Emotional imagery in high and low dentally fearful children


Material Information

Emotional imagery in high and low dentally fearful children a test of Lang's bio-informational theory
Physical Description:
viii, 175 leaves : ill. ; 29 cm.
Cohn, Lauren Kaplan, 1958-
Publication Date:


Subjects / Keywords:
Dental Stress Analysis   ( mesh )
Emotions -- Child   ( mesh )
Emotions -- Infant   ( mesh )
Pediatric Dentistry   ( mesh )
Clinical and Health Psychology thesis Ph.D   ( mesh )
Dissertations, Academic -- Clinical and Health Psychology -- UF   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1987.
Bibliography: leaves 168-174.
Statement of Responsibility:
by Lauren Kaplan Cohn.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000903088
oclc - 17889876
notis - AEL1995
System ID:

Full Text







To my parents, Doris and Irwin Kaplan, who taught me to

believe that I could do anything, and then gave me the love,

support and friendship I needed to do it.


I wish to thank my major professor, Dr. Barbara G. Melamed,

for her direction in the completion of this dissertation. Her support

and guidance during my graduate training are greatly

appreciated. Dr. Russell Bauer, Dr. Suzanne B. Johnson, Dr. Peter

Lang, and Dr. Marjorie White served with Dr. Melamed on my

doctoral committee, and I thank them for their time and for their

helpful comments. I am indebted to Robyn Ridley-Johnson, Ed Cook

and Lisa Pistone for their invaluable work on this study. Several

research assistants participated in this study. I would

particularly like to thank John Eisler, Jamie Goodman, Sanjiv

Patel, and Godard van Reede for their contributions. Thanks are

also due Paul Greenbaum and Rich Steinkohl for their help during

the final stages of manuscript preparation. I would like to express

my gratitude to Dr. Carroll Bennett, Dr. Frank Courts, Dr. Roy

Jerrell and Dr. Clara Turner who served as the dentists in this

study, and to the staff of the Pediatric Dentistry clinic at the

University of Florida for its assistance in this research. The

cooperation of the School Board of Alachua County and the parents

and children who participated in the study are thankfully


My parents, sister and grandparents played a special role in

the completion of this dissertation. Their enthusiasm about my work

and their pride in my accomplishments really helped me to achieve

my goal. I especially want to thank my husband, Alan, for his

helpful ideas and for the love and support he has given me.

The work for this dissertation was conducted while I was a

fellow on the National Institute of Dental Research Training Grant

No. 5 T32 DE07133-02.


ACKNOWLEDGEMENTS......................................... iii

ABSTRACT................................................... vii

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

Physiological Responsivity to Imagery................ 3
Lang's Bio-informational Theory ..................... 15
Responsivity to Internally Induced Imagery.......... 27
Imagery Ability in Children ......................... 30
Affect and Memory ................................... 43
Children's Responsivity to Imagined Stimuli......... 49
Children and the Bio-informational Theory........... 53
Statement of the Problem ............................ 56

METHOD...................................................... 59

Design ................................................. 59
Subjects .............................................. 60
Apparatus ............................................. 60
Procedure .............................................. 63
Screening .......................................... 63
Session 1 ........................................... 64
Session 2 ........................................... 67
Data Scoring ........................................... 69
Physiological Measures.............................. 69
Self-Report Measures ................................ 70
Observational Measures ............................. 71
Data Analysis .......................................... 72

RESULTS ..................................................... 74

Study I: Pilot Study of Scene Contents.................. 74
Pleasure ............................................ 75
Arousal ............................................. 75
Dominance........................................... 75
Fear............... ............................... 76
Similarity to Past Experience........................ 76
Similarity to Possible Future Experiences........... 76
Study II: Tests of Hypotheses .......................... 77
Subject Demographic Data ........................... 77
Imaginal Responses to Affective and Neutral Content 79
Effects of Fear Level and Imagery Training........ 86

