pleasant Pictures


7,0


450


400


350-


300-


250


200


150



600


550


500


450


400


350-


300-


I I I
Ext2 Ext3 Ext4


Unpleasant Pictures


AdI I2
Adapl Adap2 Extl


Ext2
Ext2


I I
Ext3 Ext4


Figure 9. Startle magnitudes during adaptation and extinction
slide series of classical conditioning, separately for groups (n=32
each panel) viewing either pleasant slides (above) or unpleasant
pictures (below). Note the pattern of reflex habituation in each
group relative to prelearning levels (dashed lines).


IA
Adapl


----- CS+
---o-- CS-

I
Adap2 Extl


O'












































- I


fC'
ct


ore

4t4


LO
mn


(,vsuI),(:)ua~e-l 4asuo alijuls


+a t
C8


u








potentiation scores did not support these impressions (p> .20). Finally,
retesting Slide Valence effects only for the Immediate exposure group (which
is comparable with previous studies) did not change these null findings.3
The time at which probes were delivered influenced startle magnitude

in extinction. Startle response was generally larger to probes presented late
during the slide interval (regardless of CS+ or CS-) relative to earlier probes
(Ms= 358, 340, 374, and 387 ADUs for times 1-4). Variation across probe times
was significant, F(3,186)= 3.83, e= .90, p< .02; linear F(1,62)= 5.11, p< .03;
quadratic, p< .15. Further confirming this effect, mean response at the last
two probe times (381 ADUs) was greater than overall startle response at the
first two probe times (349 ADUs), F(1,63)= 8.29, p< .01. Startle discrimination
was consistent across probe times, Stimulus X Probe Time F(3,186)< 1.
Onset latency. There were no significant slide group, stimulus or trial
block effects for startle latency before aversive learning. (See Appendix D,
Table 16, for onset latency means in all groups and the overall sample.)
Startle onset latencies were evaluated for both learned facilitation and

discrimination, analogous to startle magnitude. Figure 10 illustrates startle
latencies from the end of adaptation (i.e., the base for facilitation) through
extinction. Due to missing data, nine subjects were excluded from these
analyses (n= 55).

3 Unexpectedly, Slide Valence influenced change in CS- startle response
from adaptation to extinction, F(1,62)= 4.63, p< .04. Probe response to CS- was
significantly potentiated (A= +82 ADUs) for Negative slide groups, F(1,31)=
4.16, p= .05, but CS- response did not significantly change (A= -25 ADUs) for
Positive groups. Although this finding suggests greater generalization of CS+
startle potentiation to CS- when subjects viewed unpleasant slides, it should
be viewed cautiously: Different patterns of CS- startle change were obtained in
previous experiments assessing affective modulation of the startle response
(Greenwald et al., 1990; Hamm et al., 1991).








From adaptation to the first block of extinction, CS+ latency was
speeded, whereas CS- latency was slowed. The interaction of these changes
was significant, F(1,51)= 4.41, p< .05. Whereas CS- slowing (A= 1.4 msec) was
significant, F(1,51)= 4.15, p< .05, CS+ speeding was not (A= 1.1 msec, p= .24).
Probe response to CS+ was faster overall than CS- response during extinction
(Ms= 32.4 vs. 34.1 msec), F(1,51)= 8.31, p< .01. Although CS+ onset latencies
returned to prelearning levels, discrimination did not significantly vary over
trial blocks. Onset latency of the reflex between slides (M= 39.0 msec) was
significantly slower than during slides.
Similar to startle reflex magnitude, mean onset latency discrimination
in extinction was smaller for the Negative/Arousing group than the other
slide groups (0.1 vs. 2.2, 1.5, and 2.5 msec for Negative/Calm, Positive/Calm,
and Positive/Arousing groups, respectively). However, the statistical test of
this impression was not significant (p= .19). There were no other remarkable
effects of affective slide content on modulation of startle onset latency.
Visceral Reactions

Skin conductance. Appendix D (Tables 17-18) presents First Interval
skin conductance responses for all slide foreground groups throughout
conditioning. Electrodermal response magnitude decreased from adaptation
trial 1 to 2, increased at trial 3, and decreased again from trial 3 to 4. This
within-block response habituation was significant, F(1,60)= 24.68, p< .0000.
The increase between adaptation blocks probably represents dishabituation
due to either shock exposure or completion of questionnaires (see above).
Analogous to startle potentiation, the Immediate exposure group
showed increased conductance response from the end of the adaptation
series to the first block of acquisition, whereas the Delayed exposure group's
response decreased (As= .051 and -.054 pS; see Figure 11). This difference in















030-


0.25





0.20-


0.15


-0- Immediate
Delayed


.4.


Adap2


Acql


Blocks of 2 CS+ Trials




Figure 11. Extent of CS+ skin conductance response potentiation
in groups exposed to shock before adaptation (open circles) or
delayed exposure to shock midway through adaptation (dark
squares).



















II ..-,




oo
u u




M 0 0
(U -.4
iO w











(U
Z &=4 8










oo d .






So o
t3



I gi=





JTb4 0 ;C4

0 0 0 ft8 0

r( 9 0









conductance potentiation was significant, Group X Trial Block F(1,62)= 3.96,

p= .05. The two groups did not differ in their skin conductance responses at

the end of adaptation.

In acquisition, First Interval Response magnitude to CS+ was larger

than to CS- overall (Ms= .24 and .15 pS), F(1,60)= 38.05, p< .0000 (Figure 12).

Conductance response magnitude habituated during learning (Ms= .22, .21,

.19, and .17 pS, for blocks 1-4), F(3,180)= 5.19, e= .84, p< .005. Second Interval
Response magnitude, reflecting shock anticipation, was greater during CS+

than CS- overall (Ms= .18 and .07 pS), F(1,60)= 47.96, p< .0000. Relative to

CS-, the Second Interval Response to CS+ increased monotonically across

trial blocks, F(3,189)= 5.47, e= .97, p< .002. Not surprisingly, unconditioned

responding to shock (i.e., Third Interval Response magnitude after CS+; M=
.50 pS) was dramatically greater than activity after CS- offset (M= .06 pS),

F(1,60)= 320.53, p< .0000. These unconditioned responses to shock declined

over trial blocks (Ms= .57, .51, .45, and .45 pS), F(3,189)= 4.27, e= .80, p< .01.

In extinction, the First Interval Response was greater for CS+ than CS-

(Ms= .14 and .09 pS; see Figure 12). This overall discrimination effect was

significant, F(1,60)= 11.65, p< .002. First Interval Response magnitude, which

declined from acquisition, continued to decrease in extinction. This general

decline across extinction trial blocks (Ms= .14, .12, .11, and .10 pS) was

significant, F(3,180)= 4.09, e= .99, p< .01.

Whereas both Hamm et al. (1991) and Greenwald et al. (1990) found that

electrodermal discrimination in extinction was larger in the context of calm

slides, neither Slide Arousal nor Valence influenced conductance response

differentiation in this experiment (both main effect Fs< 1). Consistent with

these past studies, conductance discrimination was virtually absent in the

Negative/Arousing group, relative to other rmrupps, in extinction (.01 vs. .06,









.04, and .08 pS in Negative/Calm, Positive/Calm, and Positive/Arousing
groups, respectively). This impression of an isolated group difference (as

suggested by the Slide Valence X Slide Arousal X Stimulus interaction),
however, was not supported, F(1,60)= 2.50, p< .12.

Electrodermal activity was larger in anticipation of shock onset during
extinction. The Second Interval Response to CS+ was significantly greater
than to CS- (Ms= .24 and .19 pS), F(1,60)= 11.14, p< .002, and habituated

overall (Ms= .33, .22, .17, and .16 pS), F(3,180)= 36.39, e= .93, p< .0000. Third
Interval Response magnitude (i.e., due to shock omission after CS+) was

small and undifferentiated in extinction.

Heart rate. Tables 19-22 in Appendix D present heart rate level and leg

scores, by slide foreground group, throughout the experiment. Adaptation
prestimulus heart rate levels tended to be higher for Positive slide groups
than Negative groups, F(1,59)= 3.68, p< .06. Levels were also higher prior to

the CS+ slide than the CS- slide for the Negative groups, Slide Valence X
Stimulus F(1,59)= 5.01, p< .03. Because heart rate levels varied within and

between groups, analyses of leg scores during the adaptation series were

conducted using baseline heart rate for each CS at each trial block within

each affective valence group as covariates.

In adaptation, there was significantly less cardiac acceleration (Al leg)

for CS+ than CS- (Ms= 2.55 vs. 4.28 bpm), F(1,59)= 9.62, p< .005, which was

not altered by covariance analysis, F(1,58)= 5.75, p< .02. Initial deceleration

(Dl leg) and late-slide deceleration (D2 leg) tended to be larger for CS+ than
CS-, F(1,59)= 3.72, p< .06, and F(1,59)= 4.06, p< .05, but adjusting for tonic

heart rate attenuated both findings in both instances (ps> .35).

In acquisition, cardiac acceleration was the best index of learning. As
Figure 13 suggests, the initially smaller acceleration to CS+ (originating in




























;- 0


9.


C C? <


-4 ('. et 4 9 .


+ I
U U
WOl


C C


--o


S0

wU
ogy








0o










& >

B as JU


- S




-in


C4~










-LL
0






-C4


(udq) agueqD alet Je-aH








adaptation, but not significantly different from CS- response in acquisition
block 1) gradually increased; there was a concomitant decrease in acceleration

to the CS- slide over trials. Although the overall learning effect (i.e., CS+

more acceleratory than CS-) was not significant, F(1,59)= 1.84, p= .18, there
was a linear increase in discrimination with repeated slide/shock pairings.
The Stimulus X Trial Block interaction was marginal, F(3,177)= 2.60, e= .93,

p< .06, but the difference in CS+ vs. CS- linear trends across blocks was

significant, F(1,62)= 7.99, p< .01. Thus, associative learning reversed the

initial disparity in slide cardiac response from the adaptation series and did

so monotonically. Furthermore, the learning effect was a function of both

incremental acceleration to CS+ and declining acceleration to CS-.4

Unconditioned cardiac acceleration (A2 leg) to shock after CS+ offset

was greater than CS- postslide response (Ms= 10.7 vs. 5.8 bpm), F(1,59)= 45.62,

p< .0000. The unconditioned response diminished over trial blocks (Ms=

11.6, 11.0, 9.9, and 7.0 bpm), whereas cardiac response after CS- stayed smaller

and constant. This differential habituation effect was also significant,

Stimulus X Trial Block F(3,180)= 3.28, e= .94, p< .03.

In extinction, late cardiac deceleration (D2 leg) was greater during CS+

than CS- (Ms= -5.2 vs. -4.2 bpm), reflecting anticipation of shock, F(1,59)= 4.78,

p< .04. There were no other significant leg score findings during extinction.

4 Groups viewing arousing slide foregrounds showed the increasing
cardiac accelerative differentiation more than Calm slide groups, Slide
Arousal X Stimulus X Trial Block F(3,183)= 2.76, e= .93, p< .05. This was due
to incremental learning across blocks in the Arousing slide groups but also
reversed conditioning for Calm groups in trial block 2 (see Appendix D, Table
21). However, modulation of cardiac acceleration by slide foreground has not
been observed in previous conditioning experiments; its appearance here
may be fortuitous.








Positive slide groups, however, had higher heart rate levels than Negative

groups, F(1,59)= 4.03, p< .05.
Summary
Clear, significant conditioning effects were obtained across a variety of

measures. For the startle probe reflex, magnitude was potentiated during

CS+, especially for the Immediate exposure group, and was larger than CS-

throughout extinction. Similarly, CS+ startle latencies were faster than CS-

in extinction. For visceral reactions, skin conductance response during CS+

was rapidly potentiated, especially for the Immediate exposure group, and

remained larger than that of CS- during both acquisition and extinction. In

contrast, cardiac accelerative conditioning proceeded more slowly, with CS+

response gradually exceeding CS- response late in acquisition; late-slide

deceleration was greater for CS+ than CS- in extinction. For affective valence

judgments, CS+ became more unpleasant (relative to CS-) after learning,

especially for subjects who viewed pleasant slides; conditioning produced no

systematic changes in arousal ratings overall. In general, startle and visceral

responses were not significantly modulated by the affective slide foreground,

but valence and arousal judgments were influenced by this manipulation.

Individual Differences

The relationships of temperament characteristics, shock tolerance, and

rated shock aversiveness to stimulus reactivity and associative learning were

assessed. Temperamental influences are treated as temporal precursors to

experimental variables in the subsequent analyses.

Temperament Characteristics

Consistent with Gray's (1982) broad hypothesis that neurotic introverts

are more sensitive to aversive stimulation, the EASI fear and impulsivity

temperament dimensions accounted for significant variability in aversive








reactions; in contrast, dimensions of sociability, activity, and anger did not
predict measures of aversive reactivity or learning. Figure 14 illustrates the
distribution of fear X impulsivity scores for all subjects and cutpoints used in
subsequent group analyses. Fearfulness and impulsivity subscale scores were
independent in this sample of subjects (r= .09, ns) and the cutpoint-defined
groups did not differ on the other temperament dimensions.
Reactivity to aversive stimulation. Each person set his/her level of
shock to be "highly annoying but not painful" during initial exposure.
However, the physical intensity of shock at this subjective level--hereafter,
"shock tolerance"--varied widely and was significantly lower for subjects
whose temperamental disposition was fearful and inhibited.
Stepwise regression (Table 4) indicated that fearfulness and inhibition
predicted lower tolerance for shock, multiple r= -.45. Group analysis (see
Table 5) revealed a significant main effect for inhibition, F(1,60)= 5.45, p< .03,
a marginal effect for fearfulness, F(1,60)= 2.98, p< .09, and no interaction (p=
.84). In follow up tests, only highly fearful/inhibited and low fear/impulsive
subjects differed from one another in shock tolerance.
Temperamental fear and inhibition also predicted prelearning startle
magnitudes which were significantly larger (multiple r= .32). Group analysis
indicated a significant fearfulness effect, F(1,60)= 4.01, p< .05, a marginal effect
for impulsiveness, F(1,60)= 3.74, p< .06, and no interaction (p= .87). Again,
follow up tests showed that only fearful/inhibited subjects and nonfearful/
impulsive subgroups significantly differed from one another. (Mean startle
magnitude of fearful subjects continued to be higher overall throughout
conditioning, but this difference was not significant). Consistent with the
startle magnitude data, mean prelearning startle onset latencies of fearful










Fearful

0 1


0-20
I
0 0 000 -0
0 10


(b
0 1o (C 0
-- r -T
0


oj


106


00

0
(-Q


20
~6


0
Impulsive
Impulsive


0 0
0


0


0 j,

000i
(b (-10o


01


-15
Nonfearful


Figure 14. Distribution of fearfulness and impulsivity scores (EASI Temperament
Survey) for all subjects. Placement of dashed axes indicates that cutpoints of 14 on
each dimension which were used in determining subgroups for statistical analyses.


5
Inhibited


00







Table 4

Stepwise multiple regressions of temperamental fear and inhibition
(low impulsivity) on measures of aversive reactivity.


Dependent Multiple R
Variable Fearful Inhibited Statistical Summary


FSS total score .48 F(1,62)=18.31, p< .001
Shock tolerance -.33 -.45 F(2,61)= 7.86, p< .001
Prelearning startle reflex:
magnitude .22 .32 F(2,61)= 3.46, p< .05
onset latency .25 F(1,60)= 3.88, p< .06







Table 5

Means and standard deviations for measures of aversive reactivity
as a function of temperamental fear and impulsivity subgroups
(cutpoints = 14 for each subscale).


Low Fear High Fear

Dependent Inhibited Impulsive Inhibited Impulsive Mean
Variable n= 20 n= 10 n= 20 n= 14 n=64


FSS total score 87.2 b 105.3 ab 112.0 a 107.6 a 102.2
(40-200 scale) (16.4) (28.5) (19.0) (19.6) (22.3)

Shock tolerance 6.1 ab 8.3 a 4.2 b 6.8 ab 6.0
milliamperess) (4.1) (4.5) (3.1) (4.5) (4.1)

Prelearning
startle reflex:
magnitude 309.9 ab 131.6 b 465.9 a 315.8 ab 332.1
(A-D units) (243.4) (83.3) (364.3) (453.2) (336.9)

onset latency 41.6 a 42.8 a 40.2 a 37.6 a 40.4
(msec) (8.8) (7.2) (7.5) (8.1) (8.0)


Note:
Means without common letters significantly differ (p< .05; using Tukey HSD,
MSerror [df= 4,60] and harmonic mean= 14.74).

























































1 LJ
a ..

ct a 44
IE
1J- 0
h-Il


6 (D i a .







-- -~ Ue e'
J a-
a P > E 0



1 o E 4

at .5 us







to
cu t -a








.i 0





uo,


* w .1
(2) L" x
1 t z c0




0 x
F isc


1 1

li~ ~. .f








subjects were marginally faster than for nonfearful subjects (r= .25). Group
analysis yielded a somewhat weaker fearfulness effect, F(1,58)= 2.33, p= .13.
As was expected, temperamentally fearful subjects (independent of
impulsivity) reported more fear across a variety of situations (Fear Survey
Schedule; r= .48). This was confirmed by a significant difference in the group
analysis, F(1,60)= 6.69, p< .02. Follow up tests indicated that the nonfearful/
inhibited group reported less fear than other subgroups. Finally, fearful
subjects also tended to rate their shock as more aversive (r= .20, ns) which, to

some extent, explains the counterintuitive finding that less shock tolerance
was related to significantly higher shock aversion in the overall sample (r=

-.35).
Figure 15 summarizes the relationships between temperament, low

shock tolerance, high shock aversion, and high situational fear. This pattern
of data supports the conclusion that fearful/inhibited subjects are more
generally reactive to aversive stimulation--in terms of reported situational

fears, behavioral tolerance of shock, and reflex excitability.
Sensitization. Temperament characteristics were not significantly

related to sensitization learning measures.
Associative learning. Temperamentally impulsive subjects exhibited

more skin conductance response potentiation and discrimination during
acquisition. Both effects were supported statistically by regression analysis,

Fs(1,62)= 4.36 and 6.67, ps< .05. Correlations between impulsivity and both
conductance potentiation and discrimination were modest but significant
(rs= .26 and .31). Fearfulness, entered stepwise in these same regressions,
did not significantly predict electrodermal learning. Temperament did not
interact with slide affective foreground in determining aversive learning.









Rated Shock Aversiveness
The role of the shock in mediating aversive learning was explored next,

using stepwise multiple regression. Rated shock aversiveness and
behavioral shock tolerance served as predictors and psychophysiological
measures of learning (sensitization, potentiation, and differentiation scores)

were the dependent measures. Interestingly, shock tolerance did not
significantly relate to any of these learning measures.
Sensitization. Rated shock aversiveness was not significantly related to

sensitization.

Associative learning. In contrast, shock aversion systematically affected

associative learning (see Figure 15). One group of subjects judged the shock

as highly aversive (rating of 7 on the I 7 scale; n= 27). Others, forming a
Moderate Aversion group (n= 20), rated shock aversiveness equal to 6. The
remaining subjects (n= 17), with aversion ratings from 3 to 5, constituted the

Low Aversion group. SAM ratings of the shock were each highly correlated

with shock aversion (pleasure, r= -.57; arousal, r= +.51). Therefore, reported
"aversion" can be decomposed into elements of displeasure and arousal.

Stepwise regression indicated that greater shock aversiveness predicted

significantly greater CS+ aversion, F(1,62)= 17.59, r= .47; startle potentiation,

F(1,62)= 13.98, r= .43; and skin conductance response potentiation, F(1,62)=

6.87, r= .32. Table 6 presents mean data from the group analyses. The group
mean difference in rated CS+ aversion was significant, F(2,61)= 7.84, p< .001.
Follow up tests found that High and Low Aversion groups significantly
differed; these extreme groups both differed marginally from the Moderate

Aversion group (ps< .07). However, the three groups did not significantly

differ in CS- aversion ratings.








Shock Aversion groups differed overall in startle potentiation, F(2,61)=
7.83, p< .001 (see Figure 16). In t-tests, High and Moderate Aversion groups
did not differ (As= +120 vs. +82 ADUs) and both showed significantly more
potentiation than Low Aversion subjects (A= -134 ADUs).5 The lack of
startle potentiation in the Low Aversion group was not due to more of these
subjects receiving delayed shock exposure (see Appendix D, Table 23).
Regression analysis indicated that shock aversiveness was not significantly
related to startle discrimination.
Subjects who rated shock as more aversive showed larger conductance
response increases to CS+ from adaptation to acquisition (r= .32; see Figure
17). The Aversion group difference in conductance response potentiation
was significant, F(2,61)= 3.59, p< .04. As with startle potentiation, electro-
dermal response potentiation for Moderate and High Aversion groups (As=
.04 and .08 pS) did not differ and changes in both groups were significantly
greater than the decreasing response in the Low Aversion group (A= -.07 pS).
Higher shock aversion was marginally related to better skin conductance
response discrimination in acquisition, Aversion X Stimulus F(2,61)= 2.54,
p< .09, but not during extinction.
High Aversion subjects showed greater heart rate acceleration to CS+ in
the final trial block of acquisition. However, the importance of this effect
was diminished by a nearly complete reversal of this effect in the preceding
trial block (see Appendix D, Table 24, for data and a brief discussion).

5 Although Figure 16 might suggest that the Low Aversion group had
higher CS+ startle responses than the other two groups before learning, the
group difference was not significant (p= .12). Thus, a direct comparison of the
amount of potentiation across groups is permissible.







Table 6

Means and standard deviations for measures of associative learning
as a function of shock aversiveness (see Figure 15 for correlations).1

Shock Aversion Group

Dependent Low Moderate High Mean Refer to
Variable n=17 n=20 n=27 n=64 Figure:


CS+ aversiveness 4.5 b 5.5 ab 6.2 a 5.5
(1-7 rating scale; (1.5) (1.6) (1.0) (1.7)
postlearning)

CS- aversiveness 1.8 a 2.2 a 2.5 a 2.2
(1-7 rating scale; (1.1) (1.5) (1.7) (1.8)
postlearning)

Startle potentiation -134 b 82 a 120 a 41 16
(A-D units)2 (273) (98) (235) (274)

Skin conductance -.07 b .04 a .08 a .02 17
potentiation (pS)3 (.11) (.17) (.21) (.20)

Change in CS+ -1.7 a -1.2 a -3.2 b -2.2 18
pleasure4 (0.8) (1.5) (1.8) (1.7)

Notes:
1. Means without common letters are significant different (t-tests; p< .05).
2. Change in CS+ response from adaptation block 2 to extinction block 1.
3. Change in CS+ response from adaptation block 2 to acquisition block 1.
4. Change in CS+ valence from pre- to post-learning, for subjects viewing
pleasant slides only.






71












0 >.


c 0










(U
0









0 '-4*




o .
-^ -ci














Cu '
00


o 0











o 1o



in (.
0 .

















.'.&




(4 C- (
10v0 10 *'1' -'I
_________________________________________. '-^
















8-

S0 0
.o / o


S\ / oxs

S< _

ca 0

tu
0i)







( .
0E +o
A.,
I) \ I



-0



bc

o ap
0oo 0

0 C il

'i e.-

o 0P a
in ml



(Eu

(Sd I -I+ jI Bol) apnimj) asuodsaN aluel:npuoD uijS























+
UU
12


I


'a


I I --


-


(sIui-i sSuWeIP) agueiq3 azualeA








The decrease in CS+ pleasure from pre- to postlearning (i.e., negative
change scores indicating valence potentiation) tended to be greater for High
Shock Aversion (A= -1.4 units) than Moderate or Low Aversion subjects
(both As= -0.4 units) in the overall sample, Aversion Group X Stimulus,

F(2,61)= 2.88, p< .07. However, as seen in Figure 18, high shock aversiveness
potentiated CS+ valence change only for subjects viewing pleasant slides,

Slide Valence X Shock Aversion F(2,58)= 3.18, p< .05. Post hoc tests showed
that, for unpleasant slide foregrounds, shock aversion did not systematically
modulate CS+ valence change. But among subjects viewing pleasant slides,

subjects highly averse to shock judged their CS+ as more unpleasant after
learning, F(2,29)= 5.73, p< .01. Figure 18 also illustrates, in comparison, that

CS- change was not modulated by rated shock aversion.
Shock aversiveness did not interact with affective slide content in
determining psychophysiological response during associative learning.
Frequency judgments of shock occurrence. Subjects underestimated the
number of shocks they received during conditioning (i.e., excluding the

shock exposure procedure). Although each subject experienced 8 shocks, the
mean estimate was 6.5 times. This discrepancy from actual exposure was
significant, F(1,61)= 29.22, p< .0000. Although no individual difference

factors were reliably associated with shock frequency estimates, subjects

reporting less aversion of shock tended to under-report frequency of shock

occurrence (r= .23, p< .10). Frequency judgments of shock occurrence were

not influenced by the affective slide foreground during conditioning.
Contingency recall. Eighty-six percent of subjects clearly recalled their

CS+ when probed by the following questions: (1) Did you recognize a rule in

this experiment?; (2) Were some stimuli followed by shock more often than

others?; and (3) If yes, which? An additional four subjects did not clearly








understand the probe questions but, when clarified, correctly identified their
CS+. (It is not clear whether the experimenter inadvertently cued the subject
in these cases; thus, it is not certain whether correct identification was a

recall or a recognition decision. If these four subjects are included in the
correct recall category, the recall rate becomes 92%.)
Four subjects could not report which slide had been shocked and a fifth

subject incorrectly believed that CS- had been shocked. Interestingly, four of

these five subjects were in the Low Shock Aversion group (the other was in
the High Aversion group). However, due to the rather small subsample,

these data were not subjected to statistical analysis. Contingency recall was
not related to affective slide content. Re-analysis of the data set, excluding
subjects who could not recall the contingency, did not change the findings.
Summary
The correlational pattern shown in Figure 15 supports several major

conclusions. First, temperamental fear and inhibition modulated sensitivity
to aversive stimulation, but did not generally determine aversive learning.

Second, the perception of shock aversiveness--and not behavioral tolerance

of shock intensity--predicted the degree of associative learning. Third, these

relationships were present only for measures of simple (but not differential)

association. Finally, whereas rated shock aversiveness directly influenced
physiological measures of learning, CS+ aversiveness showed similar, albeit

weaker, relationships.













DISCUSSION

The present results indicate that startle reflex magnitude was sensitized
after exposure to shock and was also potentiated during a stimulus which
was paired with shock (relative to a stimulus not paired). In short, startle
magnitude is a reliable index of both nonassociative and associative learning
when an aversive reinforcer is used. Whereas startle onset latency and

visceral (heart rate and electrodermal) reactions also reflected associative
learning, these measures failed to index the sensitization learning process
which is presumed to occur when subjects are exposed to shock. Consistent
with the startle reflex data, ratings of affective valence (but not arousal) for
the conditioned stimuli reflected overall associative learning.
In addition, individual differences in aversive reactivity and learning

were identified. First, subjects who were both temperamentally fearful and
inhibited were more sensitive to aversive stimulation (i.e., they showed less

tolerance of shock, larger initial startle responses, and reported more anxiety

across a variety of situations) than other subjects. However, temperament
characteristics were not generally related to learning measures. Second,

perceived aversiveness of the shock US--and not physical intensity of shock--

modulated the degree of learning in this experiment.

Sensitization

Startle Magnitude
In this experiment, a brief series of shocks was administered to each
subject to calibrate US intensity just below the subject's reported pain
threshold level. Shock exposure prompted an immediate, vigorous (40%)








increase in startle magnitude on the first postexposure trial, whereas this

effect was absent in control conditions. Thus, significant startle sensitization
developed very rapidly and could not be attributed to either dishabituation
or the simple passage of time. In addition, the increase in startle magnitude
from pre- to postexposure in the Immediate group--for whom duration of

this sensitization effect could be more completely measured--persisted
throughout adaptation (ca. 10 mins). The present findings are consistent

with the animal literature on shock sensitization of startle (Boulis & Davis,
1989; Davis, 1989; Hitchcock et al., 1989). To the extent that similar results
were obtained in recent experiments using human subjects (Greenwald et
al., 1990; Hamm, Stark, & Vaitl, 1990), it may be concluded here that shock

sensitization of the startle response is a robust phenomenon (i.e., large

magnitude, rapid development, relatively persistent, and independent of

species, dishabituation, and time).
In contrast, neither startle latency nor visceral activation (i.e., tonic

levels or slide responses) was reliably facilitated after shock exposure. Taken

together, the data suggest that shock exposure prompts an aversive

preparatory set--uniquely indexed by the startle reflex (Lang et al., 1990)-

which, lacking visceral arousal, may not require abrupt mobilization for

action. Consistent with this idea, freezing has been shown to emerge as the

dominant tactical response following shock exposure in animals (Fanselow,

1980; Fanselow & Bolles, 1979); furthermore, the extent of motoric inhibition

correlates highly positively with increased startle response in individual

subjects (Leaton & Borszcz, 1985). Shock exposure can also influence

behavior in future aversive situations. For instance, pre-exposing animals

to a brief series of relatively intense shocks facilitates "fear" of subsequent

aversive stimulation, as measured by suppression of operant rates (Pearl,








1963; Pearl, Walters, & Anderson, 1964) and avoidance in a conflict test
(Kurtz & Walters, 1962; Walters, 1963).

Current evidence (Lang et al., 1992) suggests that startle augmentation
reflects the motivational state of the subject, specifically the presence of an

aversive response disposition. The selective effect of shock exposure on

sensitization of startle magnitude (i.e., in the absence of visceral activation)
obtained here, as well as the varied effects of shock exposure on behavior

just described, encourages the conclusion that the startle response is not
tactically specific (i.e., associated with a particular behavioral act) but, rather,

a more general measure of aversive processing.
Methodological Considerations

One question which cannot be answered from this experiment is

whether actual exposure to shock is necessary to obtain performance
changes, since exposure is confounded here with the knowledge that shock
will be presented. Thus, it remains possible that instructed exposure can

produce sensitization. Several researchers have, in fact, examined reactions

to threat of shock. Chattopadhyay, Cooke, Toone, and Lader (1980; Exp. 1)

conducted a study in which subjects were threatened or not threatened with

shock (within-subject, sessions separated by 1 week; order counterbalanced

across subjects). They found that electrodermal levels and anxiety ratings

increased during threat sessions, relative to no threat. Unfortunately, the

authors did not report whether responses increased relative to resting

baselines within the threat session, nor whether they conducted a between-

group analysis of the first session. Thus, it is not clear whether within-

session changes were significant, or whether between-session (i.e., threat

order) effects influenced the results. Farha and Sher (1989) also showed that

autonomic measures, somatic activity, and reports of anxiety all increase








after shock threat (relative to a "naive" baseline). Unlike the present study,
their experiment did not include a control group. Therefore, neither
dishabituation (i.e., after the experimenter re-entered the testing room) nor
the sheer passage of time can be ruled out as potential explanations of their
findings.
Unlike the studies above, shock exposure did not sensitize either cardiac
or electrodermal activity here. There are several possible explanations for
this discrepancy. First, it could be argued that the test of shock sensitization
of visceral activation was not powerful: Pre-exposure tonic levels and slide
responses were not available for the majority of subjects (i.e., Immediate
exposure group), yielding a weaker test in a smaller subset of subjects (i.e.,
Delayed exposure group). On the other hand, although conductance levels
and slide responses slightly increased (nonsignificantly) after shock exposure
in the Delayed group, levels and responses in the control condition (i.e.,
Immediate group) also increased. These data suggest that the increases in the
Delayed group were not specific to shock exposure and could be accounted
for by another factor (e.g., dishabituation). A second explanation is that
exposure to shock and threat of shock may differentially affect visceral

activity. During shock threat, the subject might anticipate aversive
stimulation and thus maintain visceral activity at higher levels. During
shock exposure, however, visceral activity may temporarily increase (before
it could be measured here?)1 and then rapidly habituate. The present heart

1 There was clearly an unconditioned reaction to shock presentation.
However, the issue here is whether shock exposure generated a fleeting
increase in visceral tonic level before the experimenter could continue the
procedure to measure such a change. In fact, there was a period of about 3 to
5 minutes after setting the final shock intensity when the experimenter had
to instruct the subject on the subsequent conditioning procedure, and initiate








rate tonic level data are partially consistent with this interpretation, since

levels of the Delayed group tended to increase after exposure, whereas

control levels significantly habituated. A third, equally plausible,

explanation is that the results of other studies may arise from uncontrolled
influences discussed above. In future work, one would ideally separate the
incremental effects of threat and exposure, using a paradigm similar to that

of Farha and Sher (1989), while including a control group and continuous

monitoring of visceral activation.

Finally, it is interesting that during the initial control period for shock

exposure (Delayed group) there was a mild, but nonsignificant, increase in
startle magnitude. This increase is not inconsistent with previous findings:
Some habituation experiments (Groves & Thompson, 1970; Szabo & Kolta,

1967) have shown that the startle reflex elicited by an intense acoustic probe

(in the absence of shock) may briefly increase before habituation occurs.

Thus, it is possible that reflex sensitization in the Immediate exposure group

reflects summation from two events--a strong, persistent increase from

shock exposure and a weaker, probe-related increase. One suggestion for

future research would be to repeat the present study design, adding a longer

preshock series of startle probes to induce certain response habituation. This

would insure that the amount of sensitization following shock exposure

could not be attributed to characteristics of the probe itself.

Associative Learning

Magnitude of the eyeblink reflex to acoustic probes presented during the

CS+ slide was larger at the start of extinction, relative to adaptation. Startle


the computer program. Thus, if visceral changes did occur, they could have
subsided by the time the procedure continued.








reflexes evoked during CS+ were also significantly larger and faster in onset

than those during CS- or interslide intervals in extinction. Additional
performance changes--including skin conductance response potentiationn
and discrimination), differential cardiac acceleration, and affective ratings
change--provide converging support for the conclusion that aversive
learning occurred. The present findings thus replicate and extend results

from single cue (i.e., nondiscriminative) conditioning experiments with

animals (Berg & Davis, 1985; Brown, Kalish, & Farber, 1951; Cassella & Davis,

1986; Davis & Astrachan, 1978; Kurtz & Siegel, 1966; Leaton & Borszcz, 1985)

and recent differential learning experiments with humans (Greenwald et al.,

1990; Hamm et al., 1991; Hamm, Stark, & Vaitl, 1990).
Sensitization and Habituation in Associative Learning

The differential learning paradigm was used in this experiment because

it controls for nonassociative influences such as (within-session) habituation

and sensitization. Since the assessment of associative learning involves

comparing CS+ and CS- responses, it is important to rule out nonassociative

factors as alternative explanations.

Sensitization, habituation, and potentiation. CS+ startle magnitude was

potentiated early in extinction (relative to adaptation), whereas interslide

responses decreased over this same period. Recent experiments (Greenwald

et al., 1990; Hamm et al., 1991) also demonstrated this same pattern. If both

CS+ and interslide responses had increased, this would have suggested that a

nonassociative factor influenced potentiation. Thus, the data indicate that

the CS+ increase was an associative effect (since shock was never paired with

the absence of slides). By similar logic, sensitization cannot account for CS+

skin conductance response potentiation, since CS- response decreased from

adaptation to acquisition. Also, sensitization cannot account for heart rate








accelerative changes during training, since CS+ and CS- responses developed
slowly and in opposing directions.
For subjects whose exposure to shock preceded adaptation in this study
(as in past experiments), startle probe responses habituated before learning
and there was a clear replication of the startle potentiation effect. In contrast,
for the subgroup whose exposure to shock was delayed, startle potentiation
was not only absent but significantly decreased from adaptation to extinction.
These data suggest that habituation of startle responses in the Delayed group
prior to acquisition was not as advanced which, in turn, may have masked
startle potentiation. It is therefore possible that the absence of startle
potentiation in the Delayed exposure group may be a methodological artifact
resulting from insufficient habituation. A similar pattern was observed with
skin conductance response potentiation, i.e., initial pairing of the CS+ and
shock produced increased conductance response in the Immediate group, but
CS+ response in the Delayed group decreased--even though the two groups'
slide responses prior to shock pairing did not differ. An obvious remedy for
this artifact (if one were to repeat this procedure) would be to extend the
adaptation series, thus insuring response habituation in the Delayed group.
Habituation and discrimination. The psychophysiological measures of
learning employed here--especially startle magnitude and skin conductance
response--clearly habituated during the procedure, with the exception of
transient potentiation due to the shock pairing. Yet there was significant
discrimination between CS+ and CS- in extinction for startle (magnitude and
onset latency), skin conductance, and late cardiac decelerative responses and,
during acquisition, for heart rate acceleration and electrodermal responses.
These data suggest that the associative effect during and after learning was
specific to CS+. Furthermore, CS+ and CS- responses did not extinguish at








different rates, i.e., response decline was parallel for the CS+ and CS- slides,
to the extent that Stimulus X Trial Block interactions were not obtained for
these measures. Thus, discrimination learning is assessed here with
habituation as a constant, background factor.
Bradley, Cuthbert, and Lang (1990a) demonstrated that the startle reflex
habituates with repeated probe presentations within a session, a result which
presumably reflects decremental activation in the obligatory startle circuit.
They also found that the affect-startle effect was present throughout this
general habituation, attesting to persistent activity in the modulatory circuit.
In the present experiment and other recent human conditioning studies,
CS+/CS- response discrimination throughout extinction similarly reflects
the continued influence of the modulatory system--as induced by associative
learning.
The pattern of cardiac acceleration was different from startle reflex and
electrodermal measures during associative learning. In acquisition, heart
rate acceleration during the CS+ slide increased with repeated pairings
whereas acceleration to CS- decreased with repeated nonreinforcement. This
cardiac pattern suggests that aversive learning may influence the heart
rhythm in such a way as to reflect both excitatory and inhibitory influences.
However, slightly different results were reported by Hamm et al. (1991), who
found that cardiac response was more acceleratory overall, that CS- response
became more accelerative over trials with CS+, and that the size of
differential acceleration was slightly larger than in the present study. At first
glance, these differences suggest that Hamm et al. produced a more powerful
manipulation (e.g., a more intense shock); however, the size of the startle

reflex and skin conductance learning effects are virtually identical in the two
studies. Therefore, it does not seem warranted to conclude that the present








data reflect a less effective manipulation. A more likely explanation is that
cardiac acceleration is simply less reliable than other measures of slide
response (Lang, Greenwald, Bradley, Hamm, in press). Thus, specific aspects
of the cardiac pattern may be less likely to replicate.
Finally, there was evidence that (1) affective ratings habituatee" (regress
towards neutrality) with stimulus repetition and that (2) CS+ valence and
arousal judgments generalized to CS-. On the first point, recall that adjunct
slides varying in affective content were rated pre- and postlearning, but were
not subjected to conditioning. Mean valence judgments of the pleasant and
unpleasant control slides became more neutral, whereas initially neutral
slides showed no mean change. Similarly, arousal ratings of adjunct slides
became moderate (i.e., calm slides became more exciting and vice versa).
However, changes in CS+ valence ratings (particularly for Positive groups)
and CS+ arousal ratings (especially for Calm slide groups) were greater than
changes in ratings for control slides, suggesting that learning played a greater
role in affective change than regression to the mean.
On the second point, Figure 5 clearly shows that affective ratings of CS+
generalized to CS-. This was more the case for arousal than valence ratings:

Rated arousal, similar for CS+ and CS-, was the single performance measure
which did not reflect aversive learning. In contrast, both discrimination and

potentiation of valence ratings were obtained. Although it is not clear why
arousal ratings did not differentiate the stimuli here--Hamm et al. (1991) and
Greenwald et al. (1990) found significant ratings changes along both affective
dimensions--one hypothesis is that both earlier studies used neutral/calm
CS slides, which yielded the greatest arousal change of any affective slide
type. These neutral CS slides, because of the selection of extreme affective
stimuli here, were largely excluded.








Affective Associative Change
Results of this experiment strongly support the main hypothesis that
aversive learning modifies the principal motivational (valence, arousal)
parameters of emotional response. Startle reflex and electrodermal changes
(both potentiation and discrimination), cardiac acceleration, and affective
ratings change were all significantly modulated by the classical conditioning
procedure. The results do not, however, support the secondary hypothesis
that the affective quality of the slide foreground prior to learning constrains
the degree of associative change. More specifically, the physiological
measures used here do not supply information on the extent of affective
associative modulation.
Affective modulation by specific content. As noted before, Hamm et al.
(1991) found that the amount of startle potentiation (and discrimination)
varied linearly with a priori valence of slide content: Less modulation was
observed for subjects who viewed unpleasant slide contents (mutilations,
threatening animals) than for subjects who viewed pleasant contents (nature
scenes, erotica). In addition, electrodermal discrimination varied linearly
with a priori slide arousal: Greater discrimination was observed for subjects
who viewed calm slides (household objects) than for subjects who viewed
the most arousing slides (mutilations, erotica).
Greenwald et al. (1990) attempted to replicate these associative effects,

by assigning subjects to view one of six slide content pairs distributed in the
valence X arousal space. The results of that experiment showed two changes
in the original startle modulation effect: (1) Discrimination, more so than
potentiation, tracked the amount of valence change; and (2) Rather than a
difference in magnitude, a smaller proportion of subjects who viewed

unpleasant slides showed startle discrimination than those viewing pleasant








slides. Electrodermal discrimination was also not a linear function of slide
arousal. Such differences across studies do not strongly support the affective
modulation hypothesis or, perhaps, suggest that other factors (e.g., specific
content) could mediate these effects. In contrast, changes in affective ratings
here supported findings of both earlier studies, namely that (1) CS+ valence
potentiation was greater for subjects viewing pleasant (than unpleasant)
slides and (2) CS+ arousal potentiation was greater for subjects viewing calm
(than arousing) slides. One problem with these differential ratings changes,
however, is that they may arise from floor and ceiling effects (i.e., initially
unpleasant slides cannot be rated much more unpleasant, and initially
arousing slides cannot be rated much more arousing, after aversive
learning).
Modulation along affective dimensions. In the present experiment,
stimulus content was deliberately obscured by choosing slides based on each
subject's extreme ratings. Inspection of Appendix A indicates that many
slide contents defined each of the four affective categories. Thus, it appears
that the effects obtained by Hamm et al. (1991) and Greenwald et al. (1990)
may have partly depended on specific slide contents.
The present approach to stimulus selection is more consistent with the
dimensional view of emotion, rather than relying on specific slide content to
influence the degree of associative change. Even with this more sensitive
approach, the results did not clearly support the affective change hypothesis.
On the other hand, subjects who viewed unpleasant/arousing slides tended
to show less differential learning than the other groups--not only in startle
magnitude and onset latency, but for electrodermal response, valence and
arousal ratings (Appendix D). Thus, a weaker version of the hypothesis
proposed by Hamm et al. (1991) could be advanced, in which a CS+ whose








affective character prior to learning most closely resembles that of shock (i.e.,
highly unpleasant and arousing) will undergo marginal associative change,
relative to any other effectively dissimilar stimulus. This notion-which is a
variant of the "delta rule" (Rumelhart et al., 1986)--has been applied to the
problem of affective change in conditioning elsewhere (e.g., Rescorla &
Wagner, 1972). In the present case, this hypothesis predicts a Stimulus X
Valence X Arousal effect (for discrimination scores) or a Valence X Arousal X
Trial Block effect (for potentiation scores). Despite the consistency of mean
differences across measures, however, no statistical test yielded a significant
interaction.
Temporal Specificity and Preparatory Processes
Startle response. Several studies have demonstrated that associative
learning is temporally specific, since startle potentiation is largest nearest the
CS/US training interval (Davis, Schlesinger, & Sorenson, 1989; Siegel, 1967)
or when shock is expected to occur (Grillon, Ameli, Merikangas, Woods, &
Davis, in press). Grillon et al. signaled subjects to expect shock in one

("threat") period but not in another ("safe") period. In addition, subjects

were told that shock would only occur during the last 10 sec of the threat
period. Grillon et al. found that startle was significantly potentiated during
threat (relative to safe) periods, and was maximal during the last 10 sec of the
threat period. Interestingly, this effect was obtained before shock was ever
administered and improved over trials.

In the present study, startle magnitudes generally (across CS+ and CS-)
increased linearly over the CS/US interval in extinction. Thus, temporal
sensitivity of probe response was demonstrated here. However, the lack of
differentiation between CS+ and CS- along this temporal gradient indicates
that stimulus generalization also occurred. A similar, but non-significant,








increase in probe response was found by Hamm et al. (1991). One reason that
a significant finding may have been obtained here but not in the previous
experiment is that every presentation of a slide was probed here, whereas
Hamm et al. probed only 3 of every 4 CS occurrences. Thus, the difference
across studies may be partly due to statistical power.
Visceral reactions. Electrodermal and heart rate responses obtained in
this experiment offer a clearer example of learned temporal and stimulus
specificity. In acquisition, the SIR conductance response (preceding slide
offset) was greater to CS+ than CS-, presumably indexing greater orienting to
shock onset. This differentiation developed linearly over trials and persisted
into extinction. Although overall SIR magnitude decreased in the absence of

reinforcement, this anticipatory discrimination remained constant. Also,
late heart rate deceleration in extinction (D2 leg; occurring at the same time
during slide presentation as the SIR response) was significantly greater
during CS+ than CS-. Late cardiac deceleration presumably reflects
anticipation of events at stimulus offset in an S1-S2 paradigm (Bohlin &

Kjellberg, 1979). Thus, skin conductance and heart rate response provide
convergent evidence for temporally-specific stimulus differentiation.

Exactly what subjects are preparing to do during the training phase of
aversive classical conditioning has been widely debated. The midinterval
acceleratory and late-interval deceleratory components of the cardiac
waveform have been examined for their relationship to psychological
processes. Enhanced cardiac acceleration to an aversive cue has been
proposed to measure recruitment of a defense response (Fowles, 1980; Gray,
1982; Hodes et al., 1985; Obrist, 1981), whereas greater heart rate deceleration
just prior to US delivery has been thought to index an attentional set (Deane,
1969; Lacey & Lacey, 1970). To the extent that cardiac discrimination in








acquisition develops in the acceleratory component, this suggests that
temporary activation of a defense response may occur during learning. In
contrast, cardiac discrimination in the late decelerative leg during extinction
suggests that the information processing set shifts from a defensive
(stimulus rejection) posture to a more attentive (stimulus intake) mode with
cessation of aversive reinforcement.
Patrick and Berthot (1991) recently studied the effect of active vs. passive
preparation on subjects' psychophysiological responses in anticipation of
aversive noise. They found that only late cardiac deceleration varied by
preparatory set, with greater deceleration in the "active" group (who pressed
a microswitch to escape the noise) than either a yoked "passive" group or a
control group anticipating a less intense tone. In the active group, the extent
of cardiac deceleration was significantly correlated with the speed of escape,
which supports the idea that deceleration reflects sensory enhancement (cf.
Lacey, Kagan, Lacey, & Moss, 1963). These data are consistent with the results
of an experiment by Chase, Graham, and Graham (1968), who found that

cardiac acceleration was dominant only when subjects anticipated a vigorous
motor response (i.e., a leg lift), whereas cardiac deceleration was the clear

pattern when subjects anticipated a simple reaction time response.
In contrast to the active-passive group difference they found for cardiac
deceleration, Patrick and Berthot reported that startle magnitude during the
preparatory period was equal for the two groups, increasing closer to the
aversive US (just as in the present study). In contrast, startle responses of the
control group were smaller and decreased progressively closer to the weaker
US which that group received. Thus, the startle probe reflex did not index
the specific tactical response, which supports the hypothesis that it is a more
general strategic indicator of aversive motivation (Lang et al., 1990, 1992).








Brain-Behavior Relationships in Aversive Learning
The present analysis has focused on psychophysiological manifestations
of emotional learning and used these performance measures to infer
cognitive processing in aversive behavior. Yet this study also relates to an
expansive literature on the neural substrates of aversive learning. Thus,
although the present discussion is framed at a representational level of
analysis (Figure 1), the cognitive events described here are presumed to
parallel events at an implementational level (cf. Marr, 1982). Ultimately,
one goal of this work is to develop a psychological model of emotional
learning which is (at the very least) plausible or (at its very best) isomorphic
with neural events.
Davis (1989a) has outlined a neurobehavioral model of the aversive
system using both fear-potentiated startle and sensitization paradigms. The
amygdaloid complex, especially the central nucleus, appears to be the critical
anatomical node involved in aversive learning and memory (Kapp, Pascoe,
& Bixler, 1984; LeDoux, 1987; Sarter & Markowitsch, 1985). Lesions of the
central nucleus abolish both startle sensitization and potentiation (Hitchcock
& Davis, 1986; Hitchcock et al., 1989). Furthermore, drugs which have
anxiety-reducing properties in humans block learned startle potentiation

(e.g., Berg & Davis, 1984), whereas anxiogenic agents increase the learning
effect (e.g., Davis, Redmond, & Baraban, 1979).
Behavioral signs of aversion are presumed to occur when the amygdala
central nucleus is activated and projects to a wide array of brain areas. When
the shock US is presented--as in sensitization--it activates the amygdaloid
complex, which projects to several efferent nodes including the central grey
(leading to inhibition of behavior), the dorsal hypothalamus (cardiovascular
changes), the parabrachial nucleus (respiratory adjustments), and--in the case








of startle reflex excitability--the nucleus reticularis pontis caudalis (Davis,
1989a). Electrical stimulation of the amygdaloid central nucleus elicits these
unconditioned behaviors (Applegate, Kapp, Underwood, & McNall, 1983;
Davis et al., 1982; Gentile, Jarrell, Teich, McCabe, & Schneiderman, 1986;
Harper, Frysinger, Trelease, & Marks, 1984; Rosen & Davis, 1988, 1990).
The amygdala can also act as an associative node for the conjunction of
the CS+ and US. In classical conditioning, the result of pairing a visual CS+
with the shock US is that the amygdala can then be independently activated
by the CS+, via relatively direct subcortical sensory pathways, leading to
potentiated startle (Tischler & Davis, 1983). In short, the flow of events
within this neuroanatomical circuitry is remarkably consistent with the
Konorskian cognitive mechanisms of learning adopted here.
Individual Differences
Aversive Reactivity and Sensitization

Each subject was pre-exposed to a series of shocks, the criterion level
being judged as very intense but below his or her nocioceptive threshold.
Subject temperament characteristics modulated tolerance of shock during
this procedure. Subjects who were relatively more fearful and inhibited
were less able to tolerate increments in shock intensity than other subgroups
(especially nonfearful, impulsive) defined by the fear X impulsivity
temperamental space. These trait fearful, inhibited subjects also reacted
more strongly to the startle probe before learning and reported more specific
fears across a variety of situations than trait nonfearful, impulsive subjects.
Thus, the present study identified subjects who are generally more sensitive
to aversive stimulation.
In the Chattopadhyay et al. (1980; Exp. 1) study noted earlier, auditory
evoked responses to quasi-startle probes (10 msec rise time) were measured








during threat of shock and its absence. Not unlike the larger probe startle
responses observed here for fearful/inhibited subjects, they found that the

P300 component of the auditory evoked cortical response was greater in high
trait anxious than low trait anxious subjects, independent of shock threat.
There may be some communality among those findings and the present
data, since other investigators (Roth, Blowers, Doyle, & Kopell, 1982; Roth,
Dorato, & Kopell, 1984) have found that startle blink amplitude and P300

amplitude covary reasonably well across a range of probe intensities. More
work is needed to determine whether temperamental fear potentiates both
the reflex eyeblink and evoked cortical potential to acoustic startle probes.
This would constitute a test of the hypothesis (Lang et al., 1992) that
emotional response sets are integrated efferent patterns extending from
"cortex to reflex."

It is possible that nonfearful/impulsive subjects, who were less affected
by the aversive stimuli (i.e., shock, startle probe) presented here, may be part
of a "disinhibited" subpopulation that is generally more tolerant of novel or
intense stimulation (Barratt, 1983; Zuckerman, Buchsbaum, & Murphy,
1980). In fact, Bradley (1992) found that the EASI Impulsivity dimension and
Disinhibition subscale of "sensation seeking" (Zuckerman, 1979) both loaded
on a global "behavioral constraint" factor (cf. Tellegen, 1985) in a factor
analysis of personality questionnaire data. Thus, it is worth noting that the
present data agree with a recent report (Pivik, Stelmack, & Bylsma, 1988) that
subjects scoring high on the Zuckerman Disinhibition subscale exhibited
reduced motoneuron excitability (i.e., electrically elicited Hoffman reflexes).
Naturally, one should be cautious in drawing conclusions from such diverse
reflex measures, but these findings suggest that the inhibition/ impulsivity
construct may relate to basic differences in the organization of efferent









output. Future research can determine whether these temperament group
differences are reliable, whether they apply to both aversive and appetitive
USs, and which mechanisms control these putative behavioral differences.
Overall, 77% percent of subjects in this study showed at least some
increase in startle magnitude following shock exposure. However, neither
individual differences in subjects' temperament nor shock tolerance were
significantly associated with the degree of startle sensitization. Since
Greenwald et al. (1990) also found that most subjects (i.e., 95%) showed shock
sensitization of startle magnitude, startle sensitization in humans may be a
fairly robust effect. One reason that slightly more subjects in the earlier study
showed startle sensitization may be that many more preshock startle probes
were administered (as part of a preceding experiment), presumably leading
to greater habituation and a lower prelearning baseline for assessing startle
sensitization. As argued above, this factor would enhance the effect across
subjects.
Associative Learning
Temperament and tolerance of electric shock. Individual differences in

temperament and shock behavioral tolerance did not generally influence
associative learning. Because the fearful/inhibited group attenuated their
exposure to shock initially, it is difficult to determine whether these same
subjects should learn better with an aversive reinforcer (as Eysenckian
theory might predict). More extensive pre-exposure to the US is known to
retard classical conditioning (Randich & LoLordo, 1979). Therefore, it would
be expected that nonfearful/impulsive subjects--whose greater tolerance for
shock resulted in more pre-exposure in this study--might actually show
poorer associative learning. Yet impulsive subjects performed better in skin
conductance (i.e., greater potentiation and discrimination). Thus, neither




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EKAKBN6XF_DQST1U INGEST_TIME 2017-07-12T21:06:30Z PACKAGE AA00003270_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES



PAGE 1

AVERSIVE LEARNING, INDIVIDUAL DIFFERENCES AND PSYCHOPHYSIOLOGICAL RESPONSE By MARK KENNETH GREENW ALO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGR E E OF DOCTOR OF PHILOSOPHY UNIVERSITY O F FLORIDA 19 92

PAGE 2

ACKNOWLEDGEMENTS Although my graduate training has taken me through several distinct phases, several people have been my constant supporters mentors and friends Foremost among these is Peter Lang--whose role as my supervisor and model researcher have made an indelible mark on the way I perceive the broad field of psychology, and how I will conduct my future research. I thank Peter profusely for helping me to do my best. Two equally important and outrageously talented people, Margaret Bradley and Bruce Cuthbert, provided daily inspiration and guidance, and s e r ve d as constant reminders of my own humility For their patience and willingness to listen to my problems, and for the myriad times they came to m y res cue over the ye a rs I am d ee ply indebted to them I h a ve also been v e ry fortun a te to have two fri e nds whose collabora ti o n on va rious projects m a d e m e a ppreciate their growing t a lents To Alfon s Hamm whose early teamwork ev entually resulted in this dissertation, and to Christopher Patrick, with whom I have shared countless stimulating discussions--! am very thankful. Other scientists whose ideas have also ex cited my imagination, and with whom the Lang laboratory has been a gr eat pl ace to work, are Ed Cook, M a rk McManis, Margar e t P e try Ell e n Sp e nc e, Sco tt Vrana and David Yor k. Durin g my trainin g, m y com mitt ee m e mb e r s h ave b ee n m os t h e lp fu l and suppo rti ve o f m y wo r k. Keith Be r g i s a marvelous r o l e mode l for hi s keen met h odo l og i ca l eye and technica l abi li ties; h e a l so ta u g h t me many fundamenta l s of psychop h ysiology. I ra F i sch l e r and Michael Levy were ii

PAGE 3

instrumental in helping me to negotiate the ever-complic a ted terrain of contemporary cognitive psychology Jim Cauraugh, alwa y s pleasant and enthusiastic, coached me in my emerging interest in the cognition of motor control. Finally, the hectic pace of doctoral work was actually enjoyable because three incredibly dear people kept me sane and happy. My loving wife and companion, Margaret, and my devoted parents, David and Rosemarie, have been my cheerleaders and saviors throughout my education The golden opportunities they have given me help put all this work in the proper perspective. iii

PAGE 4

TABLE OF CONTENTS ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . 11 ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . V 1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . 1 The Startle Probe in the Assessment of Emotion . . . . . . . . 1 Aversive I..,earning . . . . . . . . . . . . . . . . . . . . . . 3 Sensitization.. . . . . . . . . . . . . . . . . . . . . . 4 Associative I..,eaming . . . . . . . . . . . . . . . . . . . 5 Individual Differences . . . . . . . . . . . . . . . . . . 10 Statement of the Experimental Problem . . . . . . . . . . . . 13 METHODS .................... .... ................... .. ........ .. 20 Subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Stimulus Materials . . . . . . . . . . . . . . . . . . . . . 20 Apparatus and Psychoph ysiolog ical Recording .... .. ............ 22 Scoring and Definition s of Physiological Respon se Measures . . . 23 Procedure..... ... ... .. ........ ... . ........... .... ...... ... 24 Design and Data Analysis. . . . . . . . . . . . . . . . . . . 31 RESULTS .................. .. .................................. .. 36 S e nsiti za tion ... ..... ........ ... ....... .. ........ ..... .... 36 Startle Probe Respon se . . . . . . . . . . . . . . . . . . 36 Visceral Reactions . . . . . . . . . . . . . . . . . . . . 40 Summary . . . . . . . . . . . . . . . . . . . . . . . . 42 Associative Learning .... ........... ... ... .. .... ..... .. ..... 42 Affective Ratings . . . . . . . . . . . . . . . . . . . . 43 Startle Probe Respon se . . . . . . . . . . . . . . . . . . 47 Visceral Reaction s . . . . . . . . . . . . . . . . . . . . 54 Summary ......... ....... ....... .. ......... . .. ... .. . 61 Individual Differences . .. .. ........... .. .. ........... .. .. . . 61 Temperament Charact e ristics .. .. ...... .. ... ... ........... 61 Rated Shock A versi v e n ess ................... .... .......... 68 Summary ... . .... . .... .. ..... .. ......... .... .... ..... 75 iv

PAGE 5

DISCUSSION ...... .... ........... ... ....... .. ............ .... .. .. 76 Sensitization ........... ......................... .. ........... 76 Startle Magnitude .... .. ......... ... .... ........ ... ....... 76 Methodological Considerations . . . . . . . . . . . . . . 78 Associative Learning ... ........ ... .... ....... .. ............ . 80 Habituation and Sensitization in Associative Learning ....... 81 Affective Associative Change ... .. .. ........ .. .. ......... 85 Temporal Specificity and Preparatory Processes .............. 87 Brain-Behavior Relationships in Associative Learning ....... 90 Individual Differences ........................ .... ............ 91 Aversive Reactivity and Sensitization ..... ...... .... .. ... 91 Associative Learning ... .... ..... ........ . ... .... .. ..... 93 Summary. ....... .. .......................... . . . . . . . . 96 APPENDIX A STIMULUS SELECTION . . . . . . . . . . . . . . 98 APPENDIX B PROCEDURAL INSTRUCTIONS. . . . . . . . . . 103 APPENDIX C QUESTIONNAIRES ....... .. ....... .. ............. 109 APPENDIX D AUXILIARY DAT A TABLES . . . . . . . . . . . . 113 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . 138 BIOGRAPHICAL SKETCH . . . . . . . . . . . . . . . . . . . . 150 V

PAGE 6

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy AVERSIVE LEARNING, INDIVIDUAL DIFFERENCES, AND PSYCHOPHYSIOLOGICAL RESPONSE By Mark Kenneth Greenwald Chairman: Peter J. Lang, Ph.D. Major Department: Psychology August, 1992 In this experiment a psychophysiological analysis of human aversi v e learning was conduct e d There were three primary goals in this stud y. First sensitization of the eyeblink reflex to an acoustic startle probe (95 dBA noise burst) and visceral activation (skin conductance, heart rate) following ele c tr i c shock exposure were examined. Subjects were randomly assigned to two groups, which were exposed to shock either before an adaptation series o f two pictures or midway through this series. To control for unrelated effects e ach group completed questionnaires while the other was exposed to shock. Shock exposure significantly increased startle magnitude relative to control conditions, whereas vi s ceral activity was not influenced. Second, the degr ee to which the startl e refle x, visceral activit y, and affec tiv e r a tin g s ind ex assoc i a tive l e arnin g w as t es t e d u s in g di ffe r ent i a l ave r s i ve co nditi o nin g. The se ri es of t wo p i c tur es co ntinu e d a n d o n e s l ide was pai r ed w ith shock. Associa ti ve l earn in g pote n t i ated s t a rt le r eflex and vi

PAGE 7

electrodermal responses, and generated more cardiac acceleration and rated aversiveness of the shocked slide (relative to the nonshocked slide) This experiment also examined whether pairing shock with different affective slides would yield variable degrees of associative change Stimulus selection was based on each subject's affective ratings, not specific content. Slides varied in pleasantness and arousal across four groups The results, however, did not support the hypothesis that the affective foreground constrains the degree of associative modulation. Third, this study explored whether individual differences modulate measures of aversive learning or reactivity to the prepotent stimuli used here (i.e., shock, startle probe). Subject temperament dimensions, as well as tolerance and rated aversiveness of the shock, were examined. Although temperament characteristics were not generally related to learning, subjects who were more fearful and inhibited exhibited greater reactivity to aversive stimulation (i.e., larger startle responses, less shock tolerance, and more reported fearfulness). In addition, subjects who rated shock as more aversive demonstrated greater learning in startle reflex, electrodermal, and verbal report measures. A cognitive learning model, in which representations of the shock and conditioned stimulus are connected to the brain's aversive motivational system, is used to explain the present results. vii

PAGE 8

INTRODUCTION The startle probe reflex is reliably augmented when ongoing processing involves aversive information. This affect-startle effect, repeatedly found in aversive learning experiments with animal subjects (Davis, 1989a), now forms the centerpiece for a broader theory of human emotion (Lang, Bradley, & Cuthbert, 1990, 1992). In this dissertation experiment, a psychophysio logical analysis of human aversive learning is conducted First, sensitization of the startle reflex and measures of visceral activation (skin conductance, heart rate) during electric shock exposure (i.e., nonassociative learning) are examined Second, the degree to which the startle probe reflex, visceral activity, and affective judgments index associative learning, through pairing photographic slides with shock, is tested Third, the present experiment will e x plore the extent to which individual differences modulate psychophysio logical indicators of these two forms of aversive learning The Startle Probe in the Assessment of Emotion The startle reflex is an obligatory motoric response to a rapid-onset, relatively intense, exteroceptive (e.g acoustic) probe Although the startle response can be a generalized bodily reaction (Hunt & Landis, 1936), the magnitude and latency of the eyeblink to the probe are the characteristics most typically m e asured in ex p e riments using hum a n subjects (Anthony 1 985). Thi s r e fl ex oc c ur s in m a n y s p ec i e s a nd h as b ee n m os t w idely investigated in an i mals (e.g., th e who l e body re spo nse in rats), for w h om the n e urobiological foundat i ons of t he p rim ary sta r t l e circuit are n ow we ll understood (Davis, Gendelman, Tisc hl e r & Gende l m a n 1982) 1

PAGE 9

2 There is an additional and highly valuable use for this reflex behavior: The blink-eliciting probe ma y b e introduced unpredictably during information processing (e g perception, mentation, learning), and changes in the magnitude or latency of the probe reflex in these contexts (relative to control conditions) permit one to infer the disposition of the organism. In fact, the affective value of environmental stimuli has been repeatedly found to modulate the magnitude of the startle probe reflex. Specifically, the magnitude of the startle probe reflex is largest when perceiving aversive pictures, relatively smaller during neutral pictures, and smallest during pleasant pictures (Bradley, Cuthbert, & Lang, 1990a, 1990b, 1991; Greenwald, Bradley, Cuthbert, & Lang, 1990; Vrana, Spence, & Lang, 1988). Similarly, startle probe response is greater when imagining aversive scenes than neutral or pleasant events (Cook, Hawk, Davis, & Stevenson, 1991; Cuthbert Bradley, York, & Lang, 1990 ; Vrana & Lang, 1990). Since this valence-reflex relationship does not depend o n sensory processing it suggests that startle modulation indexes a central motivational state. A recent theoretical mod e l offers a framework for interpreting this affect-startle effect (Lang et al. 1990). In this view, emotions are conceived as tendencies to act, represented in memory as an associative network of context-specific perceptual and response information. A postulate of this model is that all affective representations are fundamentally appetitive or aversive, reflecting action dispositions of approach or avoidance (see also Dickinson & Dearing, 1979; Konorski 1967). Unpleasantness is associated with th e bias to avoid/withdr aw, w hereas pl e a s antn ess is associated with a ppro ac h/ see king beha v i or. Bot h ext rin s i c ( e n v ir o nm e nt a l) and intrin sic ( s ubj ec t) va ri a bles contribut e t o th e fo rm a ti o n o f thi s ce n t r a ll y o r ga ni ze d r espo n se se t Th e e ff ere n t syst e m w hi c h inclu des ve r ba l ove rt m oto r (e .g

PAGE 10

3 reflex), and visceral outputs, is motivationally tuned by this central set to respond in accordance with ongoing processing. Physiological measures can, therefore, be used to index this response organization in real time. Using an affective matchin g principle, Lang et al. (1990) argue that defensive behaviors (e.g., the startle response to an aversive probe) are facilitated when the processing task also occasions an aversive state (in which case a "match" occurs) and inhibited when the situation is pleasant (a "mismatch"). The affect-startle effect thus reflects the biphasic conceptual organization of emotional valence. Startle augmentation during aversive processing thus provides a compelling link to previous dimensional accounts of emotion, all of which emphasize the direction (valence) and intensity (arousal) parameters of behavior (Duffy, 1972; Hebb, 1949; Mehrabian & Russell, 1974; Osgood, Suci, & Tannenbaum, 1957; Schlosberg 1952; Schneirla, 1959; Telleg e n, 1985). Aversive Learning Since this study focuses on learning, a clarification of terms is first presented Here, learning i s defined as a hypothetical process inferred from measurable behavioral changes which result from specific experiences (Mackintosh, 1983). Learning processes have been subdivided as either nonassociative, e.g., within-session habituation and sensitization, or associativ e, e.g., classical and operant conditioning. In this experiment, sensitization and associative learning (via classical conditioning) are the main subjects of interest although habituation is also addresse d as a secondary issue. Both typ es of l ear ning studied h ere are furt h e r characterized as aversive by the type of stimulus which will be used to modify behavior: electric shock. Sensitization will be infer r ed when a r espon e increment occur to shock exposure in the absence of an

PAGE 11

4 association, whereas associative learning will be inferred from response increments which result from explicit pairings of a stimulus with shock. Sensitization It is well established that relatively intense stimulation can produce response increments which reflect sensitization. Studies of the hindlimb flexion reflex in the spinal cat (Thompson & Spencer, 1966) and siphon withdrawal reflex in Aplysia (Hawkins & Kandel, 1984) have shown that repeated tactile stimulation can result in an initial increase in reflex magnitude (sensitization), followed by decreased response (habituation)especially if stimulation is more intense This finding led to the develop ment of "dual-process theory" (Groves & Thompson, 1970; Petrinovich, 1984), which proposes that sensitization and habituation are independent neural processes, the net activation of which determines response output at any moment Dual process th eo ry--which relates neurophysiological mechanisms of habituation and sen sitization to molar theories of behaviorpresumes that habituation occurs in the stimulus-response pathway, whereas sensitization occurs because of tonic changes in the state of the organism, which then modulate the stimulus-response pathway. Startle modulation. Sensitization of the acoustic startle response has been observed in several contexts First, early habituation experiments showed that an intense startle probe can, by itself elicit transient reflex sensitization prior to habituation (Davis, 1972; Groves & Thompson, 1970; Szabo & Kolta, 1967) A second type of sensitizati on has been identified, in which incr eas ing intensity of consta nt background noise e nhances the pro be reflex response (Davis, 1974a 1 9 74b; Hoffman & Searle, 1965; I son & Hammond, 1971; but see Putnam 1975, for a qualification of this effect)

PAGE 12

5 Third, and most recently, Davis and his colleagues have convincingly shown that the startle reflex is sensitized following exposure to electric shock (Boulis & Davis, 1989; Davis, 1989b; Hitchcock, Sananes, & Davis 1989). Davis (1989b) showed that sensitization (1) is attainable with a single shock, but the effect increases with 5 or 10 shocks, (2) develops quickly, and (3) cannot be attributed to rapid association with the experimental context It is proposed here, as Davis suggested, that this effect represents the action of unconditioned properties of shock which can support aversive conditioning. Therefore, sensitization appears to be a building block of conditional learning (Hawkins & Kandel, 1984). Sensitization could also be interpreted as a type of state transition in which response codes activated by stimulus exposure become eligible for associative learning (cf Sutton & Barto, 1981) Konorski (1967) developed a similar cognitive model of learning, proposing that exposure to an unconditioned stimulus (US) leads to an internal sensory repres e ntation of the US, which th en activates a central motivational state related to the affective quality of the US (Figure 1). This motivational state produces preparatory behavior which r eflec ts the affective character of the stimulus (e g withdrawal from aversi ve stimulation). This account suggests why the startle refl ex--a protective reaction--should be sensitized after shock exposure. According to the affective matching principle (Lang et al., 1990), shock exposure activates the aversive system (vi a the US representation) and should prime related defensive behavior s. Associative Learning The mechanism of classical conditioning, Konorski (1967) argued in volves a s impl e extension of th e cognitive events described for sensitiza tion Following repeated presentation and pa iri ng of th e conditioned stimul u s (CS) with th e US, the internal representation of t h e CS

PAGE 13

unconditioned responses, e.g., reflexes ___ preexisting connection learned connection I r 6 ,I / Aversive Motivational System I I I cs aversive preparatory responses, e.g., freezing, sweating, cardiac acceleration, hyperventilation, facial expression, reports of fear Figure 1. Cognitive model of aversive learning (adapted from Dickinson & Dearing, 1979). Presentations of the CS and aversive US lead to internal representations of each stimulus (shaded circles). The US activates (via pre-existing connections) the aversive motivational system which, in turn, generates associated behaviors Reinforcing the CS with the US generates learned connections with the US representation and aversive motivational system. Thus, direct US exposure (as in sensitization), and CS-US pairings (as in associative learning) can both potentiate aversive behavior.