DISCUSSION................................................... 94
Discrimination of Affective Valence ...................... 94
The Effects of Fear Level and Imagery Training....... 97
Implications for Lang's Bio-informational Theory....... 99
Methodological Considerations........................... 102
Suggestions for Future Research........................ 104
Conclusions............................................ 106


A IMAGERY SCENES....................................... 109

DENTAL SUBSCALE.................................... 112


D SUBJECT DATA FORM................................... 119

E CHILDREN'S FEAR SURVEY SCHEDULE................... 120

F DENTIST RATING FORM................................. 123

G BEHAVIOR PROFILE RATING SCALE..................... 124


APPPR :;iEiEJSION..................................... 130

J RELAXATION TRAINING ................................. 131

K STIMULUS IMAGERY TRAINING.......................... 136

L RESPONSE IMAGERY TRAINING.......................... 140


N PILOT STUDY OF SCENE CONTENTS..................... 153


REFERENCES .................................................. 168

BIOGRAPHICAL SKETCH ........................................ 175

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



Lauren Kaplan Cohn

May, 1987

Chairman: Barbara G. Melamed
Major Department: Clinical Psychology

In a test of Lang's bio-informational theory, physiological

responses to emotional imagery were studied in 6- to 12-year-old

children who reported high or low dental fear. In the first

experimental session, each child was given a restorative dental

treatment. Next, each child was given a relaxation exercise and

then trained in one of two kinds of imagery. Stimulus training

involved teaching children to focus on descriptive aspects of the

imagined situations, while response training instructed children to

focus on their bodily responses during imagery. Measures of social

anxiety and imagery ability were also taken. In the second

session, the relaxation exercise was repeated, and then

physiological recordings were taken while children imagined

neutral, dental fear and school fear situations. Children rated

their emotional responses to each scene. This assessment was

followed by a second dental treatment. Both treatment sessions were

videotaped for later scoring of disruptive behaviors. It was found

that children do discriminate between fearful and neutral content


both in their physiology and in affective ratings. Children showed

heart rate acceleration to fearful content and deceleration to

neutral content. Fear scenes were rated as lower in pleasure and

in perceived control than were the neutral scenes. The study

revealed that for children, the imagery task requires two steps of

processing. In the first step, children attend to the material being

presented and show a physiology concordant with attention. In the

second step, once the emotional memory network has been accessed,

children respond to the affective content of the material, showing

a physiology that varies with the affective valence of the text.



Fears and phobias in children are generally held to be fairly

common occurrences. Estimates suggest that approximately 40% of

children ages 6 to 12 years have as many as seven fears (Lapouse

& Monk, 1959; Miller, 1983; Miller, Barrett, Hampe & Noble, 1972).

Children are often referred to clinics for the treatment of their

fears, although estimates of the frequency of referrals vary from 6

to 8% (Graziano & DeGiovanni, 1979) to as much as 30-40% (Miller,

1983). Many of the therapies used with fearful children involve the

use of imagery. These treatments include hypnosis, systematic

desensitization, implosive therapy and flooding. For example,

Lazarus and Abramovitz (1962) reported on the use of "emotive

imagery" with fearful children, ages 7-14 years. This treatment is

quite similar to systematic desensitization used with adults;

however, the child's own hero or alter ego is incorporated into the

stories. This procedure was successful in ameliorating the fears of

seven of nine children treated.

Although numerous studies attest to the effectiveness of

imaginal treatment with adults, there is a relative paucity of

empirical data on the utility of these techniques with children

(Elliot & Ozolins, 1983; Graziano, DeGiovanni & Garcia, 1979;

Ollendick & Cerny, 1981). Many of the procedures used in the

treatment of adults have been applied to children with little

modification (Elliot & Ozolins, 1983; Graziano, DeGiovanni &

Garcia, 1979; Harris, 1983). However, this may not be appropriate.