PAGE 14

7 develops independent e x citatory connections with both the US representation and the relevant motivational system (Figure 1) Due to i ts ass o ciation with the motivational system, the CS r e presentation comes to activate preparatory behavior related to the affecti v e quality of the US. Konorski's theory not only deals with connections and representation but unlike many extant cognitive models of associative learning (e.g Estes, 1973; Frey & Sears, 1978; Kamin, 1969; Kehoe, 1988; Pearce & Hall, 1980; Rescorla & Wagner, 1972; Wagner, 1978), also explicitly models excitator y and inhibitory interactions between hedonic stat e s In this framework (see also Dickinson & Dearing, 1979; Wagner & Brandon, 1989), aversive and appetitive motivational systems--which comprise t he directive forces of adaptive behavior--are assumed to interact via reciprocal inhibitory connections Since conditioning procedures modulate affective stimulu s as soc i a tion s, they can b e u se d to transform the affec ti ve meaning of a CS ( e.g b y p a iring an app eti ti ve st imulus with sh ock) an d thus demon s t rate th e o p e ration of these opp o n e nt s ystems. Trad itio n a ll y, t hese aversiv e ap pet itiv e interactions h ave b ee n measured in a n i m a l s using conditi o n e d su p pr es sion (e.g., Dickinson 1 9 77 ; Rescorla, 1971 ). Ho w e v er, use of startl e r e fl ex modulation afford s a numb e r of advant ages in ev aluating affec tive assoc i a ti ve learning in cl u d in g it s independen ce from o perant behav io r S t a rtl e modulation Ex p e riments using a n ima l s u bje cts have s h own th at th e s t a rtl e refl ex c a n be a u g m e nted by av e r s i ve cl ass ical conditionin g, i e th e "fea r pot e nti a t e d s t a rtl e" effe ct Durin g tra inin g, a previous ly n eutra l l ig h t C S i s p a i red w i th an ave r s iv e e l ectric shock. In s ub seq u ent test tri als (in which shock ceases), an acous ti c prob e i s presented du r in g pr ese ntation of the light cue and in the absenc e of th e light. In these studies, fear-po ten tiat e d startle i s defined as the within -subj ct differe n ce in startle

PAGE 15

8 amplitude during the cue versus its absence (Davis, 1986). Numerous investigators have demonstrated this effect (Berg & Davis, 1984, 1985; Brown, Kalish, & Farber, 1951; Cassella & Davis, 1986; Davis & Astrachan, 1978; Kurtz & Siegel, 1966; Leaton & Borszcz, 1985). However, few experiments have been conducted using human subjects. The first studies which examined potentiated startle in humans were originally completed about 30 years ago, and were primarily concerned with the temporal delay of CS/probe presentation on reflex potentiation. One conclusion from this early research (Ross, 1961; Spence & Runquist, 1958), as well as some recent related work (Vrana et al 1988), is that probe blink response to an aversive cue is less reliably potentiated when the probe follows stimulus onset by relatively short intervals (e.g., 500 msec). The selective absence of reflex potentiation at short time lags suggests that prepulse inhibition--an attentional mechanism--may obscure measurement of fear induction (Davis, Schl esinge r, & Sorenson 1989). Recently, Hamm, Green wa ld Bradley, and Lan g (1991) reintroduced the fear-potentiated startle paradigm with human subj ects, suggesting that the hiatus in experimentation (despite Konorski's enlightened views) may ha ve been due to a disinterest in conditioning during the "cognitive revolution ." However, the basis for understanding the relationship between learned fear and startle potentiation was also complicated by Hull s (1943) explanatory u se of the motivational construct drive, which was presumed to have a broad energizing effect on behavior (and is therefore analogous to arousal rather than va l e nc e). According to this position, pairin g a light with aversive shock increased organismic drive, which then l ed to the increase in s t artle ma gnitude. However, early anima l studies failed to show a u g m e nted startle effects when appetitive reinforcement was us ed: Probe startle responses

PAGE 16

9 during a stimulus paired with food were unchanged (Trapold, 1962) or even diminished (Armus, Carlson, Guinan, & Crowell, 1964; Armus & Sniadowski-Dolinsky, 1966), relative to control conditions. Therefore, these results argue against a drive theory interpretation of startle potentiation. Hamm et al. (1991) thus adopted the Konorskian model to study emotional learning. Unlike the single cue paradigm used with animals, Hamm et al. used a differential learning paradigm, in which each subject viewed two alternating slides during the experiment. For training, one slide was always followed by shock, whereas the second slide was never followed by shock. This enables within-subject control over the associative process. Before and after training, the startle probe was presented during the slide associated with shock (CS+), the slide never followed by shock (CS-), and during interslide intervals. Visceral (heart rate, skin conductance) responses and affective (valence, arousal) ratings of the stimuli were also measured. Results indicated that blink reflexes were augmented wh e n the startle probe was administered during the slide that was paired with s h o ck. Reflexes evoked in the presence of this cue were significantly larg e r than those elicited during the slide never paired with shock, or during the interval between slide presentations. These data replicated and e x tended findings from the animal literature A second objective of the Hamm et al. (1991) experiment was to explor e whether a priori stimulus affective properties influenced the degree of emotional learning. Thus different groups of s ubj e ct s we r e assigned to vi ew s lid e co nt e nt s which v a ri e d in affec ti v e valen ce a n d a r ousa l t hroughout l ea rn ing T wo a lt e rn ative hypot h ese s we r e eval u a t ed One h y p ot h esis, b ase d o n b e l o n g in g n ess th eo r y (Ga r c i a & Ko e lli ng, 1 966) pred i c t ed t h a t bett r condition i ng s h o uld be ob t ained fo r sl i des w i th aff e cti ve values wh i ch wer e

PAGE 17

10 initially similar to shock The second hypothesis, based on the "delta rule" (Rumelhart, Hinton, & McClelland, 1986), predicted that better conditioning should be obtained for slides with affective values which were initially dissimilar to shock. Results indicated that the degree of startle potentiation varied linearly with the a priori affective valence of the slide contents: Slides of unpleasant content (i.e., mutilated bodies, threatening animals) yielded significantly less startle potentiation than slides depicting pleasant content (i e nature scenes, attractive nudes). Conditioned startle potentiation thus reflected changes in hedonic value (i.e., pleasure). In contrast, the degree of skin conductance discrimination in extinction varied with the a priori level of arousal of the conditioned stimulus: Slides depicting arousing content (i.e., attractive nudes, mutilated bodies) yielded significantly less conductance discrimination than slides of calm content (i.e., household objects). These data suggest that electrodermal differentiation reflects arousal (rather than hedonic) change. Ther efore, the results supported the delta rule, rather than the belongingness hypothesis. Individual Differen ces In the myriad contemporary conditionin g studies which have exposed human subjects to electric shock USs (e.g., Cook, Hodes, & Lang, 1986; Hammond, Baer, & Fuhrer, 1980; Hugdahl & Ohman, 1977; Lanzetta & Orr, 1986; McNally & Reiss, 1984), the subject is asked--for obvious ethical reasons--to set the shock level to be moderately to highly annoying (uncomfortable), but not painful. In other words, the final shock level is corre l ated with the subject's pain toleran ce. However, none of these studies has addressed whether a subjec t's tolerance for s hock relates to personality characteristics, nor whether temperament or shock tolerance diff e r ences can

PAGE 18

11 predict the degree of learning. Even so, there is an older set of studies and theory which bears on this issue. Extensive, but inconclusive, research has been conducted on the relationship between personality and conditioning (see Levey & Martin, 1981, for a review). As noted above, much of early conditioning research was influenced by Hull's concept of drive; this approach was wedded with the study of individual differences in anxiety by Spence's group at Iowa in the 1950s and early 1960s. This work was not related to a theory of personality but, rather, tested the influence of a single trait on learning. Spence (1958, 1964) demonstrated that anxiety level (which was equated with drive) positively related to the rapidity of conditioning, especially when the subject reported acute anxiety. Although many learning (particularly eyelid conditioning) experiments were conducted within this framework, aversive shock reinforcement was not employed. In contrast to the work above, Eysenck (1967) developed a two-factor theory of personality emphasizing extraversion-introversion and neuroticism as central dimensions. This scheme evolved from Pavlov's (1957) observational typology of nervous system "strength" into modern psychophysiological variants. The basic idea is that introverts possess weaker (i.e., less strong) nervous systems than extraverts, rendering them more susceptible to intense, especially av e rsiv e stimulation. Gray (1967) and Nebylitsyn (1972) generated research testing the hypothesis that introverts should form condition e d responses m o re r ea dily than extravert s. Gray noted that the int e raction of neuroticism--which involves s e n s itivity to both r ew ardin g a nd puni s hin g sti mul a ti o n-a n d ex tr ave r s ion p e r so n ality fac tor s im pli es th at neurotic introverts s h ou l d r espond m o r e th a n a ll othe r s t o

PAGE 19

12 aversive stimulation. This quadrant of the two-factor space has been associated with high trait anxiety (Gray, 1973, 1982). Unfortunately, there is some lack of clarity in two-factor theory and research. First, it is not clear whether anxious (neurotic introverted) subjects should be more reactive to aversive stimulation (e.g., shocks, loud noises), whether they should exhibit better associative learning with aversive reinforcers (e.g better discrimination between a stimulus paired with shock and one not paired with shock), or both. Second, Eysenck's early research showed that there was a positive correlation between introversion and neuroticism, which suggests (cf. Gray, 1982) that the Eysenckian structure is impure. Third, impulsivity (or, conversely, inhibition)--considered by these researchers to be a secondary factor, but given primary importance by others (e.g., Kagan, Reznick, & Snidman, 1987)--loads positively on the extraversion dimension and, thus, may account for variance in conditioned responding (Barratt, 1971; Eysenck & Levey, 1972). To minimize these problems, the present experiment uses a related alternative framework to explore individual differences in aversive reactivity and learning. The EASI (Emotionality [fear, anger], Activity, Sociability and lmpulsivity) temperament theory of Buss and Plomin (1975) was chosen because of its basis in heritability studies, and because the factor structure of the EASI assessment instrument relates to two-factor theory. For example, Eysenck's extraversion could be roughly translated into relatively high sociability and activity in the Buss and Plomin scheme. Thus, if the sociability and activity dimensions were to control variance in aversive reactivity or l earning, this would argue aga inst Eysenckian theory. In fact, the most l ogical expec tation is that fearful and/or inhibited temperamentanalogous to neurotic introversion--would facilitate aversive responding.

PAGE 20

13 Conditioning researchers have also not addressed whether perceived aversiveness of the shock US, regardless of personality factors or shock tolerance, relates to the amount of affective associative learning. Within the cognitive model of learning proposed above, this perceptual variable would appear to be rather important. To reiterate, the US representation is presumed to develop an excitatory connection with the CS representation after repeated pairings, which activates a central motivational state and produces preparatory behavior which reflects the affective character of the stimulus. Since each representation is assumed to be influenced by perceptual factors, then a more aversive US should enhance learning. Statement of the Experimental Problem The flow of experimental events generally follows the procedure of Hamm et al. (1991) described earlier First, prior to learning, all subjects viewed and rated a series of colored photographic slides which varied widely in affective valence and arousal. Based on these ratings, picture stimuli for conditioning were selected. Second, subjects were exposed to an aversive electric shock to evaluate its effect on response sensitization (Figure 2). To control for passage of time dishabituation (from the experimenter leaving/re-entering the room), and other factors not related to sensitization there were two groups that served as each other's control. For one group shock exposure occurred prior to an 8 trial adaptation series of two slides, whereas the second group was exposed to shock mid w a y through the adaptation se ri es. During each group's exposur e to shock th e ot h e r group compl e t e d q u es tionnair e s to s e r ve as a control fo r s hock sensitization. Fo r both g r oups, startle p r obes we r e p r esente d b efo r e s ho c k expos u re, a nd durin g a n d between slide p r ese nt a ti ons t hr o u g h o ut th e a d aptation period.

PAGE 21

Preleaming Group 1: Immediate Shock Exposure V rest period var ISI I slide 1 Adaptation Block 1 slide 2 slide 3 t t t t t t t Group 2: Questionnaires (Co ntrol Period) V rest period var. ISI I slide 1 slide 2 slide 3 Questionnaires (Control Period) ,----------, V slide 4 var. ISI I slide 5 t t t Delayed Shock Exposure ,----------,V slide 4 var. ISI I slides Adaptation Block 2 slide 6 slide 7 slide 8 t t t t slide 6 slide 7 slide 8 t t t t t t t t t t t t t t Figure 2. Group differences in procedure for shock exposure. Following the startle probes (arrows) administered before learning, the Immediate exposure group (above) was exposed to shock, then viewed a series of 4 slides (2 CS+, 2 CS-; randomly arranged) Probes were pre se nted once during each slide and during two random interslide intervals in each adaptation trial block. The questionnaire control period ensued, after which a second block of the same 4 slides (again, 2 CS+, 2 CS-) was administered. For the Delayed exposure group (below), the entire sequence was reversed so that the questionnaire period preceded s hock exposure ...... _p...

PAGE 22

Acquisition us us us us I var. ISi CS+ csCSCS+ I CSCS+ CS+ CS....... ..... Extinc tion var. ISi CS+ t t CS+ CSt t t cst CSCS+ CS. . . . t t t l lr:7 t ~t I. ..... Figu re 3 Associative learning procedure. Each subject viewed a series of 2 slides (CS+, CS-) in a random stream se parated by variable interslide intervals (see Figure 2 for the adaptation slide series, which preceded acquisition) During acquisition only each of 8 CS+ presentations was followed by the shock at slide offset, whereas the 8 CS presenta tions were not. Probes (arrows) were presented during adaptation and extinction during each slide (at 3.0 3.75, 4.5 or 5.25 sec) and during interslide intervals. Unlike the adaptation series extinction (8 presentations each for CS+ and CS-) followed acquisition without interruption and was uniform for all subjects. ...... \J1

PAGE 23

16 Third, the subject continued to view this unpredictable series of two slides, which now served as the CS+ and CSin a discriminative learning procedure (Figure 3). In acquisition, one of the two habituated slides (CS+) was always reinforced with shock at slide offset, whereas the other (CS-) was not. Each slide was presented 8 times, but startle probes were not presented during associative learning This acquisition sequence was followed by a 16 trial extinction series, during which probes were again presented during and between slides (8 presentations each slide). Finally, after the learning procedure, a sequence of 12 slides varying in affective valence and arousal (including each subject's CS+ and CSpictures) was presented. Subjects rated the slides as well as the perceived aversiveness of the shock US, CS+ slide, CSslide, and the startle probe. There were three primary goals of this experiment. The first goal, based upon demonstrations in the animal literature (Boulis & Davis, 1989; Davis, 1989; Hitchcock et al., 1989), was to assess the sensitizing effects of simple, noncontingent, shock ex posure on the startle probe reflex, as well as activation in the cardiac and electrodermal systems Using human subjects, Greenwald et al (1990) recently demonstrated that exposure to electric shock significantly increased startle reflex magnitude, but that skin conductance was not similarly augmented. However, their experiment was not e x plicitly designed to test this hypothesis, e.g., the design did not control for response increases which might have resulted from the passage of time. In the present study, subjects were randomly assigned to two groups varying only in the timin g of shock expo s ure, s o that each group s e r ve d as th e other's contr o l in t es tin g sen s itizati o n l ea rnin g. It was e x p ec t e d that s ho c k e x po s ur e wou l d i n c r e a se th e s t a rtl e p r obe r es pon se in hu mans, m u c h as it doe s in a nim a l s I n co ntrast s tud ies w i t h a n i m a l s ubj ec t s h ave not

PAGE 24

17 recorded autonomic activity before and after shock exposure; thus, although electrodermal sensitization was not obtained by Greenwald et al. (1990), only a weak hypothesis of no effect" can be advanced. From the standpoint of dimensional emotion theory (Lang et al., 1990), startle reflex sensitization should reflect change s primarily in the hedonic (valence) state of the individual, whereas sensitization of visceral response should indicate a change in arousal level. The second goal of this study, based upon early human aversive conditioning experiments (Ross, 1961; Spence & Runquist, 1959) and recent follow-up studies (Greenwald et al., 1990; Hamm et al., 1991), was to conduct a psychophysiological analysis of shock associative learning using startle reflex, visceral response, and affective judgment measures The objective here was to examine the sensitivity of each of these measures to simp le association (i.e., measured by potentiation of r es ponse to the slide paired with shock [CS+]), and to associative dis crimination (i.e., measured by differences in respons e between CS+, the slide not paired with shock [CS-], and in the absence of slides). The data gener ated by Hamm et al. (1991) suggest that both startle potentiation and discrimination should occur. The results of previous autonomic learning exp e riments (e.g ., Hodes, Cook, & Lang, 1985) suggest that cardiac response discrimination should primarily occur in acquisition a nd that skin conductan ce differentiation s hould be present during both acq uisition and extinction of the association It is not clear, however, whether affective judgment s s hould better refl ect the s imple association (i.e ., CS+ increases in displeasure and aro usal) or the discriminative aspects (i e., CS+/CSdifferenc e) of the learning situa tion In addition, this experime nt examined whet her pairing aversive shock with different emotional slides would yi ld variab l e de grees of aff ctive

PAGE 25

18 associative change. Whereas Hamm et al. (1991) assigned subjects to view slide pairs which differed in affective valence and arousal, slide cont e nt was confounded with this manipulation. In this study, there were four groups who viewed either unpleasant-calm, unpleasant-arousing, pleasant-calm, or pleasant-arousing slide foregrounds during conditioning. (Unlike Hamm et al., 1991, neutral contents were not used as foregrounds in this experiment .) Each subject viewed two pictures which he or she rated as most extreme (regardless of content) in terms of his or her stimulus assignment to one quadrant of the two-dimensional (valence X arousal) affective space The present approach is a deliberate, extreme-case test of the hypothesis that slide affective evaluation should influence classical conditioning. Although pairing any stimulus with shock should generally increase its unpleasantness (and arousal), when the CS itself is aversive initiall y, the amount of affective change due to association with shock may be minimal. Conversely, a stimulus which is initially pleasant (or calm) might accrue more displeasure (or arousal). If startle blink potentiation indexes the amount of learned aversion, then greater changes in rated displeasur e and the startle reflex should occur for pleasant foregrounds, whereas smaller changes are predicted for slides initially high in unpleasantness. Similarly, if skin conductance response potentiation is an index of learned arousal then larger changes in rated arousal and electrodermal reactions should occur for calm slide foregrounds, while smaller changes are predicted for slides initially high in arousal. Such results, if obtained, would confirm the findings of Hamm et al. (1991). The third goa l, clearly exploratory at present, was to identify individual differences which mediate e motion a l l ea rnin g S in ce it is possible that a subjec t' s s h ock t ole r ance l evel is determined by personality characteristics,

PAGE 26

1 9 and that temperament or shock sensitivity may relate to aversive learning, these variables were examined. The EASI Temperament Survey (Buss & Plomin, 1975) was used here to assess differences in aversive reactivity and learning The EASI subscales measure temperament dimensions of fearfulness and impulsivity, as well as anger, activity, and sociability If research relating to Eysenck's personality theory is correct in suggesting that anxious and/ or inhibited subjects should either respond more to aversive stimulation or show better conditioning, then subjects who score relatively higher on the EASI fear subscale and lower on the impulsivity subscale should demonstrate these effects maximally. Finally, as the cognitive model of conditioning adopted here implies, perceptual factors may relate to the degree of emotional learning Therefore, it was predicted that psychophysiological measures of learning would be enhanced in subjects who report greater aversiveness of the shock US

PAGE 27

METHOD Subjects Sixty-four introductory psychology subjects (32 males, 32 females) volunteered to participate in this experiment and received experimental credits toward fulfilling their course requirements. Subjects were informed before arriving at the laboratory that the study involved "cognitive and physiological responses to stimuli in different sensory modalities ." Total participation time for each subject was two hours. Stimulus Materials Affective pictur es Thirty-nine 35 mm colored photographic slides, selected from catalogue of the Internation a l Aff e ctive Picture System (IAPS; Lang, Ohman, & Va it l, 1988), were u se d in this ex p e riment. These slides depict a wide vari e t y o f content inclu d in g w ild a nd domestic animals natural scenes, obj e ct s, human faces and b o die s, social interactions, and conditions of life. The IAPS slides have relatively simple figure-ground relationships which facilitate perceptual resolution and evoke a broad range of affective respons e s within the limits of this stimulus format. The composition of the initial slide set was based on an inspection of normative rating s for all 240 IAPS slid es ( L a n g, Greenwald, & Bradle y, 1988) Appendix A li s ts all s lide materials us e d in thi s s tudy and their norm a tive affective valu es on eac h dimension Th e s lid es we r e distribut e d a c r oss the fo ur quadr a nt s i n the coo rd i n a t e s p ace defi n ed by t h e in te r sect i on of affect i ve va l e nce and a r ousa l axes. E i g ht s lid es from each of th ese fo u r quad r ants ( unpl easa n t-arousing, unp l easant-ca l m, pleasant-arous i ng, pleasant-ca l m) 2 0

PAGE 28

21 and four neutral-calm slides comprised the slide set Additionally, three practice slides (unpleasant, neutral, and pleasant) preceded presentation of the above 36 slides during the pre-experim e ntal viewing/judgment series, to anchor subsequent ratings of these material s. Acoustic startle probe. The startle-inducing stimulus was a 50 msec burst of 95 dBA white noise (20 20,000 Hz), with instantaneous rise time (< .05 msec). The white noise was generated by a Coulbourn S81-02 Noise Generator, gated through a Coulbourn S82-24 amplifier, and delivered binaurally via matched Telephonies TDH-49P earphones. To calibrate noise intensity, the earphone was clamped to a Bruel and Kjaer Type 4153 artificial ear (1/2" Type 4134 condenser), thus directing the sound wave perpendicular to the diaphragm (Putnam, Graham, & Sigafus, 1975). Sound pressure level of the continuous noise was calibrated usin g the fast-filtered A scale of a Bruel and Kjaer Type 2203 sound level met e r. Aversive un co nditioned stimulus ( US ). Th e electric shock US was a continuous 500 ms e c train of rectangul a r pul ses g enerated by a Farrall Instruments Mark 300C Aversive Condition e r The tactile stimulus was delivered through a concentric Tursky ele c t ro d e. The electrode was attached to the volar surfac e of the left upper arm, whi c h was first rubbed lightly with U1traphonic conductive gel. Intensity of th e US was set at a level which the subject described as "highly annoying, but n o t painful" (see Procedur e ). Shock intensity was calibrated using a Micront a DC analog meter (85 ohm internal resistan ce ), a bridge circuit (to r ec tif y th e shock pulse), a resi s tance l oa d of 27 Kohm s, a nd a 100 F cap ac i to r across t h e output t e rmin a l s of th e Fa rr a ll s ho c k er. For each s t e p on t he shock c urrent k n o b th e val u e on t h e Mic r o nt a m e t er was mu l tip li ed b y the duty cycle (32.05) t o ob t a in peak ampe r age l evel s

PAGE 29

22 Apparatus and Psychophysiological Recording Stimulus presentation and data acquisition were coordinated by distributing the processing tasks (via digital input/ output connections) between PDP 11/23 (Data Translation 12 bit analog-to-digital [A-D] converter board) and Apple Ile microcomputers. The Apple Ile triggered a Gerbrands electronic shutter (for slide onset/offset; rise/fall time< 5 msec) and the slide carousel advance/ reverse switches of a Kodak Ektagraphic Ill AM projector. The projector lens was located behind a clear plexiglass sound baffle in the experimenter's room; thus, the subject had no audible cues as to which slide (CS+, CS-) would be presented next during conditioning. The PDP computer recorded all physiological data and controlled the timing of slide, startle probe, and shock presentations. Physiological signals (except heart rate) were sampled at 10 Hz, except during startle probes, during which the sampling rate was temporarily accelerated to 1000 Hz from -50 to +250 msec relative to probe onset. The eyeblink component of the startle response was measured by recording the electromyographic activity over the left orbicularis oculi site, using Sensor Medics miniature silver/ silver chloride electrodes filled with Teca electrolyte gel. The raw signal was bandpass-filtered (from 90 to 250 Hz) and amplified (20,000 gain factor for all subjects) by a Coulbourn S75-01 Bioamplifier, and integrated on-line using a Coulbourn S76-01 Contour Following Integrator (at a time constant, calibrated off-line, of 120 msec). The analog heart rate signal was measured using collarbone and arm leads with Sensor Medics standard size silver/ silver chloride electrodes, using Teca e l ectro l yte gel. Following input to a Coulbourn S75-05 Isolated Bioamplifier, the signal was filtered from 8 to 40 Hz. A Schmitt trigger detected R-wave peaks and int errupted the corn.puter to record interbeat

PAGE 30

23 intervals to the nearest msec from 3 sec prior to slide onset to 2 sec after slide offset. Skin conductance activity was recorded with Sensor Medics standard size silver/ silver chloride electrodes attached adjacently to the hypothenar eminence of the left palmar surface The skin was first cleaned with distilled water and sensors were filled with K-Y jelly electrolyte. A Coulbourn S71-22 skin conductance coupler provided a constant 0.5 volt across electrodes. The signal was calibrated prior to each session to detect activity in the range from O to 40 Siemens (S). The calibration value was used to convert the digitized raw signal off-line to analog skin conductance values. Scoring and Definitions of Physiological Response Measures Startle reflexes were scored using a computer program that scored the blink waveform for onset latency and amplitude. Eyeblinks were scored by the algorithm if (1) minimum response amplitude was 20 A-O units and if the response had a (2) minimum on s et latency of 20 msec and (3) maximum peak latency of 150 msec relative to probe onset. Null response trials (due to violations of criteria 1-3) were as s igned an amplitud e of zero A-O units. Startle reflex magnitude (including zero amplitude responses) was the principal measure of interest. For startle onset latency, null response trials were excluded from analyses. Heart rate was converted from digitized interb e at intervals to analog rate data in half-second bins, with intervals weighted in proportion to the time occupied (Graham, 1980) Initial and late decel e r a tory (01 and 02) and midint e rv a l acceleratory (Al) co mpon e nt s o f the c a rdi ac waveform were sc or e d (Gatc h e l & L a n g, 1 973), since th ese peaks a nd t r o u g h s in the w a v efo r m r e fl ec t psyc h o l ogically s i g ni f i cant eve n ts (Boh l i n & Kj e llb e r g,

PAGE 31

24 1979) These component scores were expressed in beat/minute change scores, deviated from a 1 sec prestimulus baseline. Skin conductance responses were scored using an adaptation of the Strayer and Williams (1981) SCORIT algorithm. The First Interval Response (FIR)--which presumably reflects orienting to stimulus onset--was defined as the largest response during the window from 0.9 to 4.0 sec after stimulus onset (Prokasy & Raskin, 1973). The Second Interval Response (SIR)--which reflects anticipation of the shock US during acquisition and extinction--was scored as the largest response on each trial between 4 1 and 6.8 sec after slide onset (i.e., ending 0.8 sec after slide offset). Third Interval Responses (TIRs)which index the UCR during acquisition and orienting to shock omission during extinction--were scored as the largest responses between 0.9 and 4.0 sec after slide offset Skin conductance response data were later transformed (log [SCR+ 1]) to normalize the distribution of responses Procedure Equipment preparation and instructions. After arriving at the laboratory, each subject completed a consent form to participate in a brief (ca 30 min) slide viewing/rating experiment. The subject was informed about the different emotional pictures, a "brief noise click" and recording of physiological responses. The subject was also told that that there would be a second optional experiment, which the experimenter would describe later No mention of shock was made at this time The subject was seated in a comfortable recliner in a sound-attenuated, dimly lit room adjacent to the equipm e nt room. Physiological se nsors w e r e attach e d a t thi s time, but ( unkn own to s ubj ec t s ) phy s iol og i ca l r es p o n ses t o s l ides were n o t ac tuall y r eco rd e d du r ing th e fir s t ex p e r ime n t.

PAGE 32

25 Subjects were then instructed in using the Self-Assessment Manikin (SAM; Lang, 1980). Subjects used the pencil-and-paper version of SAM (see Appendix B, Figure 19) to make affective valence and arousal judgments of experimental stimuli. To represent changes on the valence dimension, the SAM figure ranges from smiling and happy to frowning and unhappy. To represent changes in arousal, SAM ranges from active and wide-eyed to a static, eyes-closed figure. There are 5 cartoon-like figures for each dimension. Appendix B also provides the experimental instructions for using SAM. Briefly, subjects were instructed to place an "X" on the figure, or between figures, which best represented their affective reaction to the slide on each dimension. This scheme formed a 1 9 rating scale for each affective continuum. "Low" (1) on the valence dimension was unpleasant, whereas "high" (9) was pleasant. "Low" (1) on the arousal scale was calm, and "high" (9) was excited. During initial practice with SAM, the experimenter instructed subjects with verbal labels for the endpoints and middle of each scale. These labels correspond to the anchor adjectives of the semantic differential (Mehrabian & Russell, 1974). Prior to leaving the room, the experimenter instructed the subject that the noise would occur "at various times during the session" and then placed on the headphones. Pre-experimental slide series and selection of conditioning slides. Subjects paced themselves through a series of 39 slides, the first three of which (unknown to subjects) were practice slides and did not contribute to their ratings distribution The subject was instructed to press a button which initi a t e d s lid e onset, watch th e slide closely while it w as on the screen, and r ate eac h slide for valence and arousa l immediately afte r slide offset. After the subject initi a t ed slide expo s ur e it remained o n for 2 sec The s ub ject was

PAGE 33

26 only told that the slide exposure would be "relatively brief," so that after pressing the button (1 sec delay), he or she should be ready to view the slide. After each slide trial, the subject repeated this sequence. Noise probes were not delivered during this slide sequence. The subject was instructed to relax after the final slide trial (which was evident from the numbered rating form). At that time, the experimenter determined from a video monitor that the subject was finished writing and resting comfortably. When this was confirmed, two acoustic startle probes were presented (16 sec apart) to assess preleaming probe response. The subject's rating form was then collected, and the experimenter examined the two-dimensional (valence, arousal) distribution of affective judgments. Before arriving at the laboratory, the subject had been randomly assigned to view one of four slide types (unpleasant/ arousing, unpleasant/ calm, pleasant/ arousing, pleasant/ calm) during learning. Accordingly, the experimenter first selected the two conditioning slides which the subject had rated most extremely in his or her assigned category The experimenter then selected ten adjunct slid es --two in each of five affective categories (same four types as above, plus neutral/ calm)--which were not used in learning but were rated again after the experiment (see below) Appendix A explains the decision rules for choosing these conditioning and adjunct slides. Shock exposure The subject was instructed that a new slide sequence would begin, and to "watch each slide for the duration of exposure and ignore the noise." While subjects were not informed of the actual slide exposure duration (6 sec) or when the noise would occur, probes were presented equally often at each of 4 times during slide exposure --eith er 3.00, 3.75 4.50 or 5.25 sec--as well as between slides (i e., interslide intervals)

PAGE 34

27 Subjects were instructed to relax quietly during interslide intervals, which ranged from 10 to 20 sec in 5 sec increments. Next, the procedure diverged for two groups (see Figure 2). Forty-eight subjects were told, "Experiment 1 is now over." These subjects--designated as the Immediate shock exposure group--were thanked and invited to continue in the next experiment, which provided additional credits toward their class research requirements. This group was told that there would be a new consent procedure for this second experiment and that there was no obligation to participate. If the subject agreed to continue, he or she was given the next informed consent to read. Three subjects (in addition to the sample of 48) declined to continue; they were thanked for their participation, given credit, and debriefed. The second consent procedure informed the subject about electric shock. Otherwise, the description of slide viewing and noise probes was the same as before. After written consent was obtained, the experimenter attached the shock electrode and initiated shock exposure (Appendix B). In this standard procedure, the intensity of shock was adjusted for each subject to a level which was reported as "highly annoying but not painful." The shock level began just above perceptual threshold and was increased gradually in fixed steps to the subjective criterion. Increments in shock level were calibrated to approximate a power function (Appendix B, Figure 20) All subjects reached at least the second shock intensity. Once the shock level was determined, the subject was told that, at various times durin g th e remainder of th e ex p e riment, the shock would occ ur at th e int ensi t y and dur a ti o n ag r eed upo n No m e ntion w as m ade of "c onditi o n i n g," associatio n n o r timi ng of stim ulu s p r ese nt a ti o n T h e

PAGE 35

28 experimenter answered any questions from the subject, informed the subject that the next phase would begin, and exited the room. The adaptation sequence of to-be-conditioned slides began. Each slide (CS+, CS-) was presented twice and probed twice; two startle probes were presented randomly between slides. This was the first block of adaptation. Then the experimenter re-entered the room to break the procedure for an unanticipated questionnaire session. After questionnaires, the experimenter left the room for the last time. The second block of adaptation (identical to the first block) began, and acquisition and extinction slide series continued uninterrupted from this point. The remaining 16 subjects--designated as the Delayed exposure groupcompleted questionnaires first instead of being exposed to shock. Following questionnaires, they were told that a new slide sequence would be presented and were given the instructions for slide viewing above. Adaptation block 1 slide and probe presentations were th en administered in the same manner as for the Immediate exposure group. The experimenter then re-entered the room to inform subjects in this group that experiment 1 was over. The experimenter then obtained the second informed consent, conducted the shock exposure protocol, and left the room a final time. After this point, the second adaptation block, acquisition, and extinction slide series proceeded without interruption--identical to that for the Immediate exposure group. To summarize, when the Immediate group was first exposed to shock, the Delayed group completed questionnaires. This process was reversed, so that when the Delayed group was exposed to shock, the Immediate group completed qu es tionnaires. (Both shock expos ure and questionnaires lasted about 10 mins .) This interlude enabled eac h group to serve as the other's contro l group at their time of shock exposure, as well as an opportuni t y to

PAGE 36

2 9 collect individual difference data. All subjects completed an abridged version of the Fear Survey Schedule (FSS; Wolpe & Lang, 1964), consisting of 40 items describing different types of fears depicted by IAPS slide contents. In addition, the 25 item EASI Temperament Survey (Buss & Plomin, 1975) was administered The EASI subscales tap constructs of emotionality (fear and anger), activity, sociability, and impulsivity. Items on both questionnaires were rated on a 1 5 scale, yielding a 40 200 range for this FSS version and a 5 25 range for each EASI subscale. These questionnaires (see Appendix C) were used to address the relationship of temperament dimensions and situational fears to aversive reactivity and learning. Shock conditioning. The remainder of the experiment involved the discriminative learning procedure. Subjects were randomly assigned to view an unpredictable sequence of either two unpleasant/ calm, unpleasant/ arousing, pleasant/ calm, or pleasant/ arousing slides. These two slides served as CS+ and CS(see Figure 3) During acquisition, the CS+ slide was always reinforced with a 500 msec shock US which coincided with slide offset. The CSslide was never reinforced. Each slide was presented 8 times (16 total trials; 8 different CS+/CSorders, counterbalanced across subjects)-1 1 For all subjects, the last trial of adaptation was a CS+ trial, followed by the shock US. Since subjects could not anticipate this contingency, preceding events were functionally part of the adaptation phase Thus, the first CS+ presentation in acquisition accompanied a shift in the subje c t's experience of event-relations and is a pure acquisition trial. B y similar logic the last CS+ trial of acqui s ition was not r e inforc e d with shock. As b e for e, subjects could not anticip a t e thi s new contin ge nc y and th e r e for e, all prec e ding ev e nt s w e r e function a ll y p a rt o f a cqui s iti o n In a ddi t i o n to minin1i ze th e t e nd e ncy for s ubj ec t s to expect a r e v e r sa l of th e co n tinge n cy, a n unr e in force d C S bu ffe r tri a l --w hi c h was no t a n a l yze d was int rod u ce d be t wee n acq ui s i tio n a nd ex tin ct i o n for a ll subjec t s. The i nt e n t of these l atte r t wo manipu l a ti o n s was t o pur ify interpretatio n of the extinc ti on p h ase

PAGE 37

30 Since the startle probe is prepotent aver s ive US, probes were not presented during acquisition so as not to be associated with learning During prior adaptation and subsequent extinction phases only, probes were presented during and between slides As noted earlier, each of the 2 CS slides was presented and probed 4 times during the adaptation series. These 8 trials were intermixed with 4 no slide/probe trials (12 total trials). Because of the between-group shock exposure manipulation, adaptation was divided into two equal parts. In each of four extinction blocks, each slide was seen and probed twice, and 1 no slide/probe was presented. Thus, there were a total of 8 CS+, 8 CS-, and 4 no slide/probe trials. Throughout conditioning, slide exposure was 6 sec. Variable interslide intervals ranged from 10 to 20 sec in 5 sec increments Postlearning: Affective ratings and debriefing After the last extinction trial, the experimenter removed the recording sensors, shock electrode, and headphones. The experimenter inform e d the subject that the noise and shock would no longer be presented. Th e experimenter then instructed the subject that he or she would view a final se ries of slides and rat e each slide again using SAM. The procedure was id e ntical to prelearning ratings, except that there were only 12 slides. Two of th es e slides were the subject's CS+ and CS-. The remainder were the 10 adjunct slides falling into the five affective categories mentioned above. After compl e ting the slide rating s, the subject also rated (from memory) the noise and s hock stimuli with SAM. After affe c tive judgments were com p l e ted, subjects were a s ked in a debriefing sess i o n to state (1) th e ir es tim a t e the numb e r of s ho c k s received; (2) wh e th e r t h ey reco gni z ed a n y rul e during t h e pro ce dur e; (3) w h e th e r so m e s timuli were fo ll owe d by sh ock m o r e often t h a n oth e r s; (4) w h et h e r th ey h ea rd th e n oise ove r th e h eadp h ones; (5) if t h ey kn ew w hi c h phys i o l og i ca l

PAGE 38

31 measures were being recorded; and (6) whether they knew or could guess the purpose of the experiment. In addition, they were also asked to rate each of the following items for their aversiveness (i.e., "the extent to which you would avoid the stimulus") on a 1 (not at all aversive) to 7 (extremely aversive) scale: "noise over the headphones," "shock," "shocked slide," and "nonshocked slide." After these debriefing questions, the subject was informed about the purpose of the experiment, thanked, given class credit, and dismissed Design and Data Analysis In previous human aversive learning studies, subjects have received shock exposure prior to learning. In this experiment, however, designed variations in the timing of shock exposure across groups complicate interpretation of the adaptation phase data. Thus, to facilitate interpretation of the conditioning data within this context, the effects of shock exposure timing on startle reflex, tonic visceral levels, slide visceral responses, and habituation rates for these measures are reported first. Otherwise, this study largely repeats (with minor improvements) the discriminative conditioning procedure used by Hamm et al. (1991). To facilitate comparisons with those data, univariate analyses of physiological measures (startle magnitude and onset latency, electrodermal response, and heart rate response) were conducted separately for each phase of conditioning (adaptation, acquisition, and extinction). Affectiv e ratings (valence and arousal) were assessed before and after associ a tiv e l e arning The exp e rim e nt was divided, b o th from the standpoint o f d es i g n and a n a l ys i s, int o severa l p a rt s, h e r ea ft er r efe rr ed t o as (1) P relearning, or i niti a l s lid e vi ew i ng/ r ating; (2) Sensitization; (3) Associative l earning, comp ri se d of adap t at i on, acquisitio n a n d ex tin ct i on phases; a n d (4 ) Postlcarnin g. It i s

PAGE 39

3 2 important to note that the adaptation slid e sequen c e is common to both the sensitization and associative learning segments Prelearning The first objective was t o assess the ratings distribution for each subject and--based on the subject's a p riori group assignment--use these ratings to select two matched slides as foregrounds for conditioning. Thus, analyses of these ratings served as a manipulation check that the four conditioning groups significantly differed in valence and arousal, and that conditioning slides within each group did not differ prior to learning. Slide Valence (2: Unpleasant, Pleasant; between-subject) X Slide Arousal (2: Calm, Aroused; between-subject) X Stimulus (2: CS+, CS-; within subject) univariate analyses of variance (ANOVAs) were conducted on both valence and arousal ratings. Similar analyses were conducted on adjunct (i.e., nonconditioned) slides to determine their distribution in the affective space Sensitization To assess shock sensiti z ation of the startle reflex without the confound of different affective slide fore grounds analyses focused on startle probe respon s es during i nt e rslid e interva l s. Missing dat a (2 % of all prelearning and adaptation trials) were e sti mated by regression separately for each exposure group One subject in th e Immediate exposure group was excluded because more than 50 % of his re s ponses had been rejected leading to insufficient data for estimation. Chan ges in the probe startle response as a consequence of shock exposure were analy z ed in an ANOVA for Exposure Group (2 : Immediate, Delayed; betweens ubj e ct) X Interslide Probes (2) X Probe Blocks (3: pr e l ea rning, first half ad ap tation, s e cond half a daptation; within subj e ct) T o exa m ine sens i t i zatio n of t onic viscera l l eve l s d u ring shock exposure, ave ra ge h ea rt r a t e and s kin co ndu c t ance for t h e 1 sec pr i o r t o each adap t ation tr ia l ( in t he absence of slide stimu l ation) were computed for each subject.