Several important issues must be considered when using imagery

based treatments with children. The most obvious of these is the

developmental level of the child (Elliot & Ozolins, 1983; Rosenstiel

& Scott, 1977). There is some evidence to suggest that children as

young as 4 to 6 years old can use imagery to change their

behavior (Mischel, Ebbesen & Zeiss, 1972). It cannot be assumed,

however, that children are uniformly able to use imagery in

therapy. For example, Tasto (1969) was unsuccessful in his attempt

to use imaginal desensitization to treat a 4-year-old boy with a

fear of loud noises. In vivo desensitization did result in

elimination of the child's symptoms. Furthermore, there are data to

suggest that important changes in imagery ability occur between

the ages of 6 to 8 years (Paivio, 1970; Reese, 1970; Rohwer, 1970).

This would coincide with movement into Piaget's stage of concrete


The purpose of this paper is to describe a study of children's

responses to imagery. The study addresses factors that have been

shown to be important to therapeutic improvement in adults. Three

lines of research are reviewed. First, in the absence of adequate

research on imaginal therapies with children, studies on adult's

responses to imagery are considered. In this context, Lang's

bio-informational theory of emotional imagery will be described.

This theory has provided useful information regarding variables

that have influenced improvement in adult phobic patients through

the use of imaginal treatments (Lang, 1977, 1979). Second,

research on imagery ability in children will be described. Work to

be described comes primarily from the area of learning and

memory, since, as noted above, the clinical literature has

produced few studies of imagery in children. Third, studies of

children's responsivity to imagery will be presented.

Physiological Responsivity to Imagery

The use of systematic desensitization is based on the

assumption that visualized stimuli produce fear responses that are

similar to, but less intense than, those produced by the actual

situation. Relaxation is then elicited, in order to inhibit or

extinguish this fear response (Grossberg & Wilson, 1968).

Systematic desensitization and procedures like it have

stimulated a great deal of research on the physiological effects of

imagery. Specifically, investigators have questioned whether

imagined stimuli actually do evoke a fear response, whether the

response varies with the fearfulness of the stimulus, and whether

the pattern of responsivity seen in phobic subjects differs from

that seen in non-phobic subjects.

Grossberg and Wilson (1968) designed an experiment to

determine whether fearful imagery truly does generate more "initial

tension" than neutral imagery. The authors suggest that such a

difference is basic to the theory underlying systematic

desensitization, and note that if no difference were found, "it

would be difficult to maintain that something was being inhibited,

replaced or extinguished with repeated visualizations" (Grossberg &

Wilson, 1968, p. 124).

The investigators compared the physiological responses of high

and low anxious female undergraduates to fearful and neutral

scenes. Fear scenes were individualized for each experimental

subject. A control group listened to tape recorded presentations of

the scenes selected for her matched experimental counterpart.

Control subjects reported low fear for the items tested.

The authors found that both high and low anxious subjects

showed increases in heart rate and skin conductance during both

the reading and visualization of the scenes. Fear scenes produced

greater arousal than neutral scenes during the image period only.

Both heart rate and skin conductance decreased over repeated

scene readings, while skin conductance also showed a decrease

over imaging trials.

Comparison of experimental and control subjects indicated that

the experimental subjects showed a greater heart rate increase

during the reading of the fearful scenes. Imagining fear scenes

resulted in significant heart rate and skin conductance increases

to fear scenes for both groups, although no difference between

scenes had been expected for the controls. The experimental group

showed significant habituation of the heart rate response over

trials for the reading of fear scenes. Both groups showed skin

conductance habituation for the presentation of neutral scenes.

The authors concluded that imagery does have measurable

physiological effects, and that fearful imagery produced more

arousal than neutral imagery. Some habituation was noted over

repetitions, suggesting that habituation can occur even without

relaxation training. They note that these results were obtained

with a non-phobic population and suggest that phobic subjects

would be likely to produce greater differential responding.

Haney and Euse (1976) examined physiological responsivity to

positive scenes as well as to negative (fearful) and neutral scenes

in a sample of college students. They found that during imagery,

heart rate was larger Efc the negative scenes than for the positive

or neutral scenes. Skin conductance responses were larger for both

the positive and ne-ative scenes than for the neutral scenes, while

the former were not significantly different from each other. There

was a significant decrease in skin conductance across time for the

neutral image.

Positive images were rated as more intense and higher in

clarity than either the neutral or negative images. Negative images

were rated as more intense than neutral images. The relationship

between clarity, intensity and physiological responsivity was not


Results of this study are similar to those obtained by Grossberg

and Wilson (1968), since fearful scenes produced increases in heart

rate and skin conductance. It is not clear, however, why the

positive scenes in the Haney and Euse (1976) study also produced

increases in skin conductance. The authors did not specify the

content of the positive scenes. Thus, it is not possible to

determine whether the scenes were of an arousing nature (i.e.,

physical exertion) which may have resulted in increased skin

conductance responding.

The studies described above yielded information about

physiological responsivity to imagery that has relevance to the

theory of systematic desensitization; however, subjects in these

studies were all non-phobic. Van Egeren, Feather, and Hein (1971)

studied a group of speech phobic college students in order to

evaluate a number of hypotheses about systematic desensitization.

Half of the subjects (all male) were given instruction in

relaxation, and all subjects received visual imagery training. The

details of this imagery training are not specified. The authors

note only that "his visualization procedure was discussed and

suggestions made, emphasizing vividness of the imagery, constancy

of the image, and realistic projection of the self into the

situation" (Van Egeren et al., 1971, p. 215). Each subject

imagined the scenes on his individual hierarchy during the next

two experimental sessions.

Fearful imagery was associated with significant autonomic

responses; specifically, subjects showed increases in heart rate,

skin conductance, respiration, and digital vasoconstriction. These

responses were greater than those occurring to neutral stimuli. The

investigators found a positive relationship between the degree of

responsivity and the fear level of the scene, although this was

true only for digital vasoconstriction and the number and

magnitude of skin conductance responses.

Habituation of responses with repetition of stimuli was observed

for digital vasoconstriction both within and across sessions.

VanEgeren et al. (1971) expected that the least threatening scenes

would lead to more rapid extinction of responses; however, this

hypothesis was not supported by the data.

Relaxation had limited effects on physiological responding.

Relaxed subjects showed a faster habituation of digital

vasoconstriction throughout the experiment and of skin conductance

responses during Session 1. It was also found that relaxation

produced more facilitation of habituation (skin conductance

response and vasoconstriction) for the most threatening scenes. An

interesting result obtained in this study was that decreases in

autonomic responding were not accompanied by comparable

decreases in the subjective reports of anxiety. Thus, these subjects

may have shown improvement in their anxiety reactions, but yet,

may not "feel better." This discordance between physiological,

subjective, and behavioral indices of anxiety has been described

by Lang (1968).

The study by Van Egeren et al.(1971) addressed several

important points. First, the authors found some evidence that

greater levels of physiological arousal occur to scenes rated higher

in fearfulness. They showed that physiological habituation does

occur over repeated presentations of fear stimuli and that

relaxation is not absolutely essential for habituation to take

place. Finally, the finding of discordance between self-report and

physiological measures of anxiety is a common one in the literature

(Lang, 1968) and points to the importance of comprehensive

assessment of anxiety.

Similar findings were noted by Marks and Huson (1973) in their

summary of six treatment studies that included a physiological

assessment of imagery ability in phobic patients. Responses to

scenes with phobic and neutral content were compared. Patients

exhibited increases in heart rate and skin conductance to phobic.

scenes, although these results were not obtained consistently in all

six studies. Subjective ratings of anxiety did discriminate between

fearful and neutral scenes in all studies. After the completion of

treatment, differences in responsivity to phobic and neutral scenes

were decreased. Interestingly patients in four of the studies

continued to rate the phobic scenes as more anxiety provoking.

Few correlations between the physiological measures or between

physiological and subjective measures were obtained, providing

additional support for the notion that there may be some

independence between response systems in phobic patients.