PAGE 40

33 Since there were 12 adaptation "trials" --interslide intervals were treated as trials here--the change in pretrial baseline levels during shock exposure could be examined. Although tonic visceral level data were not available during the prelearning rest period, the effect of shock exposure on visceral levels could be tested for the Delayed exposure group since, for these subjects, adaptation block 1 preceded exposure, and adaptation block 2 followed exposure. These data were explored using Exposure Group (2: Immediate, Delayed; between-subject) X Trials (6: consecutive, regardless of CS+, CSor interslide) within Trial Blocks (2: first half adaptation, second half adaptation; within-subject) ANOVAs To examine effects of shock exposure on slide visceral reactivity, slide by-slide heart rate leg scores and skin conductance responses (consecutive trials, without regard to CS+ or CS-, within each half of adaptation, as well as acquisition and extinction) were analyzed, separately for each exposure group. Slide visceral responses were tested using Exposure Group (2: Immediate, Delayed; between-subject) X Slide Trials (4: consecutive, regardless of stimulus) X Blocks (2: first half adaptation, second half adaptation; within-subject) ANOVAs. Associative learning. All physiological data were analyzed in blocks of two trials separately for each phase of conditioning (adaptation series [2 trial blocks], acquisition [4 trial blocks], and extinction [4 trial blocks]) with univariate ANOVAs. Subjects were randomly assigned to receive one of four slide affective foreground s during conditioning. Slide Vale nce and Arousal (2 l eve l s eac h) constituted the between-group factors in the design. To evaluate conditio nin g within g roups over trials, Stimulus (2: CS+, CS-) and Trial Block (2 or 4 within phase) factors formed th e remaining factors of the design For ANOVAS involving the trial block repeated measure term,

PAGE 41

34 unadjusted degrees of freedom and Greenhous e -Geiss e r epsilon (e) corrections are reported. In analyses of associative learning, "discrimination" was defined as the extent to which CS+ response exceeded CSresponse during acquisition or extinction. Startle "potentiation" was defined as the degree to which CS+ response increased from adaptation block 2 to extinction block 1. Since visceral responses rapidly habituate, potentiation of visceral activity was tested as the change in CS+ response from adaptation block 2 to acquisition block 1 Postlearning. For affective judgments, changes in affective ratings of each subject's CS+ and CSbefore and after conditioning were tested in univariate Phase (2: Pre, Post) X Stimulus (2: CS+, CS-) X Slide Valence (2: Unpleasant, Pleasant) X Slide Arousal (2: Calm, Aroused) analyses of variance for valence and arousal r a tings Statisti ca l tests evaluated the hypothesis that CS+ ratings should become (or r emain) more unpleasant and arousing than CSrating s. Similar analyses were conducted on adjunc t slides to assess the degree of affective change Individual differences R e lationships of the t e mperament dimensions to both aversive reactivity and learning were initially examined with zero order correlations and stepwis e multiple regression The EASI subscale scores were entered as independent variables in the regression models to predict measures of aversive r eac tivity (e g. shock tolerance, prelearning startle response, shock avers iveness, situational fea r f uln ess) and the extent of aversive learning (e .g., startle se nsiti za tion and poten ti atio n cardiac and e l ectroderma l cond itioni ng, affect ive r atings change). Follow -u p ANOVA s were cond u cted to describe group mean differen ces

PAGE 42

35 Relationships of shock tolerance and rated shock aversion to learning measures were also examined. Stepwise multiple regression was used to enter both variables to predict the extent of learning in the psychophysio logical response measures. Again, ANOVAs were conducted to describe mean differences among groups. For all statistical tests, the size of the rejection region was set at .05.

PAGE 43

RESULTS The strategy for data analysis and presentation proceeds as follows First, the effect of shock exposure on response sensitization during the initial slide adaptation series is examined. Second, the influence of aversive classical conditioning on psychophysiological responses is evaluated, separately for initial adaptation, acquisition, and extinction slide series Third, effects of individual differences (temperament characteristics, tolerance of shock, and judgments of the aversiveness of stimuli) on psychophysiological reactivity and associative learning are assessed. Sensitization Startle Probe Response Reflex magnitude It was predicted that after exposure to shock, each group would show a marked rise in startle magnitude, and that this would not occur under control conditions. As mentioned earlier, this was tested using interslide interval data to avoid the confounding influence of slide valence or arousal on sensitization. As predicted, startle magnitude was clearly sensitized by shock exposure (see Table 1). The omnibus test of sensitization learning was significant, Exposure Group X Trial Block F(2,124)= 7.11, e= .9 1, p< 002. Planned comparisons testing the sources of this interaction used blocked averages for each subject, since interslide responses within prelearning and adap tation series probe blocks did not differ There was no group difference in startle magnitude during prele arning (i.e., before shock exposure) f(l,62)= 1.77, p= .19. For the Immediate exposu r e group, the incr eme nt in startle ma gni tud e following shock 36

PAGE 44

37 Table 1 Startle magnitude means (ADUs + standard errors) during prelearning and interslide intervals of adaptation. Exposure Group Immediate Delayed Prelearning Adapt. block 1 Adapt. block 2 364 (51) 236 (65) (shock exp.) Table 2 502 (59) 290 (72) (shock exp.) 424 (56) 415 (81) Startle onset latency means (msec + standard errors) during prelearning and interslide intervals of adaptation Exposure Group Immediate Delayed Prelearning Adapt. block 1 Adapt. block 2 38.3 ( shock (1.0) exp.) 45.9 (2 1) 35.1 (1.0) 39 2 (1.2) ( shock exp.) 36 1 (1.1) 36.7 (1.6)

PAGE 45

E 'vl o.~ J:::\:., QJ ::, "00 .2 I Cl3 Cl3 s.s ]-f ,.0 M c:: Cl3 .... Q,) Q,) Q,) bO M c:: 0... Cl3 +J ...c:: rJJ M u...:: 200 150 100 50 0 Immediate Shock Exposure / /0 / / / /\ 0 Control Period (n s change) 2nd preleaming trial 1 Block 1 Delayed 1~ Shock Exposure 2 rControl Period (n s change) 3 4 Block 2 Interslide Intervals During Adaptation Figure 4. Shock sensitization of startle reflex magnitude. Sensitization is measured as the response change from prelearning to adaptation block 1 (Immediate exposure group) or from adaptation block 1 to block 2 (Delayed exposure group) w co

PAGE 46

39 exposure (~= + 138 ADUs) was significant, F(l 47)= 18.57, p< 005. After the sensitization peak, startle response habituated, F(l,47)= 12.30, p< .005. For the Delayed exposure group, the increase from prelearning to adaptation series block 1 (control period;~= +54 ADUs) was not significant, F(l,15)= 3.22, p> 09. As predicted, the subsequent increase in startle reflex magnitude directly attributable to shock exposure (~= + 125 ADUs) was statistically supported, F(l,15)= 9.42, p< .01. As seen in Figure 4, startle magnitude increased markedly at the first probe after shock exposure for each group. The size of the sensitization effect was equal across groups (about 40%), and absolute startle magnitude in block 2 of the adaptation series--after both groups had been exposed to shock--was nearly identical (see Table 1). This is noteworthy, since the end of the initial adaptation series was intended to serve as the general baseline for the test of startle potentiation. Also, startle reflex sensitization was independent of orbicularis oculi tension (Appendix D, Table 7) and responses during slides were comparable with responses during interslide intervals Startle onset latency. Although latency differences were obtained, the findings were not consistent with a shock sensitization effect (Table 2) Shock exposure was expected to speed onset latency (first for the Immediate group, later for the Delayed group), but latencies were generally speeded from prelearning to the first block of the adaptation series for both groups f (2,122)= 15.84, e= .88, p< .0000. A Group X Block interaction did emerge (which would be anticipated for sensitization) F(2,122)= 5.0 6, e= 88 p < 02 but thi s was due to overall s lower responses in the Delayed group, F(l,61) = 6.66, p< .02, which became significan tl y quicker in onset by the second block of th e adap t ation series. Thus, th e present data do not s up po rt th e conclusion that s t a rtl e onset latency is re li ably related to shock exposure.

PAGE 47

40 Visceral Reactions Skin conductance. Electrodermal tonic levels habituated within blocks of adaptation trials for both groups. Although Table 3 suggests a group difference in levels in block 1 of the adaptation series, this effect was weak (p= .15). There was a large increase in conductance level for the Immediate group from adaptation block 1 to 2 (i.e., after questionnaires), but not for the Delayed group (i.e., after shock exposure), Group X Trial Block F(l,60)= 6.42, p< .02. The source of this interaction was the significant increase in the Immediate group. Importantly, conductance levels of the Delayed group were not sensitized by shock exposure (p= .64). Tonic levels of the two groups did not significantly differ in acquisition or extinction. Slide conductance responses of both groups increased from block 1 to 2 of the adaptation series; the slightly greater increase for the Delayed shock exposure group was not significant, Group X Trial Block F(l,62)< 1 (Table 3). Tested alone, the increase in the Delayed group following their exposure to shock was not reliable (p= .14). Since slide electrodermal response increased in both groups, dishabituation may have contributed here. Incidentally, the Immediate group had larger mean skin conductance responses to slides throughout conditioning, but this overall group effect was only significant in acquisition, F(l,62)= 3.85, p= .05. The Immediate group also had larger Second Interval anticipatory responses than the Delayed group in acquisition (Ms= .14 and .09 S), F(l,62)= 5.33, p< .03, but the groups did not differ in Third Interval Response magnitude. Shock exposure groups did not differ on Second Interval or Third Interval response measures in extinction. Heart rate. Heart rate levels of the Immediate group generally decreased from block 1 to block 2 of the adaptation ser ie s, whereas l eve l s of the Dela ye d group increased s lightl y (see Table 3). There was a Group X Trial Blo ck

PAGE 48

41 Table 3 Mean visceral levels and slide responses for Immediate and Delayed shock exposure groups. Visceral Adaptation Acquisition Extinction Measure block 1 block 2 Skin conductance tonic level (S) Immediate 6.42 7.76 7.05 6.79 Delayed 7 75 7.92 8.03 7.69 Slide skin conductance response (log [FIR+ 1) S) Immediate 0 23 0 24 0.22 0.13 Delayed 0.15 0.19 0 13 0.08 Heart rate tonic level (bpm) Immediate 71.39 69.14 68.68 69.35 Delayed 69.52 70.47 70.31 70.08 Early-interval slide deceleration (D1; L1bpm) Immediate -3.33 -3.47 -2.83 -2 81 Delayed -3.90 -3.99 -4.71 -4.00 Mid-interval slide acceleration (Al; L1bpm) Immediate 2.69 3.30 3 30 3.51 Delayed 4.96 4.35 3.70 3.62 Late-interval slide deceleration (D2; L1bpm) Immediate -3.43 -2 .89 -5 00 -4.20 Delayed -3 .85 -3.45 -6.28 -6.15

PAGE 49

42 interaction, F(l 61)= 4.33, p < .0 5 whi c h w as due solely to h a bituation o f levels across blocks in the Immediate group, F(l,46)=9 23 p< 005; that is, tonic heart rate levels did not increas e (i e subjects were not sensitized) in the Delayed exposure group (p= .53) Heart rate levels did not differ across groups during conditioning Shock exposure also did not sensitize cardiac response to the slides during the adaptation series although group differences emerged later in conditioning (Table 3). For initial deceleration (Dl leg), the Delayed group decelerated more than the Immediate group during acquisition, f(l,61)= 8.02 p< 01, and extinction, f(l,61)= 4.41, p< .05. The Delayed group showed greater mean deceleration late in the slide interval (D2 leg) in acquisition, but this was not significant (p= .13). In extinction, the group difference in late deceleration was significant f(l 61)= 4 76, p< .04. There were no group differences in Al leg respons e during an y part of conditioning Summary Onl y startle magnitude was signif i c a ntly influenced b y s hock expo s ure : Startle responses were 40 % larger immediately following s hock relative to control conditions, suggesting that this increase cannot be a ttributed to dishabituat i on or simple passage of time In contrast, neith e r startle onset latency nor visceral reactions were sensiti z ed by e x posure to shock. Associative Learning Prior to learning the CS+ slide CSslide, and 10 a d j u nct s lid e s were selected fr om each subject s r a ting s di s tribution (see Appendi x A).1 Mean 1 Th e t er m s Ne gati ve," P os iti ve," "C alm" and Ar ousi n g and th ei r co mbin a ti on (e.g., "Neg ati ve/Ca l m") refer to th e d ifferen t conditio ni ng g r o u ps v i ewing s lid es o f di ffere nt affective va l e n ce or arousal. To minimize confu s i o n th ese gro up r e f e r e n ces w ill be d i s tin g ui s h e d from th e l ower-case

PAGE 50

43 valence and arousal ratings for conditioning and two control slides--which were drawn from the same-affect category as each subject's CS slides--are presented in Appendix D (valence, Tables 8-9; arousal, Tables 10-11). Adjunct slide ratings (Appendix D, Tables 12-13) were not significantly affected by conditioning. Affective Ratings Valence judgments. Figure 5 illustrates each group's affective valence and arousal ratings of their foreground slides both before and after learning. Valence ratings of CS+ and CSslides did not differ prior to learning. Not surprisingly, affective valence ratings by subjects assigned to view pleasant CSs were higher than ratings by subjects assigned to view unpleasant CSs (regardless of arousal), F(l,60)= 1910.38, p< 0000. Regardless of content, CS+ was rated more unpleasant than CSafter learning (Ms = 3.95 vs. 4.55), f(l,60)= 9.78, p< .003; i.e., valence discrimination did not significantly vary by affecti ve slide condition. Analysis of va l e nce change scores also indicated that overall, CS+ became signi fican tl y more unpleasant than CS(~s= 81 vs. -.25 units), f(l,60)= 7 .8 7 p< .01. Figure 5 shows that the direction of ratings change was opposite for subjects viewing unpleasant vs. pleasant foreground s, F(l,62)= 54.86, p< .000 0 Follow up tests found that for Positive slide groups, CS+ became signific an tly more unpleasant(~= -2.19) whereas for Negative groups, CS+ did not significantly change (~= +.56). Dependin g on the slide foreground, conditioning and control slides (from th e same affective category as th e subjec t's CSs, but not conditio ned) terms "unpleasant," "p leasant," "calm," and "a rousing ," which denote th e affective qualities of slide pictures or subject ratings of th e pictures.

PAGE 51

pre 0 Cf) bO s ...., !'O (J) u (J) ....... !'O > 44 p os t .... 1 CS + 9 CS0 8 Positive/Calm Positive/ 7 Arousing group group 6 2 3 4 5 6 7 8 4 Negative/Calm 3 Negative/ group Arousing 2 group 1 Arousal Ratings Figure 5. Affective valence and arousal ratings of conditioned stimuli before learning (origin of lines) and after learning (terminal point of arrows), separately for each slide foreground group. 9

PAGE 52

..--.. en ... bO C .o -Q) 0... I ... en 0 0... ......., Q) bO C ,.c; u Q) u C Q) ......... > 1.5 1.0 0.5 0.0 -0.5 -LOj CS+ I'll CS-1.5 Control -2 0 -2.5 Unpleasant Pleasant Affective Valence of Slide Foreground Figure 6. Change in affective valence ratings from prelearning to post learning for CS+, CS-, and same-affect control slides. Valence change is illustrated separately for groups viewing unpleasant (left) and pleasant (right) slide foregrounds during conditioning. .j:v,

PAGE 53

46 showed different relative change, Slide Valence X Stimulus F(2,124)= 9.93, p< .0001 (see Figure 6). Follow up tests found that, for Negative groups, CS became significantly more pleasant than both CS+ and control slides. In contrast, a gradient resulted for Positive slide groups, in that CS+ became marginally more unpleasant than CS(p< .07), whereas both CSs became significantly more unpleasant than control stimuli. The shock US and startle probe were rated as equally unpleasant after learning (Ms= 1.97 and 2.16, respectively). Arousal judgments. Arousal ratings of the CS+ and CSslides did not differ before learning. Due to the experimental selection criteria, arousal judgments by subjects assigned to view more exciting CSs were significantly higher than ratings by subjects assigned to view calm CSs (regardless of valence), F(l,60)= 62 31, p< .0000. Unlike valence ratings, classical conditioning did not produce arousal judgment discrimination. The overall difference between CS+ and CSwas not significant, F(l,60)< 1, and did not vary by slide foreground. Arousal change scores also failed to provide evidence of learning. Although CS+ maintained slightly more arousal than CS-, the overall difference between CS+ and CSchange was not significant, f(l,60)< 1. However, the direction of ratings change was opposite for subjects viewing calm vs. arousing foregrounds (see Figure 5), f(l,62)= 15.96, p< .0005. For Arousing groups, CS+ became significantly more calm (L\.= -1.59), whereas for Calm groups, CS+ did not significantly change (L\.= +.60) There were no other significant findings. Depending on the slide foreground group, conditioning and control slides show e d different relative change Slide Arousal X Stimulus F (2 124) = 5 16, p< 01 (see Appendix D, Table 11 ). For Arousing slide g roup s, both CS+

PAGE 54

47 and CSbecame significantly more calm than control slides (ps< .02). Calm slide groups failed to show this effect; differences in arousal change between control slides and both CS+ and CSslides were not significant. Arousal ratings of the shock and startle probe were similarly high (Ms= 7.19 vs. 6.77); even so, this small difference was reliable, F(l,60)= 4.08, p< .05. Startle Probe Response Reflex magnitude. As expected, CS+ and CSstartle magnitude did not differ before learning (see Appendix D, Tables 14-15 for reflex magnitude means for all groups during conditioning). There were no other significant effects on startle magnitude during the adaptation slide series-2 It was predicted that startle magnitude evoked during the CS+ would increase immediately after learning Clearly, however, the ability to detect change at extinction relies on the initial level from which change is assessed Recall that delaying shock exposure for some subjects in this sample was a departure from other conditioning studies (in which subjects have routinely been exposed to shock prior to the adaptation series). Although mean startle responses of the two groups did not differ in the second block of adaptation, responses of the Delayed group were relatively high within that group due to their recent shock exposure; conversely, responses in the Immediate group had decreased from their earlier postshock exposure (see Table 1 ). Thus the larger respons es for Delayed subjects may not have fully habituated, making it more difficult to detect potentiation. 2 Subjects viewing unpleasant slides had larger mean startle responses (across CS+ and CSpictures) than subjects viewing pleasant slides during the adaptation ser i es ( Ms = 518 vs. 360 ADUs), F( l 60) = 3.18 p< .08. Covarying for prelearnin g s t art l e magnitude l evels, how e ver eliminated thi s marginal e ff ec t Th ese appare nt s lid e foreground respo nse differ ences could th e refor e be attributed to init i al subject response differences.

PAGE 55

48 Overall analysis indicated that th e CS+/ s hock pairing produced a mean increase in startle magnitude from block 2 of the adaptation series to block 1 of extinction (~= +41 ADUs, from 443 lo 484 ADUs), but this change was not significant, F(l,62)= 1.85, p< .18. A clos e r look at the trials immediately surrounding the acquisition phase (i.e., th e last trial of adaptation and first trial of extinction) showed that startle was sign ificantly potentiated across all subjects (~= + 139 ADUs, from 414 to 553 ADUs), F(l,61)= 5.20, p< .03. There was, however, a dramatic diff e rence in startle reflex potentiation between the two groups differing in the timing of shock exposure (Figure 7). The Immediate exposure group showed start le potentiation (~= +83 ADUs), whereas the Delayed group showed a commensura te decrease(~= -86 ADUs), F(l,62)= 6.62, p< .02. Tested alone, the in crea se in the Immediate group was significant, F(l,47)= 6.77, p< .02. Thu s, in a statis tical test dir ec tly comparable with previous studies, fear-potentiated start l e was replicated. Overall mean CS+ response wa s significa ntly larger th an CSresponse throughout extinction, F(l,62)= 5.84, p< .02, and this discrimi nation did not vary over slid e trial blocks (see Figure 8). Clearly, startle magnitude declined overall during extinc tion, F(3,186)= 35.94, p< 0000. There were no Slide Arousal or Shock Exposure group effects o n extinc tion startle discrimination. In two r ecent experiments, startl e discrimination tend ed to be greater among subjects viewing pleasant slid es th a n unpleasant slides In this study, this effect failed to replicate, Slid e Val e n ce X Stimulus F(l ,62)< 1 (Figure 9). An effect of Slide Valence on CS+ s t artle potentia tion (cf. Hamm et al., 1991) was also not replicated, F(l,62) < 1. Consis t ent w ith the se r ecen t studies, the Negative/ Arousing g r oup s h ovved sm,dler mean sta rtl e pot e ntiation than other groups (-7 vs. 1 22, 15, and 32 ADUs for Nega tiv e / Calm, Pos itiv e/Ca lm, a nd Po s itiv e/ Arousi n g gro up s, rcspl'clivcly). Howeve r statistical ana l yses of

PAGE 56

,,......_ Vl ..... 0 I < ..__.. (l) "O .2 ... (,:I ::;E ] p:) 49 520 500 480 --0Immediate 460 Delayed 440 420 400---------,,--------------,---Adaptation block 2 Blocks of 2 CS+ Trials Extinction block 1 Figure 7. Extent of CS+ startle potentiation in groups exposed to shock immediately before the adaptation series (open circles ) or delayed exposure to shock midway through the adaptation seri e s (dark squar es).

PAGE 57

475 .......... cJ) ..... 425 ;:::s 0 I < ..__.. Q) "O ;:::s ..... 375 ..... So ro ::E C ..... ......... i:o 325 275 50 CS+ ---0CS. Adapl Adap2 Extl Ext2 Ext3 Ext4 Figure 8. Startle magnitude differentiation in the overall samp l e (n=64). Mean responses are presented in blocks of 2 trials during the adaptation and extinction slide series. Note that probe reflexes decrease in extinction to a point which is below prelearning levels (dashed line), and that CSresponse reaches this level before CS+ response.

PAGE 58

51 450 400 Pleasant Pi ct ures 7 tr} ..... ..... 350 0 I < ..__, Qi 300 "O 2 .... So 11:S 250 ; i:o 200 S150 Adapl Adap2 Extl Ext2 Ext3 Ext4 600 550 Unpl easant Pictu r es tr} ..... ... 500 0 I < ..__, Qi 450 "O 2 ... So 11:S 400 ; i:o 350 300 Adapl Adap2 Extl Ext 2 Ext3 Ext4 Figure 9. Sta rtl e magnitudes during adaptation and extinctio n slide se ri es of classical cond iti o ning, sepa r ately for groups (n=32 each panel) viewing e ith e r pleasant slides (above) r unpl easa nt pictures (be l ow). Note the pattern of reflex habituati on i.n each group relative to prelearning l evels (dashed lin es).

PAGE 59

,-.. u Cl) (/) E: ......... >-. u I::'.: Cl) ..... ca ..... Cl) (/) 8 Cl) +:l I-< ca ..... CJ) 35 34 I 33 I 32 I IICS+ -QCS0 ------31-------------------.-----------......... -adaptation block 2 block 1 block 2 block 3 block 4 Extinction slide series (blocks of two trials) Figure 10. Startle onset latency during classical conditioning in the overall sample. Facilitation (speeding) of CS+ latency and inhibition (slowing) of CSlatency in the first extinction trial block, relative to the second trial block of adaptation. Onset latency discrimination can be observed in the difference between CS+ and CSonset latencies throughout extinction. l/1 N

PAGE 60

53 potentiation scores did not s upport these im press i o ns (p> .20) Finally, retesting Slide Valence effects only for the In1mediate exposure group (which is comparable with previous studies) did not c h a nge these null findings-3 The time at which prob e s were deliver e d influenced startle magnitude in extinction. Startle respon se was generall y l a r ge r to probes presented late during the slide interval (regardless of CS+ o r CS-) relative to earlier probes (Ms= 358, 340, 374, and 387 ADUs for times 14). Variation across probe times was significant, F(3,186)= 3.83, e= .90, p< 02; lin e ar F(l,62)= 5.11, p< .03; quadratic, p< 15. Further confirming this eff e ct, mean response at the last two probe times (381 ADUs) was greater than ov e rall startle response at the first two probe times (349 ADUs), F(l,63)= 8 2 9 p< .01. Startle discrimination was consistent across probe times, Stimulus X Probe Time F(3,186)< 1. Onset latency. Th e re w e re no significant s li de group, stimulus or trial block effects for startl e l a t e ncy before aversi ve l ea rning. (See Appendix D, Table 16, for onset lat e n c y m e ans in all g rou ps and the overall sample ) Startle onset l a t e n cies we r e evalu a t e d for both l e arned facilitation a nd discrimination, analo go u s t o s tartle magnit ude. F i g ure 10 illustrates st a rtl e latencies from the end of adaptation (i.e., th e base for facilitation) through extinction. Due to mi s sing data, nine subj ec t s were e xcluded from these analyses (n= 55) 3 Unexpectedly Slide Valence influ e n ced c h a nge in CSstartle response from adaptation to e x tinction, F(l,62)= 4 63 p< .04. Probe response to CSwas significantly potentiat e d ( Li = +8 2 ADU s ) for Negat i ve slide groups, F(l,31) = 4 16, p= .05, but CSr es pon se did not s i g nifi ca n t l y c h a nge (Li= -25 ADUs) for Positive groups Alth o u g h thi s findin g s u ggests grea ter generalization o f CS+ s t a rtl e potentiation t o CS w h e n s ubj ec t s v i ewed unp l e asant slides, it s h o uld be v i ewe d cauti o u sly: Differe n t pa tt erns of CSstartle c h a n g e wer e ob t a in ed in p r evio u s ex p e ri ments assessing affective modulc1tion of th e st a rtl e r esponse (Greenwa ld e t a l. 1990; Hamm et a l. 1991)

PAGE 61

54 From adaptation to the first blo ck of extin c tion, CS+ latency was speeded, whereas CSlatency was slowed The interaction of these changes was significant, F(l,51)= 4 41, p< .05. Wher eas CSs lowing (~= 1.4 msec) was significant, F(l,51)= 4 15, p< .05, CS+ speeding was not (~= 1. 1 msec, p= .24). Probe response to CS+ was faster overall than CSre s ponse during extinction (Ms= 32.4 vs. 34.1 msec), F(l,51)= 8.31, p< .01. Although CS+ onset latencies returned to prelearning levels, discrimination did not significantly vary over trial blocks. Onset latency of the reflex between slides (M= 39.0 msec) was significantly slower than during slides. Similar to startle reflex magnitude, m ea n onset latency discrimination in extinction was smaller for the Negative/ Arousing group than the other slide groups (0.1 vs. 2.2, 1.5, and 2.5 msec for Negative /Calm, Positive/Calm, and Positive/ Arousing groups, resp ec tivel y) However, the statistical test of this impression was not significant (p = .19). Th e re we re no other remarkabl e effects of affective slide content on modulatio n of sta r t le onset latenc y. Visceral Reactions Skin conductance Appendix D (Tables 17-18) presents First Interval skin conductance responses for all slide foreground groups throughou t conditioning. Electrodermal response ma gnitude decreased from adaptation trial 1 to 2, increased at trial 3, and decreas ed again from trial 3 to 4. This within block response habituation was significant F(l, 60)= 24.68, p< .0000 The increase between adaptation blocks prob ably represents dishabituation due to either shock exposure or completi on of questio nnaires (see above). Analogous to startle potentiation, the Im.m e diate exposure group showe d incre ase d conducta n ce r espon se fr o m the e nd of the adaptation series to th e first bl ock of acqui s i tion, wh e r e J. s t h e D e layed exposure group s r es ponse decreased (~s= .051 and -. 05 -J: 1 6; see fig u re 11 ). This differ e n c e in

PAGE 62

0.30 ,--, + 0.25 & 00 0 ...... ........., 0.20 55 -OImmediate Delayed 0 15-----.......-------------.----Adap2 Acql Blocks of 2 CS+ Trials Figure 11. Extent of CS+ skin conductance response potentiation in groups exposed to shock before adaptation (open circles) or delayed exposure to shock midway through adaptation (dark squares)

PAGE 63

-rJ) ::1. ,........ + 0::: ti: ........ b.Q 0 ...... ..__... Q) "'O ;::, ..... ..... ro c:i::: u rJ) 0 3 I 0.2 0.1 0 0 1 2 Adapt. I aCS+ --oCSo" 0~ 1 2 3 4 1 2 3 4 Acquisition Extinction Figure 12 Skin conductance response differentiation in the overall sample (n=64) Mean responses are presented in blocks of 2 trials throughout conditioning. Note the presence of both differentiation and habituation in acquisition and extinction V,

PAGE 64

57 conductance pot entiation was significant, Group X Trial Block F(l,62)= 3.96, p= 05 The two groups did not differ in their ski n conductance responses at the end of adaptation. In acquisition, First Interval Respon se m ag nitude to CS+ was larger than to CSoverall (Ms= .24 and .15 S), F(l,60)= 38.05, p< .0000 (Figure 12) Conductance response magnitude habituat ed during learning (Ms= .22 .21, 19, and .17 S, for blocks 1-4), F(3,180)= 5.19, e= .84, p< .005. Second Interval Response magnitude, refl e cting shock anticipation, was greater during CS+ than CSoverall (Ms= 18 and 07 S), F(l ,60)= 47.96, p< .0000. Relative to CS-, the Second Interval Response to CS + increased monotonically across trial blocks, F(3,189)= 5 47, e = .97, p< .002. Not surp risingly, unconditioned responding to shock (i.e., T hird Interval Response magnitude after CS+; M= .SO S) was dramati ca ll y greate r than acti vity after CSoffset (M= .06 F(l,60)= 320.53, p< .0000. These uncondit ioned responses to shock declined over trial blocks (Ms= .57, 51, 45, and .45 ~LS), F(3,189) = 4.27, e= .80, p< 01. In extinction, the First Interval R esponse was g reater for CS+ than CS (Ms= 14 and .09 S; see Figure 12) Thi s overall discrimination effect was significant, F(l,60)= 11.65, p< .002. First Interval Response magnitude which declined from acquisition, co ntinued to decrease in extinction. This general decline across extinction trial blocks (Ms= .14, .12, 11, and .10 S) was s ignificant, F(3,180)= 4.09, e= 99, p< .01. Whereas both Hamm et al. (1991 ) and Greenwa ld et al. (1990) found that e l ectr odermal di scrimination in extinction was l a rger in the context of ca lm slides, neither Slide Arousal nor Valen ce influ ence d con ductan ce response differentiation in thi s experiment (both 1T1ain ffcct Fs< 1). Consistent with the se past studi es, co n ducta nce discrimination wa virtually absent in the ega tive/ Arou s in g group, r e l a tive to othL:r groups, in extinct ion (.01 v .. 06,

PAGE 65

58 .04, and 08 S in Neg a tive/Calm, Positi ve/Ca lm an d Positive/ Arousing groups, respectively). This impression of an isol ated group difference (as suggested by the Slide Valence X Slide Arous a l X Stimulus interaction), however, was not supported, F(l,60)= 2.50 p< .12. Electrodermal activity was larger in anticipation of shock onset during extinction. The Second Interval Response to CS+ was significantly greater than to CS(Ms= 24 and .19 S), F(l,60)= 11.14, p< .0 02, and habituated overall (Ms= .33, .22, .17, and .16 S), F(3,180)= 36 .39, e= 93, p< .0000. Third Interval Response magnitude (i.e., due to shock omission after CS+) was small and undifferentiated in extinction. Heart rate Tables 19 22 in Appendix D pres e nt heart rate level and leg scores, by slide foreground group, throughout the experiment. Adaptation prestimulus heart rate levels tended to be hi gher for Positive slide groups than Negative groups, F(l,59)= 3.68, p< .06. Levels were also higher prior to the CS+ slide than the CSslide for the Negative gro ups, Slide Valence X Stimulus F(l,59)= 5.01, p< 03 Because h eart rate l eve ls varied within and between groups, analyses of leg scores durin g th e adaptation series were conducted using baseline heart rate for each CS at eac h trial block within each affective valence group as covariate s. In adaptation, there was significantly l ess cardiac acceleration (Al leg) for CS+ than CS(Ms= 2.55 vs 4.28 bpm), F(l,59)= 9.62, p< .005, which was not altered by covariance analysis, F(l,58)= 5.75, p< .0 2. Initial deceleration (D1 leg) and late-slide deceleration (D2 l eg) tended to be larger for CS+ than CS-, F(l,59)= 3.72, p< .06, and F(l,59)= 4.06, p< 05, but adjusting for tonic heart rate attenuated both findings in bolh in s t a n ce s (ps> .35). In acquisition, cardiac accele r ation wa s th e b e st index of l earning As Figure 13 suggests, the initially small e r accele r at i o n to CS+ (originating in

PAGE 66

Trial Block 1 Trial Block 2 Tria l Block 3 Trial Block 4 iJ 11 1 ~ 1 ,,-... V J "'l \. V I E o i .. 0 0... e OJ -1 ~ 0.0 i:; ro ..c:: u -2 ~ OJ ..., ro ..., -3 ~ 1-, ro OJ :r: -4 ~ -5 \... I -1 v-, .... -1 )._ )T" ~ -1 -2 i \._ -2 i "I'\ -2 -3 i 3 1 \Cl -3 -4 4 1 4 1 -5 5 I I I I I I I I I -5 I I I I I 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 Second s Post-Stimulus Onset Figure 13. Conditioned heart rate response during acquisition trial blocks. Each panel shows the 6 second cardiac w a ve form re s ponse (deviated from pretrial ba s eline) averaged across two CS+ and two CStrials. Note the gradual deve l o pment of the rnidinterval accelerative component in the CS+ waveform, whereas CSresponse becomes progre s sively more decelerative over slide trial block s 'o 'o lJl I 6

PAGE 67

60 adaptation, but not significantly diff erent from CSresponse in acquisition block 1) gradually increased; there was a concomitant decrease in acceleration to the CSslide over trials Although the overa ll learning effect (i.e., CS+ more acceleratory than CS-) was not s ignificant, F(l,59)= 1.84, p= .18, there was a linear increase in discrimination wit h r epea ted slide/ shock pairings The Stimulus X Trial Block interaction was marginal, F(3,177)= 2 60, e= 93, p< .06, but the difference in CS+ vs CSlinear t rends across blocks was significant, F(l,62)= 7.99, p< .01. Thus, associative learning reversed the initial disparity in slide cardiac response from the adaptation series and did so monotonically Furthermore, the learning effect was a function of both incremental acceleration to CS+ and declining acceleration to CS-.4 Unconditioned cardiac acceleration (A2 leg) to shock after CS+ offset was greater than CSpostslide respon se (Ms= 10.7 vs. 5.8 bpm ), F(l,59)= 45.62, p< .0000. The unconditioned respon se diminished over trial blocks (Ms= 11.6, 11.0, 9.9, and 7.0 bpm), wherea s cardiac response after CSstayed smaller and constant. This differential habi tuation effec t was also significant, Stimulus X Trial Block F(3,180)= 3.28, e= .94, p< .03. In extinction, late cardiac decel e r ation (D2 leg) was greater during CS+ than CS(Ms= -5.2 vs. -4.2 bpm), refl ecting anticipation of shock, F(l,59)= 4 78, p< 04. There were no other signific a nt leg score findings during extinction. 4 Groups viewing arousing slide foregrounds showed the increasing cardiac accelerative differentiation more than Calm slide groups, Slide Arousal X Stimulus X Trial Block F( 3,183)= 2 76, e= .93, p< .05 This was due to incremental l ea rning across block s in the Arous ing slide groups but also reversed cond i ti o ning for Calm group s in trial block 2 (see Append i x D, Table 21) Howev er, modulation of cardia c acce l eration by slide foreground has not be en observed in previous conditioning exper im ents; it s appearance here may b e fortuit ous.