Waters and McDonald (1973) noted that the type of stimulus

presented to subjects may influence the degree of fear decrement. A

number of studies had indicated that exposure to the feared

stimulus in vivo results in greater autonomic reactivity and more

improvement in phobic concerns. Thus, Waters and McDonald

examined autonomic reactivity and differences in response

decrement to fear provoking stimuli in three modalities. Rat phobic

undergraduate females were given systematic desensitization. At

each level of the hierarchy subjects heard a description of the

fear stimulus (auditory mode), saw a slide depicting the feared

stimulus (visual mode), and then closed their eyes and imagined

the stimulus (imagery mode). Each subject was exposed to each

modality in this between subjects design. Imagined stimuli

produced significantly more responsivity than either the visual or

auditory stimuli in the case of heart rate, skin resistance and

skin potential. Vasomotor response showed more responsivity to the

auditory stimuli. There were no significant differences between

stimulus modalities in the degree of response decrement over trials,

with two exceptions. Imagery produced greater decreases in skin

potential than did auditory presentation, while skin resistance

showed greater decrease to visual than auditory stimuli. When

response decrements did occur over trials, these occurred more

frequently to visual and imagined stimuli. No habituation was

produced for heart rate, and only one stimulus produced

habituation of skin resistance and skin potential. The authors

concluded that exposure to stimuli in either the visual or imaginal

mode would be sufficient to produce a decrease in the physiological

component of phobic disorders.

There is a serious confound in this study. Subjects all received

exposure to the three stimulus modalities, and all in the same

order. It is very possible that the increased responsivity to the

visual and imaginal stimuli was a result of an incremental effect

of these modalities when given subsequent to the auditory

modality. In fact, the authors note that a "small minority of

subjects indicated that their imagery occasionally was an

exaggerated form of the auditory and visual images. Effects of

each modality need to be tested separately.

In a study designed to evaluate the effectiveness of a device

for automated desensitization, Lang, Melamed and Hart (1970)

presented some interesting data on patterns of physiological

responsivity over the course of treatment. Snake phobic

undergraduate women were given eleven sessions of systematic

desensitization with the automated device. The authors examined

physiological responses across all sessions on those trials during

which the subject signaled fear. These trials were compared with

the one previous trial and two subsequent trials during which no

fear was indicated. Consistent with prediction, subjects showed

significantly greater heart rate, skin conductance and respiration

on trials during which fear was reported. Further, those subjects

manifesting the greatest improvement in the degree of fearfulness

after treatment were those who demonstrated the highest heart rates

on fear signaled trials. These same subjects showed greater heart

rate habituation to repeated presentations of a fearful stimulus.

Respiration and skin conductance data did not yield a similar

pattern of results.

In a second study, Lang et al. (1970) investigated the nature

of an anxiety hierarchy. Specifically, the authors questioned

whether the degree of anxiety manifested physiologically and in

verbal report would correspond to the ranking of each item in the

hierarchy. The relationships between image clarity, hierarchy

position and anxiety were also considered. Public speaking and

spider phobic subjects participated in this study. Scenes were

presented in a random order so as to control for any effect of

sequential presentation of the scenes on responses. There was no

significant relationshi-- between hierarchy rank and image

vividness. Subjective ratings of anxiety were greater for items

higher in the hierarchy. Heart rate and skin conductance yielded

the predicted result: responsivity on these measures was greater

for those items ranked as highly fearful. A significant linear

trend was obtained for heart rate for both groups of subjects,

while only the spider phobics produced such a trend for skin

conductance. No relationship between respiration responses and

hierarchy rank was obtained. Both groups evidenced a significant

relationship between heart rate increments and anxiety ratings.

Vividness correlated with heart rate changes for the combined

groups and for the public speaking phobics alone, but not for the

spider phobics.