PAGE 68

61 Positive slide groups, however, had higher heart rate levels than Negative groups, F(l,59)= 4.03, p< .05 Summary Clear, significant conditioning effects were obtained across a variety of measures. For the startle probe reflex, magnitude was potentiated during CS+, especially for the Immediate exposure group, and was larger than CS throughout extinction. Similarly, CS+ startle latencies were faster than CS in extinction. For visceral reactions, skin conductance response during CS+ was rapidly potentiated, especially for the Immediate exposure group, and remained larger than that of CSduring both acquisition and extinction. In contrast, cardiac accelerative conditioning proceeded more slowly, with CS+ response gradually exceeding CSresponse late in acquisition; late-slide deceleration was greater for CS+ than CSin extinction. For affective valence judgments, CS+ became more unpleasant (relative to CS-) after learning, especially for subjects who viewed pleasant slides; conditioning produced no systematic changes in arousal ratings overall. In general, startle and visceral responses were not significantly modulated by the affective slide foreground, but valence and arousal judgments were influenced by this manipulation. Individual Differences The relationships of temperament characteristics, shock tolerance, and rated shock aversiveness to stimulus reactivity and associative learning were assessed Temperamental influences are treated as temporal precursors to experimental variables in the subsequent analyses Temp erame nt Ch a ract eristics Consistent with Gray 's (1982) broad hypoth esis that neurotic introverts are more se nsitive to aversive stimu lati on, the EASI fea r and impulsivity t e mp erame nt dim e n sio n s accounted for s i gnificant variabi lity in aversive

PAGE 69

62 reactions; in contrast, dim e nsions of sociability, activity, and anger did not predict measures of aversive reactivity or learning. Figure 14 illustrates the distribution of fear X impulsivity scores for all subjects and cutpoints used in subsequent group analyses. Fearfulness and impulsivity subscale scores were independent in this sample of subjects (r= .09, ns) and the cutpoint-defined groups did not differ on the other temperament dimensions. Reactivity to aversive stimulation. Each person set his/her level of shock to be "highly annoying but not painful" during initial exposure. However, the physical intensity of shock at this subjective level--hereafter, "shock tolerance"--varied widely and was significantly lower for subjects whose temperamental disposition was fearful and inhibited Stepwise regression (Table 4) indicated that fearfulness and inhibition predicted lower tolerance for shock, multiple r= .4 5 Group analysis (see Table 5) revealed a significant main effect for inhibition, F(l,60)= 5.45, p< 03 a marginal effect for fearfulness, F(l,60)= 2 9 8, p < .09, and no interaction ( p= 84). In follow up tests, only highly fearful/inhibit e d and low fear/impul s i ve subjects differed from one another in shock tolerance. Temperamental fear and inhibition also predicted prelearning start l e magnitudes which were significantly larger (multiple r = 32). Group analy s i s indicated a significant fearfulness effect, F(l,60)= 4 01, p< 05, a marginal eff ec t for impulsiveness, F(l,60)= 3.74, p< .06, and no int e raction (p= 87) Again, follow up tests showed that only fearful/inhibit e d subjects and nonfearful/ impulsive subgroups significantly differed from one another. (Mean startl e magnitude of fearful s ubj ec t s c ontinu e d to b e hi g h e r ove rall through o ut co nditi o n i n g but t his difference was n o t significant). Consis t e nt w ith the s t a rtl e m ag nitud e da t a, m ea n pre l ea rn i n g sta rtl e onset l a t e n c i es of fea r fu l

PAGE 70

0 0 0 5 0 63 Fearful I 0 1 000 0 o
PAGE 71

6 4 Table 4 Stepwise multiple regressions of temperamental fear and inhibition (low impulsivity) on measures of aversive reactivity. Dependent Variable FSS total score Shock tolerance Prelearning startle reflex: magnitude onset latency Multiple R Fearful Inhibited Statistical Summary .48 F(l,62)=18.31, p< 001 33 -.45 F(2,61)= 7.86, p< .001 22 .32 F(2,61)= 3.46, p< .05 .25 F(l,60)= 3.88, p< 06

PAGE 72

65 Table 5 Means and standard deviations for measures of aversive reactivity as a function of temperamental fear and impulsivity subgroups (cutpoints = 14 for each subscale) Dependent Variable FSS total score ( 40-200 scale) Shock tolerance (milliamperes) Pre learning startle reflex: magnitude (A-D units) onset latency (msec) Note : Low Fear Inhibited Impulsive n= 20 n= 10 87.2 b 105.3 ab (16.4) (28.5) 6.1 ab 8.3 a (4.1) (4.5) 309.9 ab 131.6 b (243.4) (83.3) 41.6a 42.8 a (8.8) (7.2) High Fear Inhibited Impulsive n= 20 n= 14 112.0 a 107.6 a (19.0) (19.6) 4.2 b 6.8 ab (3.1) (4.5) 465.9 a 315.8 ab (364 .3) (453.2) 40.2 a 37 6 a (7.5) (8.1) Mean n=64 102.2 (22.3) 6.0 (4.1) 332.1 (336.9) 40.4 (8.0) Means without common letters significantly differ (p< .05; using Tukey HSD MSerror [df= 4,60] and harmonic mean= 14.74)

PAGE 73

Individual Differences Moderators Learning Measures temperamentally high fearfulness and inhibition (EASI) multiple r= -.45 shock tolerance (intensity at which the subject reported that the shock was highly annoying but not painful ) .47 rated aversion of the CS+ slide .48 -.35 .2 1 fearfulness of many situations (FSS) .30 rated aversiveness of the shock UCS .43 startle potentiation .33 .21 ----skin conductance potentiation Figure 15. Individual differences in aversive reactivity and associative learning. High temperamental fear and inhibition predicted greater aversive reactivity : less tolerance of s h ock, higher situational fearfulness and higher prelearning startle respo n se (not pictured here) Of the two variables presumed to mediate associative learning rated shock aversivenessbut not shock tolerance--predicted greater CS+ aversiveness, more startle potentiation, and more electrodermal response potentiation. Pearson correlations (all except the one multiple r are zero-order) greater than 25 are significant at p< 05 '

PAGE 74

67 s ubj ec ts were margin a ll y faste r than for n o n fearful s ubjects (r= .25). Gr o up anal y sis yielded a som e wh a t weaker fearfulne ss effe ct, F(l,58)= 2 33, p= 13 As was expected, temp e ramentally f e arful s ubj e cts (independent of impulsivity) reported mor e fear across a vari e t y of s ituations (Fear Survey Schedule; r= 48) This was confirmed by a si g ni fica nt difference in the group analysis, F(l,60)= 6 69 p< 02. Follow up tests indi ca ted that the nonfearful/ inhibited group reported l e s s fear than other sub gro ups Finally, fearful subjects also tended to rate their shock as more a v e rsive (r= .20, ns) which, to some extent, explains the counterintuitive findin g that less shock tolerance was related to significantl y high e r shock aversion in the overall sample (r= 35) Figure 15 summari zes th e relationships b etwee n temperament, low s ho ck toleranc e high sh ock ave rsion, and hi g h sit u a tional fear This p a tt e rn o f d a ta supports th e conclu s i o n that fearful/inh ibite d subjects are mor e gene r a ll y reactive to a ve r s i ve s timulation -in terms of reported situ atio n a l fea r s, behavioral tol e r a n ce of s hock, and refl ex excita bility. Se n s itizati o n Te m perame nt chara c t e ri st i cs were not significantl y r e l a t e d to sensitiz a tion l ea r n ing measures Associative l ea rning Te mperamentally im pu l s ive subjects exhibit e d m o r e s kin conduct a n ce r es p o n se potenti a tion and di s crimination durin g ac qui s ition. Both eff e cts w e r e supported stati st i cally by regression anal y si s, Fs(l,62 ) = 4 36 and 6 .6 7 ps< 05 Correlations betwee n impulsivit y and bo th cond u c t a nc e pot e n t i at i on and di sc riminati o n were mo d e st but si g ni fica nt (rs= 26 a nd .31). Fea r fu ln ess, e nt e r e d s t e p w i se in these sa m e re g r ess i ons, did not significantly pr ed i ct e l ec trodermal l earning. Temperament did not int e r ac t w i th slide affectiv fo r egro und i n d etermi ning aversive learnin g

PAGE 75

68 Rated Shock A versiveness The role of the shock in mediating aversive learning was explored next using stepwise multiple regression. Rated shoc k aversiveness and behavioral shock tolerance served as predictors and psychophysiological measures of learning (sensitization, potentiation and differentiation scores) were the dependent measures. Interestingly, shock tolerance did not significantly relate to any of these learning measures. Sensitization Rated shock aversiveness was not significantly related to sensitization. Associative learning. In contrast, shock av e rsion systematically affected associative learning (see Figure 15) One group of subjects judged the shock as highly aversive (rating of 7 on the 1 7 scale; n= 27). Others, forming a Moderate Aversion group (n= 20), rated shock ave rsiveness equal to 6 Th e remaining subjects (n= 17) with a v er s ion ratin gs from 3 to 5, constituted th e Low A version group. SAM ra t ings of the shoc k w er e each highly correla ted with shock aversion (pleasure r= -.57; arous a l r = +.51 ) Therefore, rep o rt e d "aversion" can be decomposed into elements of displeasure and arousal. Stepwise regression indicated that greater shock aversiveness predicted significantly greater CS+ aversion, F(l,62)= 17.59 r = .47; startle potentiation F(l,62)= 13.98, r= 43; and skin conductance resp o nse potentiation, F(l,62 ) = 6.87, r= 32. Table 6 presents mean data from th e group analyses The group mean difference in rated CS+ aversion was significant F(2,61)= 7 84, p< 0 0 1. Follow up tests found that High and Low Aver sio n g roups significantly differed; these e x tr e m e groups both di ffe r e d m argina ll y from the Mod e r ate Av e rsi o n group (ps< 0 7 ) H oweve r the t h ree groups did no t s i g ni ficantly di ffe r in CSa ve r s i on r at i ngs

PAGE 76

69 Shock Aversion groups differed overall in s tartle potentiation, F(2,61)= 7 83, p< .001 (see Figure 16) In t-tests, High and Moderate Aversion groups did not differ (11s= + 120 vs. +82 ADUs) and both showed significantly more potentiation than Low Aversion subjects (11= -134 ADUs).5 The lack of startle potentiation in the Low Aversion group was not due to more of these subjects receiving delayed shock exposure (see Appendix D, Table 23). Regression analysis indicated that shock aversiveness was not significantly related to startle discrimination Subjects who rated shock as more aversive showed larger conductance response increases to CS+ from adaptation to acquisition (r= .32; see Figure 17). The Aversion group difference in conductance response potentiation was significant, F(2,61)= 3.59, p< .04. As with startle potentiation, electro dermal response potentiation for Moderate and High Aversion groups (11s= 04 and 08 S) did not differ and changes in both groups were significantly greater than the decreasing response in the Low Aversion group (11= -.07 Higher shock aversion was marginally related to better skin conductance response discrimination in acquisition, Aversion X Stimulus F(2,61)= 2.54, p< .09, but not during extinction. High Aversion subjects showed greater heart rate acceleration to CS+ in the final trial block of acquisition. However, the importance of this effect was diminished by a nearly complete reversal of this effect in the preceding trial block (see Appendix D, Table 24, for data and a brief discussion) 5 Although Figure 16 might suggest th at th e Low Aversion group had higher CS+ startle r espo nses than the other two gro ups before learning th e group difference was not significant (p= .12). Thus, a direct comparison of th e amount of potentiation across groups is p e rmi ss ible

PAGE 77

70 Table 6 Means and standard deviations for measures of associative learning as a function of shock aversiveness (see Figure 15 for correlations). 1 Shock Aversion Group Dependent Low Moderate High Mean Refer to Variable n=l7 n=20 n=27 n=64 Figure: CS+ aversiveness 4.5 b 5.5 ab 6.2 a 5.5 (1-7 rating scale; (1.5) (1.6) (1.0) (1.7) pos tlearning) CSaversiveness l.8 a 2.2 a 2.5 a 2.2 (1-7 rating scale; (1.1) (1.5) (1.7) (1.8) postlearning) Startle potentiation -134 b 82 a 120 a 41 16 (A-D units) 2 (273) (98) (235) (274) Skin conductance 07 b .04 a .08 a .02 17 potentiation ()3 ( 11) (.17) (. 21) ( 20) Change in CS+ -1.7 a -1.2 a -3 2 b -2.2 18 pleasure 4 (0.8) (1.5) (1.8) (1.7) Notes: 1. Means without common letters are significant different (t-tests; p< .05) 2 Change in CS+ response from adaptation block 2 to extinction block 1. 3. Change in CS+ response from adaptation block 2 to acquisition block 1. 4 Change in CS+ valence from preto post-learning fo r subject s v iewing pleasant slid es onl y.

PAGE 78

625 Low A version 5001 Moderate A version 550 I High A version 8 ,-.... 575~ \ 450 j II 500 VJ ..., 0 I ......... 525 ] \ J I I Q.) 0 0"O -, -o ::J ..., p / 450 -~ ro X J '\ J J I Q.) c= Q.) p:::: 475 j I I 3501 ....... 400 p:i I d 425-'-----------.------, 300-'--------r-----r--, 350-'-----r------, Adap2 Extl Adap2 Extl Adap2 Extl Figure 16. Startle potentiation (CS+ change from adaptation block 2 to extinction block 1) as a function of group differences in rated shock aversiveness. CSstartle change (also shown here) did not significantly differ between groups --.J ......

PAGE 79

I Low A version Moderate Aversion High A version 30 j CS+ U) ::i. -0CS.-1 + e: =00 25 0 (l) -0 ::l e" 2 00 ro (l) .20
PAGE 80

---2 C ;:J V) oO C ro p:::
  • Low A versiveness Moderate Aversiveness High Aversiveness 2 unpleasant pleasant unpleasant pleasant unpleasant pleasant 1 ----=-~= :;:::::;:::::: 0 I 20 I :::::=::::: } 1 11 ~"'-"I I 1 1 I 11 ----"""'-'t 2 11 11 1 -3 -l I I 11 [[] -4-L----------------1 Figure 18. Valence potentiation (change in CS+ pleasure ratings from preto postlearning; black bars) as a function of group differences in rated shock aversiveness (between panels) and Slide Valence (within each panel). Only subjects who rated their s h oc k US as highly aversive and viewed pleasant slides demonstrated substantial decreases in CS+ pleasure. In contrast, CS vale n c e c hange (striped bars) did not significantly differ between groups I --.J w

    PAGE 81

    74 The decrease in CS+ pleasure from preto postlearning (i.e., nega t ive change scores indicating valence potentiation) tended to be greater for High Shock Aversion (~= -1.4 units) than Moderate or Low Aversion subjects (both ~s= -0.4 units) in the overall sample, Aversion Group X Stimulus, F(2,61)= 2.88, p< .07. However, as seen in Figure 18, high shock aversiveness potentiated CS+ valence change only for subjects viewing pleasant slides, Slide Valence X Shock Aversion F(2,58)= 3.18, p< .05. Post hoc tests showed that, for unpleasant slide foregrounds, shock aversion did not systematically modulate CS+ valence change. But among subjects viewing pleasant slides, subjects highly averse to shock judged their CS+ as more unpleasant after learning, F(2,29)= 5.73, p< .01. Figure 18 also illustrates, in comparison, that CSchange was not modulated by rated shock aversion. Shock aversiveness did not interact with affective slide content in determining psychophysiological response during associative learning Frequency judgments of shock occurr ence. Subjects underestimated the number of shocks they received during conditioning (i.e., excluding the shock exposure procedure). Although each subject experienced 8 shocks, the mean estimate was 6.5 times. This discrepancy from actual exposure was significant, F(l,61)= 29.22, p< .0000. Although no individual difference factors were reliably associated with shock frequency estimates, subjects reporting less aversion of shock tended to under-report frequency of shock occurrence (r= .23, p< .10) Frequency judgments of shock occurrence were not influenced by the affective slide foreground during conditioning. Contingency recall. E ighty-six percent of sub j ects clearly recalled their CS + when probed by the followi ng qu estions: (1) Did you recognize a rule in this experiment?; (2) W e re some stimuli followed by shock more often than others?; and (3) If ye s, which? An addition a l four sub je cts did not clearly

    PAGE 82

    75 understand the probe questions but, when clarified, correctly identified their CS+. (It is not clear whether the experimenter inadvertently cued the subject in these cases; thus, it is not certain whether correct identificat i on was a recall or a recognition decision. If these four subjects are included in the correct recall category, the recall rate becomes 92 % .) Four subjects could not report which slide had been shocked and a fifth subject incorrectly believed that CShad been shocked. Interestingly, four of these five subjects were in the Low Shock Aversion group (the other was in the High Aversion group). However, due to the rather small subsample, these data were not subjected to statistical analysis. Contingency recall was not related to affective slide content. Re analysis of the data set, excluding subjects who could not recall the contingency, did not change the findings. Summary The correlational pattern shown in Figure 15 supports several major conclusions. First, temperamental fear and inhibition modulated sensitivity to aversive stimulation, but did not generally determine aversive learning Second, the perception of shock aversiveness--and not behavioral tolerance of shock intensity--predicted the degree of associative learning. Third, these relationships were present only for measures of simple (but not differential) association. Finally, whereas rated shock aversiveness directly influenced physiological measures of learning, CS+ aversiveness showed similar, albeit weaker, relationships.

    PAGE 83

    DISCUSSION The present results indicate that startle reflex magnitude was sensitized after exposure to shock and was also potentiated during a stimulus which was paired with shock (relative to a stimulus not paired). In short, startle magnitude is a reliable index of both nonassociative and associative learning when an aversive reinforcer is used. Whereas startle onset latency and visceral (heart rate and electrodermal) reactions also reflected associative learning, these measures failed to index the sensitization learning process which is presumed to occur when subjects are exposed to shock. Consistent with the startle reflex data, ratings of affective valence (but not arousal) for the conditioned stimuli reflected overall associative learning. In addition, individual differences in aversive reactivity and learning were identified. First, subjects who were both temperamentally fearful and inhibited were more sensitive to aversive stimulation (i.e., they showed less tolerance of shock, larger initial startle responses, and reported more anxiety across a variety of situations) than other subjects However, temperament characteristics were not generally related to learning measures. Second, perceived aversiveness of the shock US--and not physical intensity of shockmodulated the degree of learning in this experiment. Sensitization Startle Magnitude In this experiment, a brief series of shocks was administered to each subject to calibrate US intensity just below the subject s reported pain threshold level. Shock exposure prompted an immediate, vigorou s (40 %) 76

    PAGE 84

    77 increase in startle magnitude on the first postexposure trial, whereas this effect was absent in control conditions. Thus, significant startle sensitization developed very rapidly and could not be attributed to either dishabituation or the simple passage of time. In addition, the increase in startle magnitude from preto postexposure in the Immediate group--for whom duration of this sensitization effect could be more completely measured--persisted throughout adaptation (ca. 10 mins). The present findings are consistent with the animal literature on shock sensitization of startle (Boulis & Davis, 1989; Davis, 1989; Hitchcock et al., 1989). To the extent that similar results were obtained in recent experiments using human subjects (Greenwald et al., 1990; Hamm, Stark, & Vaitl, 1990), it may be concluded here that shock sensitization of the startle response is a robust phenomenon (i.e., large magnitude, rapid development, relatively persistent and independent of species, dishabituation, and time) In contrast, neither startle latency nor visceral activation (i.e., tonic levels or slide responses) was reliably facilitated after shock exposure Taken together, the data suggest that shock exposure prompts an aversive preparatory set--uniquely indexed by the startle reflex (Lang et al., 1990)which, lacking visceral arousal, may not require abrupt mobilization for action. Consistent with this idea, freezing has been shown to emerge as the dominant tactical response following shock exposure in animals (Fanselow 1980; Fanselow & Bolles 1979); furthermore the extent of motoric inhibition correlates highly positivel y with increased startle re s ponse in individu a l s ubjects (Leaton & Bor sz c z, 1 9 85) Shock ex po s ur e c an a lso influenc e b e h av ior in fu tu re ave r s i ve s ituations For i n s t a n ce, pre-ex posin g a nim a l s to a bri e f se ri es o f relative l y i nt e n se s h ocks facili t a t es fear" o f s ub seq u e nt aversive stimu l ation as mea s ured by s u ppr essio n of operan t r a t es (Pearl

    PAGE 85

    78 1963; Pearl, Walters & Anderson, 1964) and avoidance in a conflict te st (Kurtz & Walters, 1962 ; Walters, 1963). Current evidence (Lang et al., 1992) suggests that startle augmentation reflects the motivational state of the subject, specifically the presence of an aversive response disposition. The selective effect of shock exposure on sensitization of startle magnitude (i.e., in the absence of visceral activation) obtained here, as well as the varied effects of shock exposure on behavior just described, encourages the conclusion that the startle response is not tactically specific (i.e associated with a particular behavioral act) but, rather a more general measure of aversive processing Methodological Considerations One question which cannot be answered from this experiment is whether actual exposur e to shock is necessary to obtain performance changes, since exposure is confounded here with the knowledge that shoc k will be presented Thu s, it remains poss i ble that in s tru c t e d e xposur e c an produce sensitization Several researchers ha v e in fact examined react io ns to threat of shock. Chattopadhyay, Cooke Toone, and Lader (1980 ; Exp 1) conducted a study in which subjects were threatened or not threatened with shock (within-subject sessions separated by 1 week; order counterbalanced across subjects) They found that electrodermal levels and anxiety ratings increased during threat sessions, relative to no threat. Unfortunately the authors did not report whether responses increased relative to resting baselines within the threat session nor wheth e r th ey conduct e d a b e tw ee n g roup anal ys i s o f th e fir s t s e s s ion Thu s, it is no t cl ea r w h e th e r w ithi n sess ion ch a n ges were sig nific a nt o r wh e th er be t ween-sess i o n (i .e., threa t or d e r ) effec t s in fl u enced th e res u l t s. Far h a and S h er (1989) a l so showed that autonomic measures, somatic activ i ty, and reports of anxiety all increase

    PAGE 86

    79 after shock threat (relative to a "naive" baseline ). Unlike the present study, their experiment did not include a control group. Therefore, neither dishabituation (i e ., after the experimenter re-entered the testing room) nor the sheer passage of time can be ruled out as potential explanations of their findings. Unlike the studies above, shock exposure did not sensitize either cardiac or electrodermal activity here. There are several possible explanations for this discrepancy. First, it could be argued that the test of shock sensitization of visceral activation was not powerful: Pre-exposure tonic levels and slide responses were not available for the majority of subjects (i.e., Immediate exposure group), yielding a weaker test in a smaller subset of subjects (i.e., Delayed exposure group) On the other hand, although conductance levels and slide responses slightly increased (nonsignificantly) after shock exposure in the Delayed group, levels and responses in the control condition (i.e., Immediate group) also increased. These data suggest that the increa ses in the Delayed group were not specific to shock exposure and could be accounted for by another factor (e.g., dishabituation) A second explanation is that exp osure to shock and thr eat of shock may differentially affect visceral activity. During shock threat, the subject might anticipate aversive stimulation and thu s maintain visceral activity at higher levels. During sho ck exposure, however, visceral activity may temporarily increase (before it could be measured here?) 1 and then rapidly habituate. The present heart 1 There was cl early an unconditioned reaction to shock pres e nt atio n Howeve r, the i ssue h ere i s whether shock exposure generated a fleeting increase in visceral tonic level before th e experime nter could continue the proce dure to mea sure suc h a change. In fact, th ere was a period of about 3 to 5 minutes after se ttin g the final shock intensity when the experimenter had to instruct the subject on the subsequent conditioning procedure, and initiate

    PAGE 87

    80 rate tonic level data are partially consistent with this interpretation, since levels of the Delayed group tended to increase after exposure, whereas control levels significantly habituated. A third, equally plausible, explanation is that the results of other studies may arise from uncontrolled influences discussed above. In future work, one would ideally separate the incremental effects of threat and exposure, using a paradigm similar to that of Farha and Sher (1989), while including a control group and continuous monitoring of visceral activation. Finally, it is interesting that during the initial control period for shock exposure (Delayed group) there was a mild, but nonsignificant, increase in startle magnitude. This increase is not inconsistent with previous findings: Some habituation experiments (Groves & Thompson, 1970; Szabo & Kolta, 1967) have shown that the startle reflex elicited by an intense acoustic probe (in the absence of shock) may briefly increase before habituation occurs Thus, it is possible that reflex sensitization in the Immediate exposure group reflects summation from two events--a strong, persistent increase from shock exposure and a weaker, probe-related increase. One suggestion for future research would be to repeat the present study design, adding a longer preshock series of startle probes to induce certain response habituation. This would insure that the amount of sensitization following shock exposure could not be attributed to characteristics of the probe itself. Associative Learning Magnitude of the eyeblink reflex to acoustic probes presented during the CS+ slide was larger at the start of extinction, relative to adaptation. Startle the computer program Thus, if visceral changes did occur, they could have subsided by the time the procedure continued.

    PAGE 88

    81 reflexes evoked during CS+ were also significantly larger and faster in onset than those during CSor interslide intervals in extinction. Additional performance changes--including skin conductance response (potentiation and discrimination), differential cardiac acceleration, and affective ratings change--provide converging support for the conclusion that aversive learning occurred. The present findings thus replicate and extend results from single cue (i.e., nondiscriminative) conditioning experiments with animals (Berg & Davis, 1985; Brown, Kalish, & Farber, 1951; Cassella & Davis, 1986; Davis & Astrachan, 1978; Kurtz & Siegel, 1966; Leaton & Borszcz, 1985) and recent differential learning experiments with humans (Greenwald et al., 1990; Hamm et al., 1991; Hamm, Stark, & Vaitl, 1990). Sensitization and Habituation in Associative Learning The differential learning paradigm was used in this experiment because it controls for nonassociative influences such as (within-session) habituation and sensitization. Since the assessment of associative learning involves comparing CS+ and CSresponses, it is important to rule out nonassociative factors as alternative explanations. Sensitization, habituation. and potentiation. CS+ startle magnitude was potentiated early in extinction (relative to adaptation), whereas interslide responses decreased over this same period. Recent experiments (Greenwald et al., 1990; Hamm et al., 1991) also demonstrated this same pattern. If both CS+ and interslide responses had increased this would have suggested that a nonassociative factor influenced potentiation. Thus, the data indicate that th e CS+ increase was an associative effect (since shock was never paired with th e abse n ce of slides). By si milar logic, sensitization canno t account for CS+ skin conduc tance r espo n se potentiation, since CSresponse decrea se d from adaptation to acqu i si ti on A l so, se n s iti zation cannot accoun t for heart rate

    PAGE 89

    82 accelerative changes during training, since CS+ and CSresponses developed slowly and in opposing directions. For subjects whose exposure to shock preceded adaptation in this study (as in past experiments), startle probe responses habituated before learning and there was a clear replication of the startle potentiation effect. In contrast, for the subgroup whose exposure to shock was delayed, startle potentiation was not only absent but significantly decreased from adaptation to extinction. These data suggest that habituation of startle responses in the Delayed group prior to acquisition was not as advanced which, in turn, may have masked startle potentiation. It is therefore possible that the absence of startle potentiation in the Delayed exposure group may be a methodological artifact resulting from insufficient habituation. A similar pattern was observed with skin conductance response potentiation, i.e., initial pairing of the CS+ and shock produced increased conductance response in the Immediate group, but CS+ response in the Delayed group decreased--even though the two groups slide responses prior to shock pairing did not differ. An obvious remedy for this artifact (if one were to repeat this procedure) would be to extend the adaptation series, thus insuring response habituation in the Delayed group. Habituation and discrimination. The psychophysiological measures of learning employed here--especially startle magnitude and skin conductance response--clearly habituated during the procedure, with the exception of transient potentiation due to the shock pairing Yet there was significant discrimination between CS+ and CSin extinction for startle (magnitude and onset latency), skin conductance, and late cardiac decelerative responses and during a c qui s ition for heart ra t e acc e leration and el e ctrod e rmal respon ses. The se data suggest that th e assoc iati ve effect during a nd a ft e r learning w as specific t o CS+ Furth e rm o r e, CS + a nd C Sr es pon ses di d n o t e x tingui s h a t

    PAGE 90

    83 different rates, i.e., response decline was parallel for the CS+ and CSslides, to the extent that Stimulus X Trial Block interactions were not obtained for these measures. Thus, discrimination learning is assessed here with habituation as a constant, background factor. Bradley, Cuthbert, and Lang (1990a) demonstrated that the startle reflex habituates with repeated probe presentations within a session, a result which presumably reflects decremental activation in the obligatory startle circuit. They also found that the affect-startle effect was present throughout this general habituation, attesting to persistent activity in the modulatory circuit. In the present experiment and other recent human conditioning studies, CS+ /CSresponse discrimination throughout extinction similarly reflects the continued influence of the modulatory system--as induced by associative learning The pattern of cardiac acceleration was different from startle reflex and electrodermal measures during associative learning In acquisition, heart rate acceleration during the CS+ slide increased with repeated pairings whereas acceleration to CSdecreased with repeated norueinforcement. This cardiac pattern suggests that aversive learning may influence the heart rhythm in such a way as to reflect both excitatory and inhibitory influences However, slightly different results were reported by Hamm et al. (1991), who found that cardiac response was more acceleratory overall that CSresponse became more accelerative over trials with CS+ and that the size of differential acceleration was slightly larger than in the present study. At first g lanc e, these difference s suggest that Hamm et al. produc e d a more powerful m a ni p ul a tion (e.g., a m ore int e n se s ho c k ); ho weve r th e size o f th e s tartl e reflex an d ski n co ndu c t a n ce l ea rnin g e ff e ct s a r e v irtu a ll y i d e nti ca l in th e t wo s t udies. Th e r efo r e, it does n o t see m w a rr a nt ed t o concl u de th a t th e pr ese nt

    PAGE 91

    84 data reflect a less effective manipulation. A more likely explanation is that cardiac acceleration is simply less reliable than other measures of slide response (Lang, Greenwald, Bradley, Hamm, in press). Thus specific aspects of the cardiac pattern may be less likely to replicate. Finally, there was evidence that (1) affective ratings "habituate" (regress towards neutrality) with stimulus repetition and that (2) CS+ valence and arousal judgments generalized to CS-. On the first point, recall that adjunct slides varying in affective content were rated preand postlearning, but were not subjected to conditioning. Mean valence judgments of the pleasant and unpleasant control slides became more neutral, whereas initially neutral slides showed no mean change. Similarly, arousal ratings of adjunct slides became moderate (i.e., calm slides became more exciting and vice versa). However, changes in CS+ valence ratings (particularly for Positive groups) and CS+ arousal ratings (especially for Calm slide groups) were greater than changes in ratings for control s lides suggesting that learning played a greater role in affective change than regression to the mean On the second point, Figure 5 clearly shows that affective ratings of CS+ generalized to CS-. This was more the case for arousal than valence ratings: Rated arousal, similar for CS+ and CS-, was the single performance measure which did not reflect aversive learning In contrast, both discrimination and potentiation of valence ratings were obtained Although it is not clear why arousal ratings did not differentiate the stimuli h e re--Hamm et al. (1991) and Greenwald et al. (1990) found significant ratings changes along both affective dimensions--one h y poth es is i s that both earli e r s tudies used n e utral /ca l m CS s l i d es whi c h y i e ld e d th e g r ea t es t aro u sa l change of any affec ti ve slide t ype T h ese n e ut ra l CS s li des because of th e se l ection of ex t re m e affective sti muli h e r e, were l arge l y e x cl u ded.

    PAGE 92

    85 Affective Associative Change Results of this experiment strongly support the main hypothesis that aversive learning modifies the principal motivational (valence, arousal) parameters of emotional response. Startle reflex and electrodermal changes (both potentiation and discrimination), cardiac acceleration, and affective ratings change were all significantly modulated by the classical conditioning procedure. The results do not, however, support the secondary hypothesis that the affective quality of the slide foreground prior to learning constrains the degree of associative change. More specifically, the physiological measures used here do not supply information on the extent of affective associative modulation. Affective modulation by specific content. As noted before, Hamm et al. (1991) found that the amount of startle potentiation (and discrimination) varied linearly with a priori valence of slide content: Less modulation was obs erved for subjects who viewed unpleasant slide contents (mutilations threatening animals) than for subjects who viewed pleasant contents (nature scenes, erotica). In addition, electrodermal discrimination varied linearly with a priori slide arousal : Greater discrimination was observed for subjects who viewed calm slides (household objects) than for subjects who viewed the most arousing slides (mutilations, erotica). Greenwald et al. (1990) attempted to replicate these associative effects by assigning subjects to view one of six slide content pairs distributed in the valence X arousal space. The results of that experiment showed two changes in the original startle modulation effect: (1) Discrimination, more so than potentiation, tracked th e amount of va l ence chang e; and (2) Rather than a difference in magnitud e, a sma ll e r proportion of subjects who viewed unpleasant s lid es s how ed t ar tl e discrimination th an th ose viewing plea ant

    PAGE 93

    86 slides. Electrodermal discrimination was also not a linear function of slide arousal. Such differences across studies do not strongly support the affective modulation hypothesis or, perhaps, suggest that other factors (e.g., specific content) could mediate these effects. In contrast, changes in affective ratings here supported findings of both earlier studies, namely that (1) CS+ valence potentiation was greater for subjects viewing pleasant (than unpleasant) slides and (2) CS+ arousal potentiation was greater for subjects viewing calm (than arousing) slides. One problem with these differential ratings changes, however, is that they may arise from floor and ceiling effects (i.e., initially unpleasant slides cannot be rated much more unpleasant, and initially arousing slides cannot be rated much more arousing, after aversive learning). Modulation along affective dimensions. In the present experiment, stimulus content was deliberately obscured by choosing slides based on each subject's extreme ratings. Inspection of Appendix A indicates that many slide contents defined each of the four affective categories Thus, it appears that the effects obtained by Hamm et al. (1991) and Greenwald et al. (1990) may have partly depended on specific slide contents. The present approach to stimulus selection is more consistent with the dimensional view of emotion, rather than relying on specific slide content to influence the degree of associative change. Even with this more sensitive approach, the results did not clearly support the affective change hypothesis. On the other hand, subjects who viewed unpleasant/arousing slides tended to show less differential learning than the other groups--not only in startle magnitude and onset latenc y, b ut for electrodermal respon s e valence and arou s al rating s (Appendi x 0). Thu s, a weaker v er s ion of th e h y pothe s i s propo se d by Hamm et al. ( 1 99 1 ) c o uld be advanc e d in whi c h a CS+ who se

    PAGE 94

    87 affective character prior to learning most closely resembles that of shock (i.e., highly unpleasant and arousing) will undergo marginal associative change, relative to any other affectively dissimilar stimulus. This notion--which is a variant of the "delta rule" (Rurnelhart et al., 1986)--has been applied to the problem of affective change in conditioning elsewhere (e.g., Rescorla & Wagner, 1972) In the present case, this hypothesis predicts a Stimulus X Valence X Arousal effect (for discrimination scores) or a Valence X Arousal X Trial Block effect (for potentiation scores). Despite the consistency of mean differences across measures, however, no statistical test yielded a significant interaction. Temporal Specificity and Preparatory Processes Startle response. Several studies have demonstrated that associative learning is temporally specific, since startle potentiation is largest nearest the CS/US training interval (Davis Schlesinger, & Sorenson, 1989; Siegel, 1967) or when shock is expected to occur (Grillon, Ameli Merikangas, Woods, & Davis, in press). Grillon et al. signaled subjects to expect shock in one ("threat") period but not in another ("safe") period. In addition, subjects were told that shock would only occur during the last 10 sec of the threat period Grillon et al. found that startle was significantly potentiated during threat (relative to safe) periods, and was maximal during the last 10 sec of the threat period. Interestingly, this effect was obtained before shock was ever administered and improved over trials In the present study, startle magnitudes generally (across CS+ and CS-) increased lin ea rly over the CS/US interval in extinction Thus, temporal sensitivi ty of probe r espo ns e was demonstrated h ere. However, the l ack of diff ere ntiation between CS+ and CSalong thi s temporal grad i en t indi ca t es that s timulus generalization a l so occurred. A similar, but non-significant,

    PAGE 95

    8 8 increase in probe response was found by Hamm et al. (1991) One reason that a significant finding may have been obtained here but not in the previous experiment is that every presentation of a slide was probed here, whereas Hamm et al. probed only 3 of every 4 CS occurrences. Thus, the difference across studies may be partly due to statistical power Visceral reactions. Electrodermal and heart rate responses obtained in this experiment offer a clearer example of learned temporal and stimulus specificity. In acquisition, the SIR conductance response (preceding slide offset) was greater to CS+ than CS-, presumably indexing greater orienting to shock onset. This differentiation developed linearly over trials and persisted into extinction. Although overall SIR magnitude decreased in the absence of reinforcement, this anticipatory discrimination remained constant. Also, late heart rate deceleration in extinction (D2 leg; occurring at the same time during slide presentation as the SIR response) was significantly greater during CS+ than CS. Late cardiac deceleration presumably reflects anticipation of events at stimulus offset in an Sl-S2 paradigm (Bohlin & Kjellberg, 1979) Thus, skin conductance and heart rate response provide convergent evidence for temporally-specific stimulus differentiation. Exactly what subjects are preparing to do during the training phase of aversive classical conditioning has been widely debated The midinterval acceleratory and late-interval deceleratory components of the cardiac waveform have been examined for their relationship to psychological processes Enhanced cardiac acceleration to an aversive cue has been proposed to measure recruitment of a defense response (Fowles, 1980 ; Gray 1 9 82 ; Hod e s e t al. 1985 ; Ob r i s t 1 98 1 ), w h e reas g r e at e r h e art r a t e dec e l e r a ti o n ju s t pri o r to US d e li v ery h as b e en th o u g ht to in d e x a n att e nti o n a l se t (Deane 1969; Lacey & Lacey, 1 970) To t he extent th a t cardiac d i scrimination in

    PAGE 96

    89 acquisition develops in the acceleratory component, this suggests that temporary activation of a defense response may occur during learning. In contrast, cardiac discrimination in the late decelerative leg during extinction suggests that the information processing set shifts from a defensive (stimulus rejection) posture to a more attentive (stimulus intake) mode with cessation of aversive reinforcement. Patrick and Berthot (1991) recently studied the effect of active vs. passive preparation on subjects' psychophysiological responses in anticipation of aversive noise. They found that only late cardiac deceleration varied by preparatory set, with greater deceleration in the "active" group (who pressed a microswitch to escape the noise) than either a yoked "passive" group or a control group anticipating a less intense tone. In the active group, the extent of cardiac deceleration was significantly correlated with the speed of escape, which supports the idea that deceleration reflects sensory enhancement (cf. Lacey, Kagan, Lacey, & Moss 1963). These data are consistent with the results of an experiment by Chase, Graham, and Graham (1968), who found that cardiac acceleration was dominant only when subjects anticipated a vigorous motor response (i.e., a leg lift), whereas cardiac deceleration was the clear pattern when subjects anticipated a simple reaction time response. In contrast to the active-passive group difference they found for cardiac deceleration, Patrick and Berthot reported that startle magnitude during the preparatory period was equal for the two groups, increasing closer to the aversive US Qust as in the present study). In contrast, startle responses of the control group were smaller and decreased progre ssive ly closer to the weaker US which that group received Thus, the startle probe reflex did not index the specific tactical response whic h supports the hypothesis that it is a more general strategic indicator of aversive motivation (Lang et al. 1990, 1992)

    PAGE 97

    90 Brain-Behavior Relationships in Aversive Learning The present analysis has focused on psychophysiological manifestations of emotional learning and used these performance measures to infer cognitive processing in aversive behavior. Yet this study also relates to an expansive literature on the neural substrates of aversive learning. Thus, although the present discussion is framed at a representational level of analysis (Figure 1), the cognitive events described here are presumed to parallel events at an implementational level (cf. Marr, 1982). Ultimately, one goal of this work is to develop a psychological model of emotional learning which is (at the very least) plausible or (at its very best) isomorphic with neural events Davis (1989a) has outlined a neurobehavioral model of the aversive system using both fear-potentiated startle and sensitization paradigms. The amygdaloid complex, especially the central nucleus, appears to be the critical anatomical node involved in aversive learning and memory (Kapp, Pascoe & Bixler 1984; LeDoux, 1987 ; Sarter & Markowitsch, 1985). Lesions of the central nucleus abolish both startle sensitization and potentiation (Hitchcock & Davis, 1986; Hitchcock et al., 1989) Furthermore, drugs which have anxiety-reducing properties in humans block learned startle potentiation (e.g., Berg & Davis, 1984), whereas anxiogenic agents increase the learning effect (e.g., Davis, Redmond, & Baraban, 1979). Behavioral signs of aversion are presumed to occur when the amygdala central nucleus is activated and projects to a wide array of brain areas When the shock US is presented--as in sensitiza tion --i t activates the amygdaloid complex, which projects to severa l efferent nodes including the central grey (leading to inhibition of behavior), th e dorsal hypothalamus (cardiovascular changes), the parabrachial nucleus (respiratory adjustments), and--in the case

    PAGE 98

    91 of startle reflex excitability--the nucleus reticularis pontis caudalis (Davis, 1989a). Electrical stimulation of the amygdaloid central nucleus elicits these unconditioned behaviors (Applegate, Kapp, Underwood, & McNall, 1983; Davis et al., 1982; Gentile, Jarrell, Teich, McCabe, & Schneiderman, 1986; Harper, Frysinger, Trelease, & Marks, 1984; Rosen & Davis, 1988, 1990). The amygdala can also act as an associative node for the conjunction of the CS+ and US. In classical conditioning, the result of pairing a visual CS+ with the shock US is that the amygdala can then be independently activated by the CS+, via relatively direct subcortical sensory pathways, leading to potentiated startle (Tischler & Davis, 1983). In short, the flow of events within this neuroanatomical circuitry is remarkably consistent with the Konorskian cognitive mechanisms of learning adopted here. Individual Differences Aversive Reactivity and Sensitization E ach subject was pre-e x posed to a series of shocks, the criterion level being judged as very intense but below his or her nocioceptive threshold. Subject temperament characteristics modulated tolerance of shock during this procedure Subjects who were relatively more fearful and inhibited were less able to tolerate increments in shock intensity than other subgroups (especially nonfearful, impulsive) defined by the fear X impulsivity temperamental space These trait fearful, inhibited subjects also reacted more s trongly to the startle probe before learning and reported more specific fear s ac ross a variety of situations than trait nonfearful impulsive subje cts. Thu s, th e pr es ent s tudy id e ntified subj e cts who ar e ge n e rall y more sen s it ive t o aversive sti mul a ti on. In the Cha t topad h yay et al. ( 1 980; E p. 1 ) s t udy noted earl i er, a ud i t ory evok e d responses t o q u asi -s tar tl e probes (10 m sec rise time) we r e m easured

    PAGE 99

    92 during threat of shock and its absence. Not unlike the larger probe startle responses observed here for fearful/inhibited subjects, they found that the P 300 component of the auditory evoked cortical response was greater in high trait anxious than low trait anxious subjects, independent of shock threat. There may be some communality among those findings and the present data, since other investigators (Roth, Blowers, Doyle, & Kopell, 1982; Roth, Dorato, & Kopell, 1984) have found that startle blink amplitude and P300 amplitude covary reasonably well across a range of probe intensities. More work is needed to determine whether temperamental fear potentiates both the reflex eyeblink and evoked cortical potential to acoustic startle probes. This would constitute a test of the hypothesis (Lang et al., 1992) that emotional response sets are integrated efferent patterns extending from "cortex to reflex." It is possible that nonfearful/impulsive subjects, who were less affected by the aversive stimuli (i.e., shock, startle probe) presented here, may be part of a "disinhibited" subpopulation that is generally more tolerant of novel or intense stimulation (Barratt, 1983; Zuckerman, Buchsbaum, & Murphy, 1980). In fact, Bradley (1992) found that the EASI lmpulsivity dimension and Disinhibition subscale of "sensation seeking" (Zuckerman, 1979) both loaded on a global "behavioral constraint" factor (cf. Tellegen, 1985) in a factor analysis of personality questionnaire data. Thus, it is worth noting that the present data agree with a recent report (Pivik, Stelmack, & Bylsma, 1988) that subjects scoring high on the Zuckerman Disinhibition subscale exh~bited reduced mo to neuron excitability (i e., electrically elicited Hoffman reflexes) Naturally, one should be cautious in drawing conclusions from such divers e r e fl ex measures, but these findin gs sugg e st that th e inhibition / impul s ivit y construct may relate to ba s i c di ffe rences in the organization of efferent

    PAGE 100

    93 output. Future research can determine whether these temperament group differences are reliable, whether they apply to both aversive and appetitive USs, and which mechanisms control these putative behavioral differences. Overall, 77% percent of subjects in this study showed at least some increase in startle magnitude following shock exposure. However, neither individual differences in subjects' temperament nor shock tolerance were significantly associated with the degree of startle sensitization. Since Greenwald et al. (1990) also found that most subjects (i e., 95%) showed shock sensitization of startle magnitude, startle sensitization in humans may be a fairly robust effect. One reason that slightly more subjects in the earlier study showed startle sensitization may be that many more preshock startle probes were administered (as part of a preceding experiment), presumably leading to greater habituation and a lower prelearning baseline for assessing startle sensitization. As argued above, this factor would enhance the effect across subjects Associative Learning Temperament and tolerance of electric shock Individual differences in temperament and shock behavioral tolerance did not generally influence associative learning Because the fearful/inhibited group attenuated their exposure to shock initially, it is difficult to determine whether these same subjects should learn better with an aversive reinforcer (as Eysenckian theory might predict). More extensive pre-exposure to the US is known to retard classical conditioning (Randich & LoLordo, 1979) Therefore, it would be expected that nonfearful / impulsive subjects--whose greater tolerance for shock resu lted in more pre-exposure in this s tud y--might ac tuall y show poorer asso ciativ e l earni n g. Yet impulsive subjects pe r forme d better in ski n conductance ( i .e., greate r poten tiation and discrimination) Thu s, n ei th er

    PAGE 101

    94 amount nor intensity of US pre-exposure can be used to explain the general lack of temperament differences in learning Alternatively, the exposure procedure may have equalized perceived aversiveness of shock across subjects (who reported it to be "highly annoying but not painful") before learning. Although more fearful subjects in this sample tended to rate the shock as slightly more aversive after learning, this relationship was not significant. It remains possible that any potential learning effects might be indirectly mediated by temperament characteristics from shock exposure. If subjects were to be exposed to the same shock level (which is not practical due to ethical concerns, unless US intensity is quite low), temperament might influence learning but the overall magnitude of learning would diminish. Thus, the present procedure involves a tradeoff which limits our ability to study temperament group differences in learning. However, in future research, such differences could be assessed by using an aversive reinforcer which is constant across subjects (e g ., intense noise) Rated shock aversiveness. In contrast to individual differences in temperament and tolerance for shock rated shock aversiveness clearly influenced the degree of learning. Since this estimate was obtained aft e r the experiment, the rating reflects the extent to which the US maintained its aversiveness The startle reflex, skin conductance response, and CS+ aversiveness were all augmented as subjects' ratings of US aversiveness increased Because shock aversiveness directly modulated measures of simple association (startle reflex and electrodermal potentiation, CS+ aversion) but not discrimination or rat e of e x tinction, the results suggest that US aver s ion mo dulated th e i ncrease in CS + associabilit y r at h e r th a n th e pe r s i s t ence ( or l oss ) of as s ociabilit y. T his r ela ti o n s hi p was also s timul us s p e cifi c: S ho ck ave r s i ve n ess did n o t i nflue n ce r e s p ondin g t o th e CS s li de

    PAGE 102

    95 Since responses from multiple measurement domains (i.e., verbal report, reflex behavior, and visceral physiology) were systematically influenced by variations in shock aversiveness, this association is probably due to the operation of a central motivational set. The cognitive model of learning advanced at the outset anticipates such an effect, because perception is assumed to organize representations of both conditioned and unconditioned stimuli and their connections with relevant motivational systems. The coherence among psychophysiological indices and aversion which was observed in this study highlights the structural role of perception in emotion and is congruent with prior experiments which have found concordance of affective judgments with measures of facial action and visceral activation (Greenwald et al., 1989; Lang et al., in press). Furthermore, the present data suggest that it is the subject's interpretation, rather than raw physical intensity, of shock that modulates the strength of association between CS+ and US representations. Although subjects who rated the shock as less aversive showed weaker simple associative (potentiation) effects, they did not show statistically different levels of discriminative conditioning in either startle reflex and electrodermal response. Thus it cannot be argued that these subjects were poor learners or generally less affected by shock. Instead, it seems that what they learned may have differed from that of subjects who felt that the shock was more aversive. Such a distinction was raised by Hodes, Cook and Lang (1985) in their analysis of indi vid ual differences in aversive learning The y found that subjects who demonstrated cardiac acceleration to CS+ during the final half of acquisition w ere more resistant to electroderrnal extinction and rated the ir CS+ as more un p l easan t and arousing, relative to subjects whose response was dec e lerativ e. All th eir s ubject groups howed physiological

    PAGE 103

    96 signs (i.e., anticipatory cardiac deceleration and skin conductance responses ) that they learned the contingency. Thus, Hodes et al. interpreted their findings to mean that heart rate "accelerators" differed from "decelerators" in having learned a fear response. In this study, the fact that all subjects (i.e ., regardless of shock aversion) showed discriminative performance suggests they learned the basic association. However, since potentiation scoreswhich reflect only the CS+ /US relationship--were modulated by shock aversion, this suggests that subjects who loathed shock may have, like the "accelerators" of Hodes et al. (1985), learned a fear-like response. Unlike the findings of Hodes et al. (who pooled data across four experiments), the heart rate data from this single experiment did not produce reliable results in the final part of acquisition. Summary The present psychophysiological analysis of shock-induced sensitization and classical conditioning re vea led that startle magnitud e i s a uniquely sensitive measure of both non ass ociative and associ a ti v e forms of aversi ve learning Startle onset latency slide visceral reactions and affective rating s of the excitatory stimulus (CS+) reflected associative learning. Finally, individual differences in the d e sire to avoid the shock modulated the degr ee of learning These results support a cognitive mod e l of l e arning in which a representation of the US is connected with the brain s a v ersive motivation a l system. Presentations of the US sensitize this system to impending aversiv e stimulation When a CS+ is ex plicitly and repeatedl y p air e d with the US i t s mnemonic image connect s w ith the US r e present a ti o n a nd the moti v ati o n a l sys t e m T h ese associa ti ve l inks me di a t e s ub se qu ent excitatory resp on ses to p resentations of th e CS+ cue, espec i a ll y in s ubj ec t s who perceive th e US as h ig hl y ave r s i ve. A lth o u g h avers i ve reac ti o n s w ere generally larger a m ong

    PAGE 104

    97 temperamentally fearful and inhibited subjects in this study (i.e., behavioral avoidance of electric shock, larger startle probe responses, and more reported situational anxiety), temperament characteristics were not systematically related to the degree of aversive learning

    PAGE 105

    Slide (IAPS#) 112 128 130 175 192 207 236 253 254 273 300 317 323 466 468 469 520 601 602 623 733 736 803 808 818 820 849 900 929 941 942 944 APPENDIX A STIMULUS SELECTION Listing of Slide Numbers and Stimulus Content Stimulus Content attacking snake oily-looking rat pit bull snarling two brown rabbits dolphin happy baby happy family, domestic scene healthy elderly couple riding bicycles mother holding child boy licking cow's anus mutilated face baby with tumor over eye hospital patient on respirator heterosexual couple kissing nude couple in water nude couple on bed purple flowers black man in jail electric chair gun aimed at viewer ice cream with chocolate sauce pumpkin pie with flies all over it ski jumper man counterbalancing sailboat cliff divers water skier roller coaster cemetery trash heap soldier carrying wounded boy wounded soldier skulls lined up in a row

    PAGE 106

    99 Normative Affective Ratings of Slide Contents by Category Calm Slides Arousing Slides Affective Category Slide Valence Arousal Slide Valence Arousal Pleasant 520 7.4 3.2 803 7.3 7.3 236 7.9 3.4 849 7.5 7.2 253 7 8 4.0 468 7.1 6 9 254 7.6 4.0 818 6.7 6.9 175 8.3 4.1 808 7.7 6.7 192 8.4 4.9 466 7.4 6.6 207 8.4 4.9 469 6.8 6.5 733 7.7 5.3 820 7.5 6.3 ---------------------------------------------------------------------------------Neutral 701 4.9 2 7 709 4 7 2.7 710 5.2 2.9 713 4.8 3.4 -------------------------------------------------------------------------------------------------Unpleasant 900 2 6 4 1 300 1.7 7.3 601 3 3 4.2 623 2.4 7.3 944 3 7 4.3 941 1.4 7 2 929 3 0 4.5 317 1.5 7.2 736 3 3 5 0 112 3 8 6.9 128 3.4 5.1 273 2.2 6 8 323 2 0 5.4 130 3 6 6.8 602 3.2 5.7 942 1.9 6 7

    PAGE 107

    100 Decision Rules for Choosing Conditioning Slides 1. If subject is assigned to one of the Negative slide groups, choose the six most unpleasantly rated slides from the distribution of valence ratings If subject is assigned to one of the Positive slide groups, choose the six most pleasantly rated slides from the distribution of valence ratings. If there are ties, use the normative valence ratings to break ties 2. If the subject is assigned to view calm slides, choose the two slides (among the six "candidates") with the lowest arousal ratings. If the subject is assigned to view arousing slides, choose the two (among the six candidates) with the highest arousal ratings. If there are ties, use the normative arousal ratings to select the two most extreme CS slides. 3. Within each of the four conditioning groups, which slide (CS+ or CS-) was more arousing was counterbalanced across subject gender. Decision Rules for Choosing Adjunct Slides 1. To choose two slides in each of the five adjunct affective categories, first multiply the subject s valence and arousal ratings for each slide 2. For the unpleasant/calm slide category, choose the two slides with the product closest to the number 1. 3 For the pleasant/ calm and unpleasant/ arousing slide categories, choose the two slides in each category with a product closest to the number 9. 4. For pleasant/arousing slide category, choose the two slides with the product closest to the number 81. 5. For the neutral/ calm category, choose two slides whose valence rating is closest to the number 5 and arousal rating is closest to the number 1. 6. Use the product of the norm a ti v e ratings to break any tie s.

    PAGE 108

    101 Listing of Individual Subjects' Conditioned Stimuli Pleasant/Calm Pleasant/Arousing Subject No. CS+ CSSubject No. CS+ CSCS+ more calm than CS31111 175 520 32112 469 207 31113 207 820 32114 468 466 31116 207 254 32115 468 469 31118 236 520 32117 468 469 31122 207 254 32121 469 468 31124 236 254 32123 207 468 31125 520 207 32126 803 849 31127 520 236 32128 818 254 CS+ more arousing than CS31212 207 520 32211 468 820 31214 820 466 32213 468 469 31215 236 192 32216 803 468 31217 253 520 32218 849 468 31221 236 175 32222 849 803 31223 236 520 32224 849 207 31226 849 520 32225 849 803 31228 254 236 32227 803 468

    PAGE 109

    102 Unpleasan t!Calm Unpleasant/Arousing Subject No. CS+ CSSubject No. CS+ csCS+ more calm than CS11111 710 128 12112 300 273 11113 273 317 12114 317 300 11116 929 623 12115 623 941 11118 736 130 12117 317 130 11122 929 736 12121 300 317 11124 128 323 12123 317 300 11125 601 323 12126 300 941 11127 944 317 12128 941 130 CS+ more arousing than CS11212 944 601 12211 941 300 11214 944 900 12213 317 941 11215 602 736 12216 941 323 11217 941 128 12218 942 300 11221 942 273 12222 317 300 11223 300 929 12224 112 317 11226 944 601 12225 942 317 11228 736 944 12227 623 300

    PAGE 110

    APPENDIX B PROCEDURAL INSTRUCTIONS Slide Rating Instructions (Prelearning) In today's session, you'll be viewing a series of slides while your physio logical responses are recorded. I'll be in the next room operating the equipment during the session, and the slides will be presented on the screen in front of you. Right now, I'd like you to take a moment to get the seat into a reclining position. You'll be viewing different slides and rating each on two dimensions. First, you will be controlling when the slide comes on the screen. To initiate a slide trial, just press the red button on this game paddle. A slide will then appear briefly on the screen. It is very important that you pay attention to it for the entire time it is on the screen Second after the slide goes off I'd like you to rate how ple a sant or unpleasant, and how excited or calm yo u felt while you viewed the slide. To do these ratings, use the rating sheet which I've given you On it are two rows of animated figures, which we call SAM. SAM is a reflection of how you feel during slide viewing Let me explain The top row of figures represents the Pleasure dimension which is one type of feeling SAM can represent. At the left, that's the way SAM looks if you feel completely happy, pleased, satisfied, contented, or hopeful when viewing the slide At the right, that's th e way SAM looks if you fe e l complet e ly unhappy annoyed, unsatisfi e d melancholic desp a irin g, or bored whil e yo u wa t c h t he s lid e. The cent er of th e sca l e indic a t es feeling c om p l ete l y neutra l-ne ith e r h a ppy n or un h appy. T o indi c at e h ow p l easa nt o r unp l easan t you fee l n o t e th e n u m ber unde rn ea th th e fi gure whic h bes t 10 3

    PAGE 111

    104 Pleasure Arousal Pleasure Arousal 1. 21. 2. 22. 3. 23. 4 24. 5 25. 6 26. 7. 27. 8. 28. 9. 29. 10 30. 11 31. 12 32. 13 33. 14 34 15 35 1 6. 36 1 7. 37 1 8. 38 1 9. 39 20. Figure 19. The Self-Assessment Manikin (SAM), which subjects used to rate the pleasantness (top row of figures) and arousal (bottom row of figures) of experimen tal stimuli. This is a sample rating sheet for the prelearning slide series, in which subjects viewed and rated 39 slides

    PAGE 112

    105 describes yo ur f ee l i ng Write that numb e r in the "Pleasure" blank for the trial. Notice that there are even numbers which correspond to points between the figures This permits continuous ratings along the 1 to 9 scale. The lower row of figures represents the Arousal dimension, which is another type of feeling SAM can represent. This time, SAM ranges from excited to calm At the left, that s the way SAM looks if you feel completely stimulated e x cited frenzied, jittery wide-awake, or aroused when viewing the slide. At the right, that's the way SAM looks if you feel completely relaxed, calm, sluggish, dull, sleepy or unaroused while you watch the slide. The center of the scale indicates feeling neutral--neither excited nor calm. To indicate how aroused you feel note the number underneath the figure which best d e scribe s your feeling Write that number in the Arousal" blank for the trial. Again, there are even numb e rs which correspond to points between th e figu res. This permits continuous ratings along th e 1 to 9 scale. Aft e r I l eave th e room, I'd lik e yo u t o start viewing th e s lid es. Rememb e r th a t to v i ew the n ex t slid e, y ou ll press the r e d b u t ton. Th e slid e will only appe ar bri e fly, so watch it for th e entire time it i s on th e screen After the slide goes off rate both how pleasant or unpleasant you felt and how e x cit e d or ca lm y ou felt. Mak e sur e to write the numb e r s for both the pleasur e and a r o u sa l rating s in the s pace s provided on th e rating s heet. Please don t d we ll on your ratings: Your immediate rea c tion i s w hat's import a nt. Shock E x p os u re In s truction s "OK, now comes t he pa rt o f t he experi m e nt in w hi c h I will attac h th e s h ock electrode As I showed you before this box (show Farrall shocker) and e l ectrode (show Tursky concentric electrode) w ill deliver tactile stimu l ation during the experiment. As you read in the informed consent we will go

    PAGE 113

    106 through a standardized procedure to set the level of tactile stimulation. You will be setting the shock to a level which you describe as highly annoying, but not painful. I want to reassure you that this procedure will not harm you. This experiment has been approved by a 23-member panel of doctors at Shands Hospital." After attachment of the Tursky shock electrode, the experimenter sits next to the subject with the Farrall stimulator. From left to right on the front panel of the 300-C, the switches should be set as follows: CONTINUOUS STEADY ON (POWER) OPERATE The only other setting you will use is the shock current knob, which is a 10-turn potentiometer. The level should be set at 2.52 (the beginning level, which is at about perceptual threshold). During the procedure, the shock current control on the stimulator is used to set the level. This can generally be achieved in several trials. The program is set to do 10 trial loops. On each trial, the computer delivers a 500 msec shock at whatever level is set anytime the experimenter presses a button on the computer game paddle. The following procedure is used in setting the level for each subject. 1. "I'm going to start the shock at a very low level. I'll warn you before each one, so that you'll know exactly when it's coming, OK? After each trial, I'll ask you how it felt and get your permission before administering another shock at a higher intensity. We'll work our way up to a level which you describe as highly annoying, but not painful. The first one will just be noticeable. Are you ready? OK--I'll let you know right before I do each one." 2 "Ready?" Administer shock by pressing the game paddle key. 3. For the first trial, ask: "How was that? Did you feel it?" On su bsequent trials, ask: "Was it highl y annoying?" -> if not, go to (4). If yes, go to step (5)

    PAGE 114

    107 4. "OK, then I'll turn it up slightly." If yes, increase to next mark and return to machine room, looping back to step 2. 5. "Naturally, this experiment depends on the shock being an aversive stimulus. Do you think that it could be increased once more without actually becoming painful? There won't be many shocks and each will be just as brief as those you are receiving right now." If yes, turn up the shock one level and verify that it is not painful. If no, say, "OK--that's fine" and terminate program. 6. If the subject judges a shock to be painful, say: "Let's go halfway back from here to the last shock which you said was not painful." Back off shock intensity and stop when a "highly annoying but not painful" response is reported. Record the final level on the subject data sheet. Conditioning Instructions "OK, the level of stimulation will remain exactly as we've set it. Now let me explain what will happen in this part of the experiment. After I leave the room, the slide series will begin. When the slide comes on the screen in front of you, simply watch it for the entire time it is on. Occasionally, noise will be presented over the headphones (which I will rest on your head in just a moment) During the slide series, always watch the slide for the entire time it is on and ignore the noise Also, at various times during the experiment, you will receive the shock at the level which you have already determined. At the end of this part of the experiment, I'll come back in and talk with you again. Do you have any questions before we begin?"

    PAGE 115

    12 1 0 'Z' t::: Q) 8 M M ::s u u Cl 6 ca Q) 0... '-' en 0... s < 4 s 2 0 108 power function: R-squared = 0.98 dial readings on shocker for each trial: (1) 252 (2)2 68 (3) 2.80 (4)2.90 (5)2.98 (6)3 12 (7)324 (8)338 (9) 354 (10) 3 76 0 1 2 3 4 5 6 7 8 9 10 Trial Number of Shock Workup Figure 20. Calibration of shock intensity for shock exposure procedure. The increment in shock intensity on steps of the workup approximate a power function. Dial settings on the Farrall Aversive Conditioner are listed to reproduce these intensity changes.