The studies by Lang et al. (1970) complement the results of

studies already described. It had previously been shown that both

normal and phobic subjects produce more physiological and

self-reported anxiety to phobic content than neutral content. These

investigators demonstrated that scenes rated as higher or lower in

fearfulness produce differential physiological responses. Further, it

was found that habituation of heart rate responding was associated

with a favorable treatment outcome. This then suggests that

habituation is a mechanism that may underlie the effectiveness of

this therapeutic approach. The study also provided support for the

notion of a fear hierarchy, in that greater physiological

responsivity and greater subjective anxiety occurred to those items

with higher ranks.

In a thought provoking paper, Grayson (1982) described two

studies that could provide an explanatory model for differential

responsiveness to phobic and neutral stimuli, as well as for the

mechanism underlying systematic desensitization. Grayson postulated

that a stimulus that was paired with a phobic stimulus would

elicit defensive responses in fearful subjects. He noted that it

would be necessary to study second-by-second heart rate responses

in order to evaluate this phenomenon.

In the first experiment described, Grayson presented speech

anxious subjects with a slide of either a phobic or neutral

stimulus, and then asked subjects to form a visual image of that

stimulus. For one group of subjects, each slide was preceded by

the presentation of a tone: high frequencies for phobic scenes, and

low frequency tones for neutral scenes. Another group of subjects

had no signal preceding the slides. For the fear stimuli, heart

rate wave forms characteristic of a defensive response occurred to

presentation of the signal and of the slide for only those subjects

receiving a signal. Defensive responses occurred to visualization of

fear scenes regardless of whether a signal was given. Contrary to

expectation, orienting responses were not obtained when neutral

stimuli were presented. These data support Grayson's notion that

that the increased heart rates observed when subjects imagine

fearful stimuli are manifestations of a defensive response.

Grayson's second experiment investigated two models that could

explain the incremental stimulus intensity effect. This effect is "a

demonstration that habituation to repeated intense stimulation is

more rapid when the stimuli are presented with gradually

increasing intensity than when an equal number of presentations of

the most intense stimulus are given" (Grayson, 1982, p.104).

Grayson presented subjects with a hierarchy of phobic stimuli (the

incremental series) or with a constant series which consisted of

repetitions of the stimulus item at the high end of the hierarchy.

Results of the study were expected to provide support for one of

two theories. Groves and Thompson's dual process theory predicts

that greater habituation would occur with an incremental series

since any sensitization or arousal would be small and would

habituate over a few trials. A constant series would be expected to

produce marked, lasting sensitization. Sokolov's neuronal

comparator model, however, would predict that greater habituation

would occur to the constant series, assuming that presentation of a

new stimulus could elicit an orienting response, or possibly even a

defensive response.

Speech anxious college students were presented with either four

presentations each of four graded fearful stimuli (incremental

series condition), or sixteen presentations of the most fearful

stimulus (constant series condition). Comparisons were made on

responses to the last four presentations, since these were identical

for both groups. The method of stimulus presentation was the same

as in the first study.

Some support was obtained for the use of graded hierarchies,

since the constant series produced larger and more durable heart

rate accelerations in the pre-slide period and a greater number of

spontaneous fluctuations in skin conductance during visualization.

Clear empirical support was not obtained for either the dual

process or neuronal comparator model. Both experimental groups

evidenced heart rate wave forms and cephalic vasomotor

constriction characteristic of a defensive response. Nevertheless,

this study is an important one, since it points to the defensive

response as a potential explanation of the increased physiological

responsivity that has been observed to occur to phobic stimuli.

Lang's Bio-Informational Theory

In his bio-informational theory, Lang (1977, 1979, 1985)

suggests that images are not pictures in the mind's eye, but

rather that images are the product of a propositional structure in

memory. An image consists of information units, or propositions,

that are linked together into an informational network. There are

stimulus propositions, which refer to the descriptive features of the

situation, response propositions, which refer to the subject's

behavior in the situation, and meaning propositions, which refer to

the subject's appraisal of the stimulus and response data (Lang,

1985). There are three types of response propositions: overt motor

behavior (i.e., I run away), verbal report of subjective

experience (i.e., I feel afraid), and patterns of visceral

reactivity and somatomotor tonus (i.e., my heart pounds). The

network is processed as a unit when a critical number of the

propositions are accessed. Propositions can be accessed in a

variety of ways, such as through actual exposure to the feared

stimulus, through pictures or stories of the stimulus, and even

through thoughts. Since the different propositions are linked

together in memory, accessing, or activating one proposition in the

network increases the probability that a proposition linked to it

will also be accessed via a spread of activation (Bower, 1981).

Thus, if a person imagines a snake, remembers that he/she is

afraid of snakes and typically runs away from them, then

physiological responses associated with this fear behavior (i.e.,

increased heart rate) are likely to be evoked as well.

The bio-informational theory has important implications for the

treatment of phobic disorders. It has been shown that treatments

such as systematic desensitization can decrease physiological

responding to phobic stimuli (Lang, Melamed & Hart, 1970; Marks &

Huson, 1973). Lang posulates that these treatments work by

weakening the associations between propositions, such as "SNAKE"

and "MY HEART BEATS FASTER." Foa and Kozak (1986) suggest three

mechanisms for these effects. First, short term habituation results

in a dissociation of responses from stimulus propositions. Second,

there is a change in the meaning of the stimulus, and third there

is long term habituation.

According to Lang (1977; Lang et al., 1980) it is important

that the images formed be as vivid as possible if they are to have

this therapeutic value. Vividness is defined as completeness of the

evoked propositional structure. One way to maximize image

vividness is to train subjects to include response propositions in

the images that they form. This is expected, then, to lead to

enhanced physiological responsivity to imagined stimuli. Recall

that Lang et al. (1970) found maximal therapeutic improvement in

those subjects who showed the largest heart rate increases in

response to fearful images.

To test this hypothesis, Lang et al. (1980) trained one group

of subjects to attend to specific stimulus details of the imagined

scene (the stimulus-trained group), and the other group of subjects

to focus on their physiological responses during imagery (the

response-trained group). Subjects were then asked to imagine a

series of neutral, active and fearful scenes. For both groups,

neutral scenes contained only stimulus propositions. Fearful scenes

differed for the two groups. Scenes for the stimulus group

contained only stimulus propositions, while scenes for the response

group contained both stimulus and response propositions.

Physiological responses were examined at three points during each

trial: during the reading of the scene (READ), while subjects

imagined the scene (IMAGE), and while subjects relaxed after

imagining the scene (RECOVER).

Both groups showed minimal responsivity to neutral scenes.

However, a training effect was obtained for the fearful and neutral

scenes. Response subjects showed significant increases in heart

rate, respiration, and muscle tension during the image period.

This group also showed a marginally significant increase in eye

movement. The pattern of responsivity for the response group was

an inverted V: there was a slight increase in responding during

the read period, a further increase during the image period, and

a decrease during the recover period. Response curves for the

stimulus-trained subjects were essentially flat. Skin conductance

showed a slight tendency to decrease during imagery trials,

although all subjects showed less of a decline during fear scenes.

The distribution of responses during a given scene seemed to

mirror the content of the script. Thus, muscle tension responding

was greater for action scenes than for fear scenes, while the

reverse was true for respiration.

Lang et al. (1980) noted that in this experiment there was a

confound between training and script. In other words,

stimulus-trained subjects were tested on stimulus scripts, and

response-trained subjects were tested on response scripts. This

made it impossible to determine whether the observed responses

were a function of training, script, or both. Thus, the

investigators did a second study in order to replicate and extend

their findings. Four groups of subjects were used in a 2 (Stimulus

vs. Response Training) X 2 (Stimulus vs. Response Script) design.

The Stimulus-Stimulus and Response-Response groups provided for

replication, while the remaining two groups permitted study of the

independent effects of training and script. A second change was

made in the experimental design. Response propositions were

distributed equally across scenes so that, for example, action

scenes would not be confounded with muscle tension propositions,

or fear scenes with heart rate response propositions. Thus, each

scene contained one response proposition from each physiological


Replication of the findings of the first study was obtained for

the Stimulus-Stimulus and Response-Response groups. Examination of

all four groups indicated that response propositions produced

significant increases in responsivity only if both response training

and response scripts were combined. Stimulus-Response and

Response-Stimulus groups had patterns of responsivity very much

like those seen for the Stimulus-Stimulus group.

As was seen in the first experiment, muscle tension responses

were greater for action scenes than for fear scenes. Respiration

was slightly higher for fear scenes. Lang and his colleagues

(Lang et al., 1980; Lang et al. 1983) note that the finding of

differential reactivity of individual response systems in spite of

the equal distribution of response propositions across scene types

suggests that the imagery response is not just a function of the

script. Rather, it is a function of propositional elements relevant

to the situation that were accessed from the subject's own long

term memory.

Robinson and Reading (1985) studied small animal phobics in

order to examine the effects of imagery training and script content

on physiological and subjective arousal, concordance between

physiological systems, and habituation of responding over repeated

presentations of fearful stimuli. They compared stimulus and

response trained subjects after one presentation and after four

successive presentations of fear scenes. They found that response

trained subjects did show greater physiological arousal than

stimulus trained subjects for muscle tension, heart rate and skin

conductance, but not for finger temperature. Reported arousal was

significantly correlated with image vividness for both groups. For

the response trained subjects arousal and vividness were correlated

with muscle tension and heart rate, but not with skin conductance.

Muscle tension and heart rate were correlated for the response

group only. Finally, there was some tentative support for the

notion that response trained subjects show slower habituation. This

was not a clear finding however, because stimulus trained subjects

showed a pattern of small and inconsistent responding.

Bauer and Craighead (1979) investigated the effects of

instructions on physiological responses to fearful and neutral

imagery. In a two-way factorial design, subjects were given one of

two instructional sets (focus on scene details versus focus on

bodily reactions to the scene) and one of two orientation sets

(imagine actually participating in the scene versus imagine

viewing it as an observer). Imagery training was not given.

Scenes used were selected for each subject on the basis of

responses to a fear questionnaire. Fearful and non-fearful

(neutral) scenes were presented.

As was expected, fearful scenes produced greater heart rate

liability and marginally greater skin conductance responses than

did neutral scenes. A marginally significant main effect for

orientation was obtained for skin conductance, with participant

subjects showing greater reactivity than those imagining themselves

to be observers. Attentional focus produced significantly different

heart rate liability scores for the two groups; consistent with the

findings of Lang et al. (1980), subjects focusing on their bodily

reactions to the stimuli manifested greater heart rate changes.

Finger pulse volume did not differentiate between any of the

conditions. Although Lang (1977; Lang et al., 1980) suggests that

response training serves to "amplify" physiological responses so

that they can be detected, results of this study suggest that

training may not be necessary to produce these responses. A

problem in the Bauer and Craighead study is that there was no

check on the actual images generated by each subject, particularly

those in the body focus group. Such a check would be important

because the scenes presented to the subjects contained only

stimulus propositions. It would be helpful to know whether the

subjects had a tendency to include response propositions only from

certain systems in their images. Failure to find consistency

between physiological measures may have resulted from

inter-subject variation on the number and type of bodily responses


The actual content of images was assessed in a study by

Anderson and Borkovec (1980). These investigators presented speech

anxious subjects with either stimulus or stimulus plus response

imagery scripts. After the first scene presentation, subjects were

asked to describe their images. This permitted scoring of scene

content, in terms of stimulus and response propositions. Following

four more presentations of each scene, subjects again described the

content of each image. Physiological assessment was conducted

during each scene presentation.

The authors found that greater response detail was reported by

subjects in the stimulus plus response group. However, both groups

of subjects reported a decreased number of stimulus details with a

concomitant increase in response details over the course of the

study. Interestingly, the amount of response detail reported was

significantly correlated with both heart rate responses and

post-scene fear ratings. This finding