    PAGE 116

    APPENDIX C QUESTIONNAIRES EASI Temperament Survey Instructions: Please answer the next questions by rating your behavior on a scale of 1 to 5, where 1 = Strongly Disagree, and 5 = Strongly Agree. No item will apply to you in every situation, but try to consider your usual behavior. Please answer quickly and honestly--there are no right or wrong answers Subscale (scoring) 1. I make friends very quickly sociability ( +) 2 I like to keep busy all the time. activity(+) 3 I like to plan things way ahead of time. impulsivity (-) 4. I often feel insecure. fear(+) 5 I have trouble controlling my impulses. impulsivity ( +) 6. It takes a lot to get me mad anger(-) 7. I usually prefer to do things alone. sociability(-) 8. I often feel as if I'm bursting with energy. activity ( +) 9. I often feel like crying. fear(+) 10 I always like to see things through to the end impulsivity (-) 11. I am known as hot-blooded and quick-tempered. anger(+) 12 I am very sociable. sociability ( +) 13. I like to wear myself out with exertion activity ( +) 14. I am easily frightened. fear ( +) 15. There are many thin gs th at annoy me. anger ( +) 16. I have many friends sociability ( +) 109

    PAGE 117

    110 17. I'll try anything once impulsivity (+) 18 When I do things, I do them vigorously activity ( +) 19. When displeased, I let people know it right away anger(+) 20. When I get scared, I panic fear(+) 21. I usually seem to be in a hurry. activity(+) 22. I yell and scream more than most people my age. anger(+) 23 I tend to be impulsive. im pulsivity ( +) 24. I tend to be a loner. sociability(-) 25. I am almost always calm--nothing ever bothers me. fear(-)

    PAGE 118

    111 Fear Survey Schedule Instructions : The items in this questionnaire refer to things and experiences that may cause fear or other similar, unpleasant feelings. Read each item and indicate how much you are disturbed by it nowadays, using the following scale: 1. 2. 3. 4 5 6. 7. 8. 9 10 11. 12 13 14 15 16 17. 18 19. 20 1 = Not at all 2 = A little 3 = A fair amount 4= Much 5= Very much open wounds 21. being alone 22. loud noises 23. dead people 24 people who seem insane 25. receiving injections 26. being in a strange place 27. strangers 28 feeling angry 29. dentists 30 flying insects 31. seeing other people injected 32. sudden noises 33. being touched by others 34 witnessing surgical operations 35. angry people 36. mice 37. one person bullying another 38. human blood 39. tou g h lookin g p e opl e 40. animal blood prospect of surgery dead animals weapons journeys by airplane pollution crawling insects harmless snakes cemeteries ugly people sick people nude men nude women harmless spiders doctors people with deformities sight of knives sharp objects making mistakes being with a member of the opposit e sex b eco min g nauseo u s

    PAGE 119

    11 2 Correlations between questionnaire mea s ures (FSS and EASI total scores and subscales ) Correlations wi t hi n the Fear Survey Schedule (total score and fear subscales) FSS mutil med phob nudes noise other se lf objects FSS mu til 71 m e dic a l .73 .75 ph ob ia .60 65 5 7 nudes 29 .24 .4 6 .23 noise 41 .19 26 0 5 -.03 others 6 1 65 .68 6 0 47 21 self .47 .4 2 .6 1 .39 46 40 6 6 o bjects 3 2 6 1 .66 54 36 3 3 .6 1 .57 sex .39 3 4 .2 1 41 19 0 2 16 0 6 2 9 (l = m, 2=f) Correlations within the EASI Temperament Survey (total score and subscales) EASI fear anger acti v social impuls EASI fear 4 7 anger 72 .36 activity .54 -. 0 3 .24 sociability 4 8 -. 0 3 19 1 6 impuls .52 .0 9 30 06 .1 8 sex 20 .4 0 -.04 .1 6 0 2 -. 03 (l=m, 2=f) Correlations between Fear Survey Schedule and EASI Temperament Sur v ey FSS mutil med phob nudes noi se other self obje c ts EA S I .32 4 6 28 .16 .10 .0 8 2 0 24 .2 0 fe a r .4 8 .38 40 .21 .0 2 23 .17 29 27 anger 20 26 22 .05 .17 1 9 08 .26 .06 activi t y 1 2 1 7 .12 11 0 1 10 .15 00 17 soc i a b ili t y .01 28 .09 .08 .1 5 0 8 .10 10 .1 4 imp ul s . 06 .09 05 .00 .10 .03 -.01 09 09

    PAGE 120

    APPENDIX D AUXILIARY DATA TABLES Table 7 Orbicularis oculi means ( + standard errors) during prelearning and interslide intervals of adaptation. Exposure Group Prelearning Adapt. Block 1 Adapt. Block 2 Immediate 149 (s hock (15) ex p.) Delayed 129 (18) 132 ( 9) 128 (16) (shock exp. ) 131 ( 9) 144 (17) Shock expos ure might alter baseline tension in orbicularis o c uli, and such a change could be confounded with sensitization of startle magnitude However, in testing the effect of shock exposure on baseline orbicularis levels, the only significant finding was trial within block variance, F(2,122)= 6.86, p< .005. The Group X Trial Block interaction--which would be expected to be significant if shock exposure systematically changed orbicularis levelswas not significant (p= 34). Baseline levels were not significantly correlated with the magnitude of startle responses evoked betwe e n slides in each block of th e se n sitiza tion test although the association was positive ( rs= .17, 0 4 a nd 10, re spec tively, for baseline and a d aptation blocks 1-2). To s umm arize, orbi c ulari s oculi tension was independen t of startle reflex magnitude and cannot account for th e observed s h ock se nsitization effects. 113

    PAGE 121

    11 4 Table 8 Valence judgments of conditioning and control stimuli for all groups. CQnQitioning Change GrQup Stimulus Prelearning PQstlearning (pQst-pre) Negative/ CS+ 1.31 1.88 0.57 Calm CS1.43 2.81 1.38 difference -0 12 -0.93 -0.81 control 1.44 1.94 0.50 Negative/ CS+ 1.19 1.75 0.56 Arousing CS1.13 1.88 0.75 difference 0 06 -0.13 -0.19 control 1.22 1.81 0.59 Positive/ CS+ 8 19 5.75 -2.44 Calm CS8 25 6.19 -2 06 difference -0 .0 6 -0.44 -0.38 control 7.78 7.16 -0.62 Positive/ CS+ 8.38 6.44 -1.94 Arousing CS8 38 7 31 -1 07 difference 0.00 -0 .87 -0 87 control 7 81 7 19 -0 62 Table 9 Valence judgments of conditioning and control stimuli for Negative and Positive slide groups CQnditioning GrQups Negative (Calm and Arousing) Positi ve (Calm and Arou sing) Stimulus CS+ CS difference control CS+ CS difference co ntrol Prelearning 1.25 1.28 -0 03 1.33 8 28 8.31 -0.03 7.8 0 Change Postlearning (post-pre) 1.81 0.56 2.34 1.06 0.53 -0 50 1.88 0.55 6 09 2 19 6 75 -1.56 0.66 -0.63 7.18 -0.62

    PAGE 122

    115 Table 10 Arousal judgments of conditioning and control stimuli fo r all groups. CQnditiQning Change Group Stimulus Prelearning Postlearning (post-pre) Negative/ CS+ 3 13 4.44 1 31 Calm CS3.69 4.62 0.93 difference -0.56 -0.18 0 38 control 4.56 4.53 -0.03 Negative/ CS+ 7.13 5 63 -1 50 Arousing CS7.00 5.44 -1.56 difference 0 13 0.19 0.06 control 6.44 5 66 -0.78 Positive/ CS+ 4.00 3.88 -0.12 Calm CS4.19 4.31 0.12 differenc e -0 19 -0.43 -0.24 control 4.59 4.72 0.13 Positive/ CS+ 7.50 5.81 -1 69 Arousing CS 7.69 5.50 -2.19 dif f ere nc e 0 1 9 0 31 0.50 control 6 7 5 5.81 -0.94 T a ble 11 Arou s al judgments of conditioning and control s timuli for Calm and Arousing slide groups. Conditioning Change Groups Stimulus Prelearning Postlearning (post-pre) Calm CS+ 3 .5 6 4.16 0 60 (Positiv e and CS3 .94 4.47 0.53 N ega ti ve) diffe r ence 0 38 0.31 0 0 7 co nt r ol 4.58 4 .63 0.05 Arousing CS+ 7.31 5.72 -1 59 (Positive and CS7.34 5.44 -1 90 Nega t ive) difference -0 03 0.28 0.31 contro l 6 60 5.74 0.86

    PAGE 123

    116 Table 12 Valence judgments of adjunct slides before and after conditioning. Adjyn,t Affective Cond'g Category Groyp Prelearning PosUearning Change Unpleasant/ Neg/Calm(*) 1.44 1.94 0.50 Calm Neg/Aro 1.41 2.28 0.87 Pos/Calm 1.38 2.31 0.93 Pos/Aro 1.50 2 03 0 53 mean 1.43 2.14 0.71 Unpleasant/ Neg/Calm 1.19 1.41 0.22 Arousing Neg/ Aro(*) 1.22 1.81 0 59 Pos/Calm 1.09 1.56 0.47 Pos/Aro 1.25 1.31 0.06 mean 1 19 1.52 0.33 Neutral/ Neg/Calm 4.97 5 06 0.09 Calm Neg/ Aro 4 9 4 4.44 -0.50 Pos/Calm 4 88 4 97 0 09 Pos/Aro 5 00 4 75 -0 25 mean 4 95 4 80 -0.15 Pleasant/ Neg/Calm 8 16 7.50 0 66 Calm Neg/Aro 7 88 7 03 -0.85 Pos/Calm (*) 7 78 7.16 0.62 Pos/Aro 7 94 7 00 -0 94 mean 7 94 7 17 -0.77 Pleasant/ Neg/Calm 8.44 7.78 -0.66 Arousing Neg/ Aro 8.41 7 88 -0.53 Pos/Calm 8.28 7 97 -0 31 Po s / Aro( *) 7 81 7 1 9 -0.6 2 mean 8.23 7.70 0.53 sa m e-a ff ec t co ntr o l s lid es for each conditio nin g group

    PAGE 124

    117 Prelearning. As expected, adjunct slide categories significantly differed from one another in valence ratings before learning, F(4,240)= 2228.22, p< .0000. In post hoc tests, it was found that pleasant slides (calm and arousing) were rated significantly higher in valence than all other adjunct slides and that neutral slides were rated significantly higher than unpleasant (calm and arousing) slides. Some additional small, but significant, differences bear mention. First, despite the apparent equality of mean valence ratings for calm and arousing unpleasant slides (Ms= 1.43 and 1.19), these two categories statistically differed. Second, calm and arousing pleasant slides significantly differed (Ms= 7 94 and 8.23) in valence ratings. Finally, ratings of control slides and CSs were statistically equal for Negative groups, but control slides were rated slightly but significantly less pleasant than CSs for the Positive groups (Ms= 7 80 vs 8.30). These comparisons were made following a significant Slide Valence X Stimulus [mean of conditioning slides vs. mean of control slides] interaction in the overall analysis, F(l,60)= 17.11, p< .0002. Postlearning. Valence ratings of adjunct slides regressed towards the center of the affective space when repeated. Mean ratings of pleasant and unpleasant slides became more neutral overall from preto postlearning This was supported by a significant Pre/Post X Slide Category interaction, F(4,240)= 789.65, p< .0000. Despite this effect, adjunct slide categories still differed from one another aft e r conditioning, F(4,240)= 594.62 p< .0000. Compari sons revealed that th e ratings differences between adjunct slide categorie s after conditioning perfec tl y matched thos e found prior to conditioning. Therefore, aversive conditioning did not differentia ll y change th e affective va l e nce of s li des whic h were not involv ed in cond itioning.

    PAGE 125

    118 Table 13 Arousal judgments of adjunct slides before and after conditioning. Adj:un,t Affective Cond'g Category Gro:up Prelearning Postlearning Change Unpleasant/ Neg/Calm(*) 4.56 4.53 -0.03 Calm Neg/Aro 3.81 4.41 0.60 Pos/Calm 2.75 4 16 1.41 Pos/Aro 3.03 3 38 0.35 mean 3.54 4.12 0.58 Unpleasant/ Neg/Calm 7.66 6.31 -1.35 Arousing Neg/ Aro(*) 6.44 5 66 -0.78 Pos/Calm 7.25 6.31 -0.94 Pos/Aro 7.66 6.56 -1.10 mean 7.25 6.21 -1 04 Neutral/ Neg/Calm 2.66 3.06 0.40 Calm Neg/ Aro 2.50 2.19 -0.31 Pos/Calm 2.72 3.22 0.50 Pos/Aro 1.75 2.13 0.38 mean 2.41 2.65 0 24 Pleasant/ Neg/Calm 4.13 4.56 0.43 Calm Neg/Aro 3.63 4 .0 0 0 37 Pos/Calm (*) 4.59 4.72 0.13 Pos/ Aro 3 72 4 09 0.37 mean 4.02 4.34 0 32 Pleasant/ Neg/Calm 7.69 6.50 -1.19 Arousing Neg/Aro 7.38 6.16 -1.22 Pos/Calm 7.63 6.75 -0.88 Pos/ Aro(*) 6.75 5.81 -0.94 mean 7 36 6.30 -1.06 same-affect con trol sli des for each conditioning group

    PAGE 126

    119 Prelearning. Slide categories differed significantly in arousal, F( 4,240)= 158 71, p< 0000. Post hoc comparisons which compared adjunct slide types revealed that arousal ratings of unpleasant/ calm and pleasant/ calm slides (Ms= 3.54 and 4.02) did not differ, and both were significantly greater than ratings of neutral slides (M= 2.41). Rated arousal of unpleasant/ arousing and pleasant/ arousing slides (Ms= 7.25 and 7 36) did not differ, and was significantly greater than all other slides Group differences were found for initial arousal ratings of control slides (see Table 11); these differences appear to be a consequence of the decision rules for selecting conditioning slides. Recall that CSs were chosen to be the two most extremely rated ( calm or arousing) slides within each subject's distribution of slide ratings. Thus, control slides should be rated less extreme in relation to CS slides--but in opposite directions--for Calm vs. Arousing slide conditioning groups. This group difference between conditioning and control slides was significant, F(l,60)= 29.13, p< 0000. Follow-up t-tests indicated that Calm groups rated their control slides as significantly more arousing than their CSs (Ms= 4 58 and 3.75), whereas Arousing groups rated their control slides significantly calmer than their CSs (Ms= 6.60 and 7 33) Also, the Negative groups rated their unpleasant/ calm adjunct slides as more arousing than the Positive groups (Ms= 4.19 and 2 .89 ) This was indicated by a Slide Valence X Adjunct Category interaction F(4.240)= 3.55, p< .02 and follow-up tests. This effect was greatest for the Negative/Calm group (M = 4.56); however, the Slide Valence X Slide Arousal X Adjunct Category interaction was margin al, F(4,240)= 2.26, p< 08 This rating pattern is not surprising, since there are few s lid es having strongly normative unpleasant and calm ratings (Brad l ey, Greenwald, & Hamm in press). After choosin g C Ss for each subject in the Negative slide group (es pecially the

    PAGE 127

    120 Negative/Calm group), it is unlikely that the subject will have rated other slides as strongly unpleasant and calm. Thus, this effect probably arises from stimulus attrition in the wake of selecting CSs. Postlearning. Judgments of the different adjunct slides regressed towards moderate arousal when subjects were asked to re-evaluate the stimulus materials This change was supported by a significant Pre/Post X Slide Category interaction, F(4,240)= 149.14, p< .0000. Even so, the five adjunct slide categories continued to differ after learning, F(4,240)= 73.04, p< .0000. Again, comparisons showed that the rating differences between adjunct slide categories which were present prior to learning were preserved after conditioning Thus, the rated arousal of slides which were not involved in aversive conditioning did not differentially change.

    PAGE 128

    Table 14 Startle magnitude means (in A-O units) for all conditioning groups. Adap. Ext. Group Stirn. Adap1 Adap2 Mean Ext1 Ext2 Ext3 Ext4 Mean Negative/ CS+ 604.8 555.9 580.3 678 1 516.5 486.9 423.2 526.2 Calm CS594.9 573.4 584.2 639.9 412.6 480.9 449.1 495.6 Diff 9.9 -17 5 -3 9 38.2 103.9 6 0 -25 9 30.6 ITI 596.6 526 9 561.8 380.8 336.1 338 9 279 8 333.9 Negative/ CS+ 486.6 484.3 485 3 477.4 452.4 313.7 297.2 385.2 I-' N Arousing CS450.7 398.0 424.3 494.9 367.5 315.0 296.0 368.4 I-' Diff 35.7 86.3 61.0 -17.5 84.9 -1.3 1.2 16.8 ITI 500.7 411 4 456 1 256.0 205.0 203.7 172.4 209.3 Positive/ CS+ 325 1 364.0 344.6 378 5 289.6 256.2 209 6 283.5 Calm CS355.8 398 9 377.4 365.8 268 6 220.9 199.8 263.8 Diff -30.7 -34.9 -32.8 12.7 21.0 35.3 9 8 19.7 ITI 350 3 406.3 378.3 225.6 174.7 172 0 127.7 175 0 Positive/ CS+ 314.3 368.3 341.3 400.7 291.9 297.4 274.8 316.2 Arousing CS392.1 364.5 378.3 347.1 296.1 247.0 205.9 274.0 Diff -77.8 3.8 -37 0 53 6 -4.2 50.4 68.9 42.2 ITI 347 0 342.2 344 6 240 8 223.5 216.3 168.1 212.2

    PAGE 129

    Table 15 Startle magni t ude means (in A-D uni t s) for Slide Valence groups and t h e overall samp l e. Adap. Ext. Group Stirn Adapl Adap2 Mean Extl Ext2 Ext3 Ext4 Mean Negative CS+ 545.6 520.1 532.8 577.8 484.4 400.3 360.2 455.7 CS522.8 485 7 504 3 567.4 390.0 398.0 372.5 432.0 Diff 22.8 34.4 28.5 10.4 94.4 2.3 -12 3 23.7 ITI 548.7 469.2 508.9 318.4 270 6 271.3 226.1 271.6 I-' Positive CS+ 319.7 366 2 343.0 389 6 290 7 276 8 242.2 299.8 N N CS374.0 381.7 377.8 356.4 282.4 234.0 202.8 268 9 Diff -54.3 -15.5 -34.8 33.2 8.4 42.8 39.4 30.9 ITI 348.7 374 2 361.4 233 2 199.1 194.2 147.9 193.6 Overall CS+ 432.6 443.1 437.6 483.7 387.6 338 5 301.2 377.8 CS448 4 433.7 441.1 461.9 336.2 316.0 287.7 350.4 Diff -15.8 9.4 -3.2 21.8 51.4 22.5 13.5 27.4 ITI 448.7 421.7 435.2 275.8 234.8 232.7 187 0 232.6

    PAGE 130

    123 Table 16 Startle onset latency means (in msec) for all groups and the overall sample (n= 55). Adap. Ext. Group Stirn. Adapl Adap2 Mean Extl Ext2 Ext3 Ext4 Mean Negative/ CS+ 32.6 33.6 33.1 31.6 32.4 31.4 34.0 32.4 Calm CS33.9 33.2 33.6 33.6 35.1 34.5 35.2 34.6 (n=14) Diff 1.3 -0.4 0.5 2.0 2.7 3.1 1.2 2.2 Negative/ CS+ 31.l 31.9 31.5 30.6 29.0 30.1 32.1 30.5 Arousing CS33.0 31.0 32.0 31.4 31.9 28.0 31.0 30 6 (n = 14) Diff 1.9 -0.9 0.5 0 8 2 9 -2.1 -1.1 0.1 Positive/ CS+ 37.2 33.2 35.2 33.9 32.5 34.2 35.2 34 0 Calm CS36.l 31.9 34.0 35.1 35 8 36.1 35.1 35.5 (n = 12) Diff -1.1 -1 3 -1.2 1 2 3.3 1.9 -0.1 1.5 Positive/ CS+ 36 .3 34.1 35.2 32.5 33.3 33.7 32.9 33.1 Arousing CS33 4 32 5 33.0 34 3 35 8 36.7 35.4 35.6 (n = l5) Diff -2. 9 -1.6 -2.2 1.8 2 5 3.0 2.5 2.5 Overall CS+ 34 2 33 2 33.7 32.1 31.8 32.3 33.4 32.4 (n = 55) CS34 0 32.2 33.1 33 6 34.6 33.8 34 2 34.1 Diff -0.2 1 0 -0.6 1 5 2.8 1.5 0.8 1.7 1. Only subjects who were missing data relevant to the tests of conditioning (i.e., potentiation and discrimination) were excluded from these means 2 Positive difference scores indi c ate CS+ latencies faster than CSlatencies

    PAGE 131

    Table 17 Skin conductance response magnitude means (log [FIR+l] S) for all slide conditioning groups. Group Negative/ Calm Negative/ Arousing Posi t ive/ Calm Positive/ Aro u sing Adap. Stirn. Adp 1 Adp2 Mean CS+ .233 .246 .240 CS.256 .269 .263 Diff -.023 023 -.023 ITI 104 .092 098 CS+ .219 .234 .'227 CS.'223 248 .236 Diff ~004 ~014 ~009 ITI .057 .110 .084 CS+ .189 .175 .182 CS 133 190 162 Diff 161 -.015 .0 20 ITI .062 .038 050 CS+ .212 244 '228 CS.216 '225 .'221 Diff -.004 .019 .007 ITI .047 .073 .060 Acq. Acql Acq2 Acq3 Acq4 Mean 285 301 .254 199 .260 .196 .183 142 097 .154 .0 89 118 .112 .102 106 -------------------------.275 .'226 .201 206 '227 .196 201 .128 142 .167 079 025 .073 .064 .060 -------------------------.165 200 .264 .2 39 217 141 122 .105 097 116 024 .078 .159 .142 101 -------------------------. 272 .253 .275 .215 .254 .248 .192 125 .121 .171 .024 .061 .150 .094 .083 Ext. Extl Ext2 Ext3 Ext4 Mean 181 .102 146 .153 146 .103 .068 .093 .071 .084 .078 034 .053 .082 .062 .035 .091 .012 .060 049 .139 .141 .130 .076 121 .128 .108 .111 080 107 .011 .033 .019 -.004 .014 .015 .026 .054 .028 .031 155 .125 106 .118 126 104 110 .075 .067 089 .0 51 .015 .031 .05 1 .0 37 029 076 .042 016 .041 212 .167 140 .132 .163 090 .097 .089 .060 .084 .122 .0 70 051 .072 .079 .005 028 058 .061 .038 ,.... N .i:,.

    PAGE 132

    G ro up Calm Arousing O v erall Table 18 Skin conductance response magnitude means (log [FIR+l] S) for Slide Arousal groups and the overall sample. Adap Stirn. Adpl Adp2 Mean CS+ 211 .211 .211 CS. 195 229 212 Diff 016 018 -.001 ITI 083 .065 .074 CS+ .215 .239 .227 CS.220 .237 .229 Diff -.005 .002 -.002 ITI .052 .092 .072 CS+ .213 .225 .219 CS.207 .233 .220 Diff .006 -.008 -.001 ITI .068 078 .073 Acq Acql Acq2 Acq3 Acq4 Mean .225 250 .168 152 .057 .0 98 ----------.274 .240 .222 .196 .052 044 ----------.250 245 .195 .174 .055 .071 259 219 .123 097 136 122 ----------.238 .211 127 .131 .111 .080 ----------.249 .215 .125 .114 .124 .101 238 .135 103 -----.241 .169 .072 -----. 239 .152 .087 Extl 168 103 .065 .032 .175 .109 .066 .010 .172 .106 066 .021 Ext2 Ext3 114 .126 .089 .084 .025 .042 .084 .027 154 135 .103 100 .051 .035 .027 .056 134 .130 .096 .092 .038 .038 055 .041 Ext4 .136 .069 .0 67 .038 .104 .070 .034 .045 120 070 0 5 0 041 Ext. Mean .136 086 .0 5 0 045 .142 .095 .047 .035 .139 091 0 48 .040 Note : Spontaneous skin conductance responses during acquisition interslide intervals were not recorded. ...... N \Jl

    PAGE 133

    Table 19 Baseline heart rate means (in bprn) for each conditioning group and the overall sample. Adap Acq. Ext. G roup Stirn. Adpl Adp2 Mean Acql Acq2 Acq3 Acq4 Mean Extl Ext2 Ext3 Ext4 Mean Neg ative / CS+ 67.0 67.2 67.1 66.8 65 4 65 3 65 1 65 7 66.8 65.0 66.5 66.9 66 3 Calm cs66.1 65.2 65 7 66 3 64.3 65.7 65.8 65.5 66.4 66.0 67.1 67.4 66.7 Diff 0.9 2 0 1.4 0.5 1.1 0 .4 0. 7 0.2 0.4 -1.0 -0.6 -0 5 0.4 Neg ati v e / CS+ 71.4 69 6 70 5 70.1 66.0 68.0 70 8 68.7 69 1 69.0 67.2 68.5 68.5 Aro u s ing CS68.1 67.3 67 7 68 1 67.9 68.5 68 8 68.3 68.0 66.3 68 6 68 6 67.9 ...... Diff 3.3 2.3 2 8 2 0 -1 .9 0.5 2 0 0 .4 1 1 2.7 -1.4 -0 1 0 6 N Pos iti ve/ CS+ 77.0 72.9 75 0 72 3 75.7 73 6 74 0 73 9 74 7 76.0 75.2 75 8 75.4 Ca lm CS75.9 74.7 75.3 72.8 72.2 75.4 75.0 73.8 73.4 74.5 75.4 74 9 74.6 Diff 1.1 -1 8 -0 3 -0 5 3.5 1. 8 -1 .0 0 .1 1 3 1 5 -0 2 0 9 0.8 Posi t ive/ CS+ 70 0 70 9 70.5 66 6 68 2 69.5 6 8.5 68.2 67 8 68 2 69.0 68.0 68 .3 Aro u s in g CS71.9 69.2 70 6 67 0 7 0 .3 69 2 70 3 69.2 69.1 70.0 70.4 67.6 69 3 Diff -1.9 1.7 -0 1 -0.4 2 1 0 .3 -1.8 -1.0 1.3 1 8 1.4 0.4 -1. 0 Overall CS+ 71.3 70.1 70 7 68 9 68.7 69 0 69 5 69.0 69 5 69.5 69.4 69 7 69.5 CS70.4 69.0 69.7 68 5 68.6 69.6 69.9 69 2 69 2 69.2 70.3 69 6 69 6 Diff 0 9 1.1 1.0 0.4 0. 1 0. 6 -0.4 -0 2 0 3 0 3 -0.9 0 1 -0 1

    PAGE 134

    Table 20 Initial heart rate deceleration means (D1 leg; ~bpm) for each conditioning group and the overall sample. Adap Acq. Ext. Group Stirn. Adpl Adp2 Mean Acql Acq2 Acq3 Acq4 Mean Extl Ext2 Ext3 Ext4 Mean Nega tive / CS+ -4.37 -4.91 -4.64 -3.14 -3.13 -2.84 -1.70 -2.70 -2.23 -1.93 -2.89 -3.77 -2.71 Calm CS-2.32 -2.78 -2 55 -2.35 -1 59 -2.97 -3.42 -2.58 -3.13 -2 75 -1.37 -4 24 -2.87 Diff -2.05 -2.13 -2.09 0.79 1.54 0.13 1.72 -0.12 0.90 0 82 -1.52 0.47 0.16 Nega tive / CS+ -6.71 -4.52 -5.62 -6.97 -3.68 -2.51 -3.63 -4.20 -5.80 -5.06 -2.17 -3.31 -4 09 Aro u si ng CS-3.77 -3.43 -3.60 -4.31 -4.72 -5.12 -4.98 -4 78 -2.85 -2 22 -2.81 -3.30 -2.80 ,..... Diff -2.94 -1.09 2 02 2 66 1 04 2.61 1.35 0 58 -2.95 -2.84 0 64 -0 01 -1.29 N -..J Positive/ CS+ 2.94 -3.22 -3.08 -2.23 4.76 0.90 -4.30 -3 05 -2.36 -4.60 -5.08 -2.91 -3.74 Calm CS-1.53 -4 12 -2 83 -1.96 -2.39 -3.15 -2.99 -2.62 -2.60 -4.28 -4 01 -2.11 -3.25 Diff -1.41 0 90 -0.25 -0.27 -2.37 2.25 -1.31 -0.43 0 24 -0.32 -1.07 -0.80 -0.49 Positive/ CS+ -2.77 -3.02 -2.90 -1.74 2.78 -3 .25 -3.52 2 82 -1.27 -2.30 -1.96 3.37 -2.23 Aro u si n g CS-3.23 -2.83 -3.03 -0.47 -5.51 -4.46 -4.10 -3.64 -2.83 -4.03 -3.43 -2.73 -3.26 Diff 0.46 -0.19 0.13 -1.27 2.73 1.21 0.58 0.82 1.56 1.73 1.47 -0 64 1.03 Overall CS+ -4 22 -3.93 -4.08 -3 54 -3.57 -2 .4 0 -3.27 -3.20 -2.93 -3.46 -2.99 -3.35 -3.18 CS 2.73 3.28 3 01 2.27 -3 .57 -3 .94 3 88 3.42 2.86 -3.30 -2.89 3.11 -3.04 Diff 1.49 -0.65 -1 07 1 .27 0.0 1.54 0.61 0.22 -0.07 -0.16 -0.10 -0.2 4 -0.14

    PAGE 135

    Table 21 Heart rate acceleration means (Al leg; ~bpm) for each conditioning group and the overall sample. Adap. Acq. Ext. Group Stirn. Adpl Adp2 Mean Acql Acq2 Acq3 Acq4 Mean Extl Ext2 Ext3 Ext4 Mean Negative/ CS+ 1.40 1.78 1.59 2.18 3.35 3.15 4.55 3.31 5.25 2 99 3.94 4.32 4.13 Calm CS3.45 3.71 3.58 2.79 5.11 4.58 2.79 3.82 3.69 3.72 5.10 2.77 3 .82 Diff -2.05 -1.93 -1.99 -0.61 -1.76 -1.43 1.76 -0.51 1.56 -0.73 -1 16 1 55 0.31 Negative/ CS+ 1.71 2.11 1.91 0.61 2.87 4 22 5 01 3.18 3.44 2.39 4.84 2.92 3 40 Arousing CS4.50 5.04 4.77 1.60 1.84 0.79 2 65 1.72 4.92 5.33 4.35 2.90 4.38 ,-. Diff -2.79 -2.93 -2.86 -0.99 1.03 3.43 2.36 1.46 -1.48 -2.94 0.49 0.02 -0.98 N O'.l Positive/ CS+ 1.78 4.59 3.19 4.38 0.06 5.42 3.75 3.40 2.28 0 56 2.07 2.46 1.84 Calm CS4.50 3.03 3.77 3.27 3.08 3.08 3.35 3.20 3.32 2.12 2.06 3.38 2.72 Diff -2.72 1 56 -0.58 1 11 -3.02 2.34 0.40 0.20 -1 .0 4 -1 56 0.01 0.92 -0 .88 Positive/ CS+ 4.35 2.73 3 54 4.49 5.23 4.53 5 06 4.83 5.77 4.04 4.62 3.51 4.49 Arousing CS4.39 5.59 4.99 7 65 2.36 3.00 2.00 3.75 3 26 3.27 2.27 4.73 3.38 Diff -0.04 -2 86 -1.45 -3.16 2.87 1.53 3.06 1.08 2.51 0.77 2 35 -1 22 1.11 Overall CS+ 2.32 2.77 2.55 2.89 2 .9 2 4.31 4.61 3.68 4.21 2 53 3.90 3.32 3.49 CS4.21 4.36 4.29 3 84 3.10 2.86 2.69 3.12 3.81 3.63 3.47 3 45 3.59 Diff -1.89 -1.59 -1.74 -0.95 -0.18 1.45 1 92 0 56 0.40 -1 .10 0.43 0.13 0.10

    PAGE 136

    Table 22 Late heart rate deceleration means (D2 leg; L\.bpm) for each conditioning group and the overall sample. Adap. Acq. Ext. Group Stirn. Adpl Adp2 Mean Acql Acq2 Acq3 Acq4 Mean Extl Ext2 Ext3 Ext4 Mean Nega t ive/ CS+ -3 79 -3.08 -3 44 -3.67 3.77 -4.16 -2.70 -3.58 -3.17 -3.54 -4.67 -3 35 -3 68 Calm CS-2.12 -1.74 -1.93 -3.53 -4.13 -4.77 -5 92 -4.59 -3.86 -3.36 -4.25 4 96 4 11 Diff -1.67 -1.34 -1.51 -0.14 0 36 0.61 3 .22 1 01 0.69 -0.18 -0.42 1.61 0 43 Nega ti ve/ CS+ -5 11 -5.48 -5.30 -7 62 -4.62 -7.21 -6.83 -6.57 -6.32 -7.13 -4.93 -7.23 -6.40 Aro u si ng CS-3.28 -2.32 -2.80 -7 41 -6.59 -6.84 -6.67 -6 88 -4.72 -1.50 -4 51 -5.89 -4 16 Diff -1.83 -3.16 -2.50 -0.21 1 97 -0.37 -0.16 0.31 -1.60 -5 63 -0.42 -1.34 -2 24 ...... N \D Posi ti ve/ CS+ -5 09 -1.72 -3 41 -4 13 -7.27 -6 19 -5.80 -5 85 -4.59 -6.04 -5 75 -6.40 -5 70 Calm CS-3.07 -3 45 -3.26 -4 74 -4.77 -4 78 -6.24 -5 13 -4.00 -3.75 -5.48 -4.37 -4 40 Di ff -2.02 1.73 -0.15 0.61 -2.50 -1.4 1 0.44 -0.72 -0 59 -2.29 -0 27 -2 .0 3 -1 3 0 Posi t ive/ CS+ -2.08 -4.64 -3 36 -2.93 -4.55 -6.50 -5 88 -4 97 -3.17 -5 57 -4.96 -5.93 -4 91 Arousing CS-3.84 -1.79 -2.82 -1.97 -6 08 -5.51 -6.68 -5.06 -3.85 -5.11 -5.10 -2.94 -4.25 Diff 1.76 -2.85 -0.54 -0.96 1 53 -0.99 0 80 0 09 0.68 -0.46 0.14 -2 99 -0 66 Overall CS+ -4.00 -3.76 -3.88 -4.59 -5. 02 -6.01 -5.29 -5.23 -4.31 -5.60 -5.07 -5.72 -5.18 CS-3.07 -2.31 -2.69 -4.41 -5 .4 1 -5 .4 9 -6.38 -5.42 -4 11 -3.42 -4.83 -4.54 -4 23 Diff -0.93 -1.45 -1.19 -0.18 0.39 -0.52 1.09 0. 19 -0.20 -2.18 -0.24 -1 18 -0.95

    PAGE 137

    Exposure Group Immediate Delayed 130 Table 23 Distribution of subjects in shock aversion groups across experimental grouping variables. Low Moderate High Conditioning Group Aversion Aversion Aversion Total Neg/Calm 3 2 7 12 Neg/ Arousing 3 7 2 12 Pos/Calm 4 5 3 12 Pos/ Arousing 2 4 6 12 Subtotal 12 18 18 48 Neg/Calm 2 0 2 4 Neg/ Arousing 2 0 2 4 Pos/Calm 0 1 3 4 Pos/ Arousing 1 1 2 4 Subtotal 5 2 9 16 Total 17 20 27 64 Shock Aversion grouping was independent of the a priori experimental grouping variables (Slide Valence, Slide Arousal, and Shock Exposure). The point-biserial correlation between exposure group and shock aversiveness ratings was .04 (ns) In addition, shock exposure timing and aversion did not interact in determining startle or skin conductance potentiation. Thus, these factors were independent. (Note: the cell sizes in Table 23 are too small to perform a chi-square test of independence.)

    PAGE 138

    131 Table 24 Cardiac acceleration for shock aversion groups during adaptation and acquisition. Aversion Adap. Acq. Group Stirn. Adapl Adap2 Mean Acql Acq2 Acq3 Acq4 Mean Low CS+ 2.63 3.79 3.21 3.14 2 70 5.52 4.52 3.97 (n=17) CS2.81 5 55 4.18 3.49 2.67 3.24 3.97 3.34 Diff -0 18 -1.76 -0.97 -0.35 0.03 2 28 0.55 0.63 Moderate CS+ 0.60 1.78 1.19 1.80 2.60 3.88 2.45 2.68 (n=20) cs3.74 4.52 4.13 2.99 2.50 0.51 3.02 2.25 Diff -3 14 -2.74 -2 94 -1.19 0 10 3.37 -0.57 0.43 High CS+ 3.44 2.87 3.16 3.57 3 32 3.85 6.33 4.27 (n=27) CS5.48 3.47 4.48 4.71 3.84 4.42 1.59 3.64 Diff -2.04 -0.60 -1 32 -1.14 -0.52 -0.57 4.74 0 63 High Aversion subjects showed greater cardiac acceleration to CS+ than CS in the last block of acquisition, compared with subjects rating the shock as less aversive. This observation was supported by an Aversion Group X Stimulus X Trial Block F(6,180) = 2.19, e= 98, p< 05, and a significant group difference at trial block 4, F(l,61)= 6.56, p< .02. In the preceding trial block however, the Moderate Aversion group tended to show better conditioning than the High Aversion group, but this impression was not supported statistically (p> .07). The difference between Low and High Aversion groups at trial block 3 was also not significant (p = .25). However, the pattern of cardiac response is sufficiently reversed to call the stability of the effect into question. Interpretation of these data is also difficult in this experiment due to pre e x i s ting diff e rences betw ee n CS + and CScardiac response in adaptation.

    PAGE 139

    132 Table 25 Descriptive statistics for central experimental variables. Varis!l2le Standard Range: Name Mean Deviation Low High CS+ valence, pre learning 4.77 3.61 1.00 9.00 valence diff. -0.59 1.52 -5.00 2.00 valence pot. -0.81 2.02 -6 00 4.00 US valence 1.97 1.27 1.00 5.00 CS+ arousal, pre learning 5.44 2.70 1.00 9.00 arousal diff. -0.03 1.69 -4.00 4.00 arousal pot. -0.50 2.44 -8.00 7.00 US arousal 7 19 2 03 1.00 9.00 FSS 102.22 22.28 59.00 148 00 EASI fear 14.42 3.32 6.00 22 00 EASI anger 13.55 4.03 7 00 25.00 EASI activity 16.03 2.74 10.00 24.00 EASI sociab. 17.48 3.98 6 00 24 00 EASI impuls. 13 64 3.33 7.00 22.00 # shocks 6.52 2.17 2.00 15.00 US aversion 6 05 1.03 3 00 7.00 CS+ aversion 5.50 1.46 1.00 7 00 CS avers ion 2.20 1.51 1.00 6.00

    PAGE 140

    133 Table 25--continued Variable Standard Range: Name Mean Deviation Low High probe avers. 4.78 1. 7 2 1.00 7 00 US intensity 6.00 4.13 1.30 12.20 str. mag. potentiation 40 53 236.87 -866.50 800 00 str. mag. discrim. 27.31 89.78 -140.88 439.50 HR potent. (adap->acq) 1.72 6.37 -11.46 19.88 HR discrim. (acquisition) 2 03 7 62 -14 27 25 17 SCR potent. (adap->acq) 0.02 0.19 0 28 0 80 SCR potent. (adap->ext) -0.05 0.20 -0.63 0.48 SCR discrim. (acquisition) 0.09 0.11 -0.09 0.46 SCR discrim. ( extinction) 0.05 0 11 -0.12 0.68

    PAGE 141

    Variable sex pr ev al v aldif v al pot s hkval prearo arodif aropot shkaro e x pgrp fss efear ea n ge r eactiv eso c e imp n s hock s hock e s p cs m pro b e s h ki nt s t rp r e se n s s trpot s trdif hrp o t hrdif s capot sce p o t sca dif s ce dif f ss grp Table 26 Correlations Among Experimental Variables Meaning gender (male = 1, female=2) CS+ valence, pre-learning CS+ /CSvalence difference post-learning (valence discrimination) change in CS+ valence from preto post-learning (valence potentiation) rated valence of the shock CS+ arousal, pre-learning CS+ /CSarousal difference, post-learning (arousal discrimination) change in CS+ arousal from preto post-learning (arousal potentiation) rated arousal of the shock US shock exposure group (1 = immediate, 2 = delayed) Fear Survey Schedule total s core EASI fear subscale EASI anger subscale EASI activity subscale EASI sociability subscale EASI irnpulsivity subscale subject's estimate of the number of shocks received during conditioning rated aversion of the shock US rated aversion of the CS+ slide rated aversion of the CSslide rated aversion of the s tartle prob e phy s ical inten s it y of el ec tr ic s h o ck (rnA) pre-learning s tartle magnitud e ( ave ra g e of fir s t two probe re s pon s e s ) startle magnitude sen s itizat io n startle magnitude potentiatioh startle magnitude discrimination ~CS+ acceleration from adaptation block 2 to acquisition block 4 (heart rate potentiation) difference in heart rate acceleration in acquisition block 4 (heart rate discrimination) SCR change from adaptation to acquisition (FIR potentiation) SCR change from adaptation to e x tinct i on SCR difference in acqui s ition SCR differ e nce in e x tin c ti o n ( F IR di s crimination) fear group u s in g FSS c utp o int s within sex (male s, 89 ; females 111) ...... w .p...

    PAGE 142

    Table 26--continued sex preval valdif valpot shkval prearo arodif aropot shkaro expgrp sex 1. 0000 preval 0 .0044 1 0000 valdif -0.2488 0 0635 1.0000 val pot -0.2180 -0 7097 0 .35 70 1. 0000 shkval -0 2476 0.1610 -0 00 15 0 0887 1. 0000 prearo 0 0701 0.1560 0.0402 0 0852 -0 0330 1 0000 arodif 0 2050 0 0118 0 0135 0 1143 0 .0964 0.229 5 1. 0000 aropot -0.0517 -0.1599 -0.0643 -0.0451 -0 0768 0 5365 0 1889 1. 0000 shkaro -0 1086 -0 .0633 0. 11 25 -0 0164 -0 2128 0 2806 0 09 42 0 2085 1. 0000 expgrp 0 0000 -0.0328 0 0120 0 0360 0 07 15 0 0000 0 0 968 0 0597 -0 0179 1 0000 fss 0 .3916 -0 0977 -0 .1096 -0 0738 -0 2350 0 0989 0 0 453 0 1708 0.0478 0 1853 efear 0 .4029 0.1082 0 .1 445 0 0306 -0 1658 0 .1953 0 .1833 -0.1539 0.0354 0.1012 eanger -0.0430 0 0068 -0.0576 0.0206 0 0313 0.1443 0 .11 22 0 0412 0 0222 0 1648 eactiv 0 .1608 0 0986 -0 1326 -0.1870 -0 1499 0 0582 0 1509 0 2138 -0.0495 0 1392 esoc 0.0198 0.1995 0 .2 297 -0.0273 0 0281 0 0527 -0.0497 0.0287 -0.0743 -0 0617 ...... eimp -0.0331 0 0071 0.0293 0.0879 -0.0889 0 0072 -0 0359 0 03 62 0.1017 0.0519 w V, nshock 0.0218 0.1538 0.0559 -0 .14 91 -0.09 1 9 0 0 45 0 0 1255 0 05 5 6 0.1003 -0 1217 s h ock 0 2599 0 .1 098 0.1645 0 19 4 7 -0.5682 0.2840 0 20 14 0 0664 0 5117 0 0441 esp 0 2592 -0.1041 -0.1218 0 0538 -0 3679 0 2746 0.1224 -0.0268 0 2626 0 0249 csm 0 0104 -0 4040 0 0809 0 2103 -0 1450 0 0 596 -0 00 99 0 2216 0.2611 0 1261 probe 0 0731 -0.2077 -0.1838 0 0712 0 .141 7 0 0953 0 1496 0 0719 0 0482 0 1584 s hkint -0 5739 0.1903 0 0048 0 0302 0.2688 0 0622 0 .1548 0 0 44 4 -0 0617 0 0348 s trpre 0.2359 -0 2591 0 0029 0 0810 -0.1823 0 0755 -0 0306 0 0826 0.0582 0 .1 666 se ns 0.2019 -0 0088 -0.1817 0 1050 0.0397 0 1001 0 0766 0 2418 -0. 0130 -0. 0256 s trp o t -0 0763 0.0290 0 1633 0 0009 -0.2643 0 0081 -0.0908 -0 0748 0 2329 -0 .3 106 strdif -0 .0823 0 0815 -0 1773 -0 0342 -0.0673 0 0036 -0 0 45 0 -0 1448 -0.0598 0 0255 hrpot -0 0540 -0 .2906 -0 0074 0 3291 -0. 0671 0 .1129 0 0895 0 0854 0 0792 0 1755 hrdif 0.1202 -0.0012 -0.1 25 4 0 083 6 -0 0533 0 1907 0.1315 0 0593 0 0038 0 .1 675 sea pot 0 1957 -0 0341 -0 0376 -0 1779 -0 4363 0 1376 0.3383 0 0 134 0 2170 -0.2451 scepo t -0 .0506 0 .1540 0.2055 -0 1432 -0 2177 0 .1 514 0 0277 0 0726 -0.0423 -0.21 04 scadif 0 3550 0.0706 0 1942 0 2027 -0 1933 0.0015 0 1460 0 0266 0 0631 -0.18 84 scedif 0.1061 0 1114 -0.0309 -0 1643 -0 1004 0.0055 -0 0630 -0 1285 -0 1707 0 0878 fssgrp 0.0000 -0.0742 0.1244 0 0467 -0.0743 -0.0351 -0. 1305 0 0388 0.1241 0 .14 43

    PAGE 143

    Table 26--continued fss efear eanger eaetiv esoc eimp nshoek shock esp esm fss 1 0000 efear 0 .4775 1. 0000 eanger 0 .203 5 0.3620 1. 0000 eactiv 0 1214 -0.0293 0.2427 1. 0000 esoc 0.0131 -0.027 7 0.1874 0 1572 1. 0000 eimp 0 .0646 0.0899 0.3001 0 0621 0.182 4 1.0000 nshoek 0 0213 -0.1365 0 .1381 0.1174 0 .1197 0 2130 1. 0000 shock 0 3031 0.2028 0.0473 0.2129 0.0525 0.1207 0 13 12 1.00 00 esp 0 1578 0.2309 -0. 0365 0 0794 0. 1437 0.0866 0.2285 0.4701 1. 0000 esm 0 .1883 -0.0836 -0.0992 -0.0742 0.0746 -0 1648 0.0692 0.1770 0 2265 1 0000 probe 0 .1733 0.1467 0.1015 -0.1665 -0 1743 -0 .3238 0 0094 -0 1194 0 1200 0 2730 shkint -0.3135 -0.3326 0.0571 -0.1244 0. 1352 0 .2759 0.0788 -0.3505 -0. 2969 -0 1401 strpre 0.0 59 0 0 2188 -0.1793 -0.16 03 0 .1133 -0.2117 -0.1348 0 0005 0.1232 0 3378 sens 0 1358 -0 .14 75 0 1283 0.1719 0 .1 228 0. 0596 0. 0377 0.0176 -0.0342 0 0 49 3 strpot 0 0215 -0 .0 620 0 0068 0. 003 6 -0.0272 0.2100 0 2140 0 4 279 0 .2102 0 1819 strdif -0.17 46 0 0395 0 0431 0 0301 -0 0551 0.0601 -0 1006 0 0860 0 0399 -0.0583 ..... hrpot -0.1090 0.0490 0 .1 019 0.0058 0.0729 0 .1100 -0.2465 -0.0989 0 2071 0.0964 w a, hrdif -0 0598 0 0512 0 0050 0 2441 0 0865 0 0185 -0 0381 0 0789 0 23 40 0 0885 sea pot 0 03 40 0.1826 0.0373 -0.0232 0 .0664 0 2564 0 1032 0 3158 0 2065 -0 0128 seepot -0 07 14 0 .1619 -0 0470 -0.0137 0 1590 0 .163 7 0.0857 0 0237 0 012 1 -0.0841 seadif -0.0 525 0 0163 0 .11 30 0.1948 0 0693 0 3116 0.1913 0 2845 0 1122 -0.0561 sce dif -0 0907 -0.0102 0 .111 0 0 .34 05 0 0 469 0 1199 -0 0317 0 03 18 0 0520 -0 0174 fssgrp 0.7407 0.3175 0.1838 0.1378 0 0198 0 0804 -0 0509 0 .1682 0 0432 0.1977

    PAGE 144

    Table 26--continued probe shkint strpre sens strpot strdif hrpot hrdif scapot scepot pr o b e 1 0000 s h kin t -0.1408 1. 0000 s trpre 0 .2398 -0 2246 1. 0000 s en s 0 1573 -0 0340 -0 0266 1 0000 s trpot -0 0420 -0 0608 0 1171 0 0090 1.0000 strdif -0.0005 0.026 4 0 0034 -0 0608 0 3490 1 0000 hrpot -0 .1380 0 1618 -0.0695 -0.1995 -0 1171 0.1579 1. 0000 hrdif 0 .2025 -0 0523 0.0220 -0 0195 -0 1284 0 0340 0 5192 1 0000 s ea pot -0 0707 -0 2018 0 .1511 -0.0794 0 .3266 0 0337 -0 1090 0.0845 1.0000 s cepot -0 .0860 0 1016 -0 1317 0 .125 0 0 3325 0 0971 -0.0317 -0 0250 0 3992 1 0000 sc adif -0 .1091 -0 2397 0.0281 0 0 444 0 3896 0.1688 -0.0464 0.1476 0. 6211 0 2911 see d i f -0.1556 -0 0509 0.0987 0 0300 0 22 43 0 4096 0 1138 0 2042 0 1583 0 5042 fss grp 0 2560 -0 1259 0.0 40 7 0 .11 88 0 .1763 -0 1747 -0 1472 -0 0491 -0 0211 0 0835 scadif scedif fssgrp ..... w --.J s cadif 1 0000 s cedif 0 3699 1. 0000 f ss grp -0 1659 -0 0797 1 0000

    PAGE 145

    REFERENCES Anthony, B.J. (1980). In the blink of an eye: Implications of reflex modification for information processing In P.K Ackles, J R. Jennings, & M.G.H. Coles (Eds.), Advances in psychophysiology, Vol. 1 (pp 167218). Greenwich, CT: JAI Press Applegate, C.D., Kapp, B.S., Underwood, M.D., & McNall, C.L. (1983). Autonomic and somatomotor effects of amygdala central nucleus stimulation in awake rabbits. Physiology and Behavior, al, 353-360. Armus, H.L., Carlson, K R., Guinan, J.F., & Crowell, RA (1964). Effect of a secondary reinforcement stimulus on the auditory startle response. Psychological Reports, H., 535-540. Armus, H.L., & Sniadowski-Dolinsky, D. (1966) Startle decrement and secondary reinforcement stimulation. Psychonomic Science, 1-, 175-176 Barratt E.S. (1971). Psychophysiological correlates of classical differential eyelid conditioning among subjects selected on the basis of impulsiveness and an x iety Biological Psychiatry .3., 339-346. Barratt E.S. (1983). The biological basis of impulsiv e ness : The significance of timing and rhythm disorders. Personality and Individual Differences, 1-, 387-391. Berg, W K., & Davis, M (1984). Diazepam blocks fear-enhanced startle electrically elicited from the brainstem. Physiology and Behavior, 12., 333-336. Berg, W.K., & Davis, M. (1985). Associative learning modifies startle reflexes at the lateral lemniscus Behavioral Neuroscience, 22., 191-199. Bohlin, G., & Kjellberg, A. (1979). Orienting acti v it y in two-stimulus paradigms as reflected in heart rate In H.D Kimm e l E H. van 01st & J F. Orlebeke (Eds ) Th e orienting refle x in hu mans (p p 168-189) Hills d a l e, NJ: E rlb aum. 1 3 8

    PAGE 146

    139 Boulis, N M., & Davis, M. (1989). Footshock-induced sensitization of electrically elicited startle reflexes Behavioral Neuroscience. lQ3., 504508. Bradley, M M (1992). [Factor analysis of temperament questionnaire data]. Unpublished raw data Bradley, M M Cuthbert, B.N & Lang, P.J (1990a). Habituation and the affect-startle effect. Psychophysiology. 2L Sl8. (Abstract) Bradley, M.M Cuthbert, B.N., & Lang, P.J (1990b) Startle reflex modification : Emotion or attention? Psychophysiology. 2L 513-522 Bradley, M M., Cuthbert, B.N & Lang, P.J. (1991). Startle and emotion: Lateral acoustic probes and the bilateral blink. Psychophysiology. 2.8., 285-295. Bradley, M M Greenwald, M.K., & Hamm, A O (in press). Affective picture processing In N. Birbaumer & A. Ohman (Eds ), The structure of emotion: PsychophysiologicaL cognitive and clinical aspects. Toronto: Hogrefe-Huber. Brown, J.S., Kalish, H.I., & Farber I.E (1951) Conditioned fear as revealed b y magnitude of startle response to an auditor y stimulus Journal of E x perimental Psycholog y. il., 317-328. Buss, A ., & Plomin, R. (1975 ). A temperament th e ory of personality de v elopment. New York : Wiley-Interscience Cassella, J.V & Davis, M. (1986) Habituation, prepulse inhibition, fear conditioning, and drug modulation of the acoustically elicited pinna refle x in rats. Behavioral Neuroscience.1.QQ., 39-44. Chase, W.G., Graham, F K. & Graham, D T. (1968). Components of HR response in anticipation of reaction time and exercise tasks. Journal of Experimental Psychology Zfl, 642-648 Chattop a dhyay, P Cooke E ., T o one B & Lader M. (1980) Habituation of physiological respons e s in a nxiety. Biological P s y c hiatry, 711-721. Cook, E. W III Hawk L.W ., Dav i s, T.L. & St eve n so n V.E. (1991) Aff ective i ndividua l di ff erences and star tl e r eflex m odulation. J o urn a l o f Abnormal Psyc h o l ogy. 100 5-13.

    PAGE 147

    140 Cook, E.W III, Hodes, R.L., & Lang, P J (1986). Preparedness and phobia: Effects of stimulus content on human visceral conditioning. Journal of Abnormal Psychology,~ 195-207. Cuthbert, B.N., Bradley, M.M., York, D J., & Lang, P J (1990) Affective imagery and startle modulation. Psychophysiology ll, S24. (Abstract) Davis, M (1972). Differential retention of sensitization and habituation of the startle response in the rat. Journal of Comparative and Physiological Psychology,~ 260-267 Davis, M. (1974a). Signal-to-noise ratio as a predictor of startle amplitude and habituation in the rat. Journal of Comparative and Physiological Psychology,~ 812-825. Davis, M. (1974b). Sensitization of the rat startle response by noise. Journal of Comparative and Physiological Psychology, 8Z., 571-581. Davis, M (1986). Pharmacological and anatomical analysis of fear conditioning using the fear-potentiated startle paradigm. Behavioral Neuroscience, l.QQ, 814-824. Davis, M (1989a) The role of the amygdala and its efferent projections in fear and anxiety. In P Tyrer (Ed.), Psychopharmacology of anxiety (pp. 52-79). New York: Oxford University Press. Davis, M. (1989b). Sensitization of the acoustic startl e reflex by footshock. Behavioral Neuroscience 103, 495-503. Davis, M., Gendelman, D.S Tischler, M D., & Gendelman, P M. (1982). A primary acoustic startle circuit: Lesion and stimulation studies. Journal of Neuroscience, .6., 791-805 Davis, M., Hitchcock, J M., & Rosen, J B (1989). Anxiety and the amygdala: Pharmacological and anatomical analyses of the fear-potentiated startle paradigm. In G.H. Bower (Ed.), The psychology of learning and motivation, Vol. 21 (pp. 263-305). New York: Academic Press Davis, M. Redmond, D.E., & Baraban, J.M (1979). Noradrendergic agonsists and antagonists : Effects on conditioned fear as measured by the potentiated startle paradigm. Psychopharmacology, 65., 111-118. Deane, G.E (1969). Cardiac activity during experimentally induced anxiety. Psychophysiology, .6., 17-30.

    PAGE 148

    141 Dickinson, A. (1977). Appetitive-aversive interactions: Superconditioning of fear by an appetitive CS Quarterly Journal of Experimental Psychology, 21., 71-83. Dickinson, A., & Dearing, M F. (1979). Appetitive-aversive interactions and inhibitory processes. In A. Dickinson & RA Boakes (Eds.), Mechanisms of learning and motivation (pp. 203-231). Hillsdale, NJ: Erlbaum. Duffy, E. (1972). Activation and behavior. New York: Wiley. Estes, W.K. (1973). Memory and conditioning. In F.J. McGuigan & D.B. Lumsden (Eds.), Contemporary approaches to conditioning and learning (pp. 265-286). New York: Halsted Press. Eysenck, H.J. (1967). The biological basis of personality. Springfield, IL: Thomas. Eysenck, H.J., & Levey, A.B. (1972). Conditioning, introversion-extraversion and the strength of the nervous system. In V.D. Nebilytsin & J.A. Gray (Eds.), Biological bases of individual behavior (pp. 206-220). New York: Academic Press. Fanselow, M.S (1980) Conditional and unconditional components of post shock freezing. Pavlovian Journal, Th 177-182 Fanselow, M.S., & Bolles R.C. (1979). Naloxone and shock-elicited freezing in the rat. Journal of Comparative and Physiological Psychology,~ 736-744. Farha, J.G., & Sher, K J. (1989) The effects of consent procedures on the psychophysiological assessment of anxiety: A methodological inquir y. Psychophysiology 2.6., 185-191. Fowles, D.C. (1980). The three arousal model: Implications of Gray's two factor learning theory for heart rate, electrodermal activity and psychopathy. Psychophysiology, 1Z 87-104 Frey, P W., & Sears, R.J. (1978) Model of conditioning incorporating the Rescorla-Wagner associative axiom, a dynamic attention process and a catastrophe rule. P sy ch o logical Review, B.2, 321-340 Ga r c i a, J. & Ko e lling R. A ( 1966 ) Relation of c u e t o co n se qu e n ce in av oid a n ce l ea rni ng P sy cho nomic Sci e n ce,~ 1 23-124

    PAGE 149

    142 Gatchel, R.J., & Lang, P.J. (1973) Accuracy of psychophysical judgments and physiological response amplitude. Journal of Experimental Psychology 2.8., 175-183. Gentile, C.G., Jarrell, T.W., Teich, A., McCabe, P.M & Schneiderman, N. (1986). The role of the amygdaloid central nucleus in the retention of differential Pavlovian conditioning of bradycardia in rabbits. Behavioral Brain Research, 2..Q, 263-273. Graham, F.K. (1980). Representing cardiac activity in relation to time. In I. Martin & P. Venables (Eds.), Techniques in psychophysiology (pp 192-197). Chichester: Wiley. Gray, J.A. (1967). Strength of the nervous system, introversion-extraversion, conditionability and arousal. Behaviour Research and Therapy, :2., 151169. Gray, J.A. (1973). Causal theories of personality and how to test them. In J.R. Royce (Ed.), Multivariate analysis and psychological theory (pp. 409-463). London: Academic Press. Gray, J.A. (1982). Precis of The neuropsychology of anxiety: An enquiry into the functions of the septo-hippocampal system Behavioral and Brain Sciences, 469-534 Greenwald, M.K., Bradley, M.M., Cuthbert, B.N., & Lang, P.J. (1990). The acoustic startle response indexes aversive learning Psychophysiology, '2L S36. (Abstract) Greenwald, M.K., Cook, E.W. III, & Lang, P.J (1989) Affective judgment and psychophysiological response : Dimensional covariation in the evaluation of pictorial stimuli. Journal of Psychophysiology, 3., 51-64 Greenwald, M.K., Hamm, A.O., Bradley, M.M & Lang, P.J. (1988). The acoustic startle probe in the assessment of classical aversive conditioning. Psychophysiology, 451. (Abstract) Grillon, C., Ameli, R., Merikangas, K., Woods, S.W., & Davis, M. (in press) Measuring the time course of anticipatory anxiety using the fear potentiated startle reflex. Psychophysiology Groves P M., & Thompson RF (1970) Habituation: A dual-process theor y Psychological Review, TL 419-450.

    PAGE 150

    143 Hamm, A O., Greenwald, M K., Bradley, M.M & Lang, P.J. (1991) Emotional learning, hedonic change and the startle probe Manuscript submitted for publication. Hamm, A.O Stark, R., & Vaitl, D. (1990). Startle reflex potentiation and electrodermal response differentiation: Two indicators of two different processes in Pavlovian conditioning. Psychophysiology, 27, S37. (Abstract) Hammond, G.S., Baer, P.E & Fuhrer, M.J. (1980). Retention of differential autonomic conditioning and memory for conditional stimulus relationships. Psychophysiology, 1Z, 356-362. Harper, R.M., Frysinger, R.C., Trelease, R.B., & Marks, J.D. (1984). State dependent alteration of respiratory cycle timing by stimulation of the central nucleus of the amygdala. Brain Research, 306, 1-8. Hawkins, R.D., & Kandel, E R. (1984). Is there a cell-biological alphabet for simple forms of learning? Psychological Review, .21, 375-391. Hebb, D.O (1949). The organization of behavior: A neuropsychological theory New York: Wiley Hitchcock, J.M., Sananes, C., & Davis, M. (1989). Sensitization of the startle reflex by footshock: Blockade by lesions of the central nucleus of the amygdala or its efferent pathway to the brainstem. Behavioral Neuroscience, 103, 509-518 Hodes R.L., Cook, E W III & Lang, P.J. (1985) Individual differences in autonomic response: Conditioned association or conditioned fear? Psychophysiology, 21, 545-560. Hoffman, H.S & Searle, J L. (1965). Acoustic and temporal factors in the evocation of startle. Journal of the Acoustical Society of America,~ 269-282. Hugdahl, K & Ohman, A (1977). Effects of instruction on acquisition and extinction of electrodermal responses to fear-relevant stimuli. Journal of Experimental Psychology : Human Learning and Memory,~ 608-618. Hull C L. (1943) Princi p l es of behavior New Y o rk : Appleton-Centur C r o ft s Hunt W A., & L a nd is, C. ( 1 9 3 6 ). T he ove rt b e h avior pattern in s t ar tl e Journa l of Ex p er i m e ntal P s yc h o l ogy, 12., 309-315

    PAGE 151

    144 Ison, J.R., & Hammond, G.R. (1971). Modification of the startle reflex in the rat by changes in the auditory and visual environments Journal of Comparative and Physiological Psychology, a 435-452. Kagan, J., Reznick, J.S., & Snidman, N. (1987). The physiology and psychology of behavioral inhibition in children Child Development, !ili, 1459-1473. Kamin, L.J. (1969). Predictability, surprise, attention and conditioning. In B.A. Campbell & RM. Church (Eds.), Punishment and aversive behavior (pp. 279-296). New York: Appleton-Century-Crofts. Kapp, B.S., Pascoe, J.P., & Bixler, M.A. (1984). The amygdala: A neuro anatomical systems approach to its contribution to aversive conditioning. In N. Butlers & LS. Squire (Eds.), Neurophysiology of memory (pp. 473-488). New York: Guilford Press. Kehoe, E.J. (1988). A layered network model of associative learning: Learning to learn and configuration. Psychological Review, 2_-2, 411-433. Konorski, J. (1967). Integrative activity of the brain : An interdisciplinary approach. Chicago, IL: University of Chicago Press Kurtz, K.H., & Siegel, A (1966) Conditioned fear and magnitude of startle response: A replication and extension Journal of Comparative and Physiological Psychology .6L 8-14 Kurtz, K.H., & Walters, G.C. (1962). The effects of prior fear experiences on an approach-avoidance conflict. Journal of Comparative and Physiological Psychology, 5~ 1075-1078. Lacey, J I., Kagan, J., Lacey, B.C & Moss, H.A. (1963) The visceral level: Situational determinants and behavioral correlates of autonomic response patterns. In P.H. Knapp (Ed.), Expression of emotions in man. (pp. 161-196). New York: International Universities Press. Lacey, J.I., & Lacey, B.C. (1970). Some autonomic-central nervous system relationships. In P Black (Ed.), Physiological correlates of emotion (pp 205-227). New York: Academic Press Lang, P J (1980). Behavioral treatment and bio-behavioral assessment: Computer application s. In J B Sidowski J H J o hnson & T.A. William s (Eds ), Technology in h ea lth care delivery s ys t e m s ( p p 11 9137) Norwood NJ: Abl ex.

    PAGE 152

    145 Lang, P.J., Bradley M M., & Cuthbert, B.N (1990). Emotion, attention, and the startle refle x. Psychological Review. 2Z 377-395 Lang, P J Bradley, M M., & Cuthbert, B.N (1992) A motivational analysis of emotion : Reflex-cortex connections. Psychological Science, .3., 44 49. Lang, P.J., Greenwald, M.K & Bradley, M M. (1988) The International Affective Picture System (IAPS) standardization procedure and initial group results for affective judgments : Technical reports lA-1D Gainesville, FL: Center for Research in Psychophysiology, University of Florida. Lang, P.J., Greenwald M.K Bradley, M.M., & Hamm, A.O. (in press). Looking at pictures : Affective, facial, visceral and behavioral reactions Psychophysiology Lang, P.J., Ohman, A ., & Vaitl, D. (1988) The International Affective Picture System (IAPS [photographic slides]). Gainesville, FL: Center for Research in Psychophysiology, University of Florida. Lanzetta, J T., & Orr S. (1986). Excitatory strength of expressive faces: Effects of happy and fear expressions and context on the extinction of a conditioned fear response Journal of Personality and Social Psychology, _2Q, 190-194 Leaton, RN., & Borszcz, G.S. (1985). Potentiated startle: Its relation to freezing and shock intensity in rats. Journal of Experimental Psychology : Animal Behavior Processes, 11, 421-428. LeDoux, J.E (1987 ). E motion. In V.B. Mountcastle, F. Plum, & St.R. Geiger (Eds.), Handbook of physiology Section 1 : The nervous system (Vol. 5, pp. 419-459). Bethesda, MD: American Physiological Association Levey, A B & Martin, I. (1981). Personality and conditioning. In H J Eysenck (Ed .), A model for personality (pp 123-168) New York: Springer-Verla g. Mackintosh, N.J. (1983) Conditioning and associative learning. Oxford : Oxford Uni versity Press. Marr, D. (1982) Vision. San Francisco: Freeman. McNally, R.J., & Reiss, S. (1984). The preparednes s theory of phobia s: The effects of initial fear level on safety-signal conditioning to fear-relevant stim uli. Psychophysiology. 21, 647 652.

    PAGE 153

    146 Mehrabian, A., & Russell, J.A. (1974). An approach to environmental psychology. Cambridge, MA: MIT Press Nebilytsin, V.D (1972). Fundamental properties of the human nervous system. New York: Plenum Press. Obrist, P. (1981). Cardiovascular psychophysiology: A perspective. New York: Plenum Press. Osgood, C., Sud, G., & Tannenbaum, P. (1957). The measurement of meaning. Urbana, IL: University of Illinois. Patrick, C.J., & Berthot, B.D. (1991). Potentiated startle during active and passive anticipation of an aversive stimulus. Psychophysiology, S43 (Abstract) Pavlov, LP (1957). Experimental psychology and other essays. New York: Philosophical Library. Pearce, J.M., & Hall, G. (1980) A model for Pavlovian conditioning: Variations in the effectiveness of conditioned but not unconditioned stimuli. Psychological Review. B.Z, 532-552 Pearl, J. (1963) The effects of preshock on and additional punishment on general activity Psychological Reports lb 155-161. Pearl, J Walters, G C. & Anderson, D.C ( 19 64) Suppressing effect s of aversive stimulation on subsequently punished behavior. Canadian Journal of Psychology, Ia 343-355. Petrinovich, L. (1984) A two-factor dual-process theory of habituation and sensitization In H V.S Peeke & L. Petrinov i ch (Eds.), Habituation, sensitization, and behavior (pp 17-55). New York: Academic Press Pivik, RT., Stelmack, RM., & Bylsma, F.W (1988) Personality and individual differences in spinal motoneuronal excitability Psychophysiology, Z5., 16-24. Prokasy, W.F., & Raskin D (1973). Electrodermal activity in psychological research New York: Academic Press P utnam LE. (1975 ) Th e human startl e reaction : Mec hani s m s of modification by back gr o und a c ou stic stimulation. Di s s erta t ion Abs t racts Int e rnational 3.6., 6419 B

    PAGE 154

    147 Putnam, L.E., Graham, F.K., & Sigafus, D.L. (1975) Calibration of earphone transmitted sounds. Psychophysiology. ll, 106-109. Randich, A., & LoLordo, V.M. (1979). Associative and nonassociative theories of the UCS preexposure phenomenon: Implications for Pavlovian conditioning. Psychological Bulletin. 523-548. Rescorla, R.A. (1971). Variations in the effectiveness of reinforcement and nonreinforcement following prior inhibitory conditioning. Learning and Motivation, 2.., 113-123. Rescorla, R.A., & Wagner, A R. (1972). A theory of Pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement. In A.H. Black & W.F. Prokasy (Eds.), Classical conditioning II: Current research and theory (pp. 64-99). New York: Appleton-Century-Crofts. Rosen, J.B., & Davis, M. (1988). Enhancement of electrically elicited startle by amygdaloid stimulation. Physiology and Behavior, .4.8., 343-349. Rosen, J.B., & Davis, M. (1990). Enhancement of the acoustic startle by electrical stimulation of the amygdala Behavioral Neuroscience, 102, 195-202 Ross, LE. (1961). Conditioned fear as a function of CS-UCS and probe stimulus intervals Journal of Experimental Psychology, fil., 265-273 Roth, W.T., Blowers, G H., Doyle, C M & Kopell, B.S. (1982) Auditory stimulus intensity effects on components of the late positive compl ex. Electroencephalography and Clinical Neurophysiology.~ 132-146 Roth, W T Dorato, K.H., & Kopell, B.S. (1984) Intensity and task effects on evoked physiological responses to noise bursts Psychophysiology, 21., 466-481. Rumelhart, D.E., Hinton, G E., & McClelland, J.L. (1986). A general framework for parallel distributed processing. In D.E Rumelhart, J L. McClelland, and the PDP processing research group (Eds.), Parallel distributed processing, Volume 1 (pp 45-76) Cambridge, MA : MIT press Sarter M ., & Marko w it s ch H J (1985). Involv e m e nt of the amygd a la in l ea rn i n g a nd m emo r y: A critical re v i ew, wi th e mpha s is on an a t omical r e l atio n s Be h avioral Neu ro sc i e nc e, 2.2., 342-380 Sc hl osberg, H (1952) The d esc r i pti o n of fac i a l exp r ess i o n in t e r ms of two dimension s. Iournal of Expe r imen t a l Psycho l ogy, 11, 229 237

    PAGE 155

    148 Schneirla, T.C. (1959). An evolutionary and developmental theory of biphasic processes underlying approach and withdrawal. Nebraska Symposium on Motivation (pp. 1-42). Lincoln, NB: University of Nebraska Press. Siegel, A. (1967). Stimulus generalization of a classically conditioned response along a temporal dimension. Journal of Comparative Physiological Psychology,~ 461-466. Spence, K.W. (1958). A theory of emotionally based drive (D) and its relation to performance in simple learning situations. American Psychologist. 1J, 131-141. Spence, K.W. (1964). Anxiety (drive) level and performance in eyelid conditioning. Psychological Bulletin, fil., 129-139. Spence, K.W., & Runquist, W.N. (1958). Temporal effects of conditioned fear on the eyelid reflex. Journal of Experimental Psychology,~ 613-616. Strayer, D., & Williams, W.C. (1981). SCORIT. Paper presented at the Twenty-First Annual Meeting of the Society for Psychophysiological Research, Washington, DC. Sutton, R.S., & Barto, A.G. (1981). Toward a modern theory of adaptive networks: Expectation and prediction. Psychological Review, .8.8., 135170. Szabo, I., & Kolta, P. (1967). Transitory increase of the acoustic startle reaction during its habituation. Acta Physiologica Academiae Scientarium Hungaricae, aL 51-56 Thompson, R.F., & Spencer, W.A. (1966). Habituation: A model phenomenon for the study of neuronal substrates of behavior. Psychological Review, za, 16-43. Tellegen, A. (1985). Structures of mood and personality and their relevance to assessing anxiety, with an emphasis on self-report. In A H. Tuma & J.D. Maser (Eds.), Anxiety and the anxiety disorders (pp. 681-706). Hillsdale, NJ: Erlbaum. Tischler, M D., & Davis, M (1983). A visual pathway that mediated fear conditioned enhancement of acoustic startl e. Brain Research, 276, 55-71.

    PAGE 156

    149 Trapold, M A (1962). The effect of incentive motivation on an unrelated reflex response. Journal of Comparative and Physiological Psychology, .25., 1034-1039 Vrana, S R., & Lang, P.J. (1990). Fear imagery and the startle-probe reflex Journal of Abnormal Psychology,~ 189-197. Vrana, S.R., Spence, E.L., & Lang, P.J. (1988). The startle probe response: A new measure of emotion? Journal of Abnormal Psychology, 97, 487-491. Wagner, A.R. (1978). Expectancies and the priming of STM In S H. Hulse H. Fowler, & W.K Honig (Eds.), Cognitive aspects of animal behavior (pp. 177-209) Hillsdale, NJ: Erlbaurn Wagner, A.R., & Brandon, S.E (1989). Evolution of a structured connectionist model of Pavlovian conditioning (AESOP). In S.B Klein & R.R. Mowrer (Eds.), Contemporary learning theories: Pavlovian conditioning and the status of traditional learning theory (pp. 149-189). Hillsdale, NJ: Erlbaum. Walters, G.C. (1963). Frequency and intensity of pre-shock experiences as determinants of fearfulness in an approach-avoidance conflict. Canadian Journal of Psychology. lZ, 412-419 Wolpe, J., & Lang, P J (1964). A fear survey schedule for use in behavior therapy Beh aviour Research & Therap y, 2., 27-30. Zuckerman, M. (1979). Sensation seeking : Beyond the optimum l evel of arousal. Hillsdale, NJ: Erlbaurn. Zuckerman, M ., Buchsbaum, M.S ., & Murphy, D.L. (1980). Sensation seeking and its biological correlates. Psychological Bulletin, 88., 187-214.

    PAGE 157

    BIOGRAPHICAL SKETCH Mark Kenneth Greenwald was born April 8, 1960, in Baltimore, Maryland. In 1983 he earned a Bachelor of Arts degree in psychology at Oberlin College in Ohio, and a Bachelor of Music degree in composition at Oberlin Conservatory of Music. In 1987, he received a Master of Science degree in clinical and health psychology from the University of Florida. He is presently a postdoctoral fellow in behavioral pharmacology at the Johns Hopkins University Department of Psychiatry and Behavioral Sciences. 1 50

    PAGE 158

    I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. -'1.ra S. Fischler, Cochairman Professor of Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree lll-Yt1't!b~1L...Lhilosophy. W. Keith Berg Professor of Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. ~32 Professor of Psycholo g y

    PAGE 159

    I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy !!_C/A1Uh ii a f,:mes H Cauraugh A~sociate Professor of Exercise and Sport Sciences This dissertation was submitted to the Graduate Faculty of the Department of Psychology in the College of Liberal Arts and Sciences and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August, 1992 Dean Graduate School

    PAGE 160

    UN IV ERS I TY OF FL OR I DA II I II IIIII I Ill I l l ll l ll l l lll I I I II I I I I II I I I llll l l ll 11 1 11 1 111 1 1 111 1 3 1262 08553 7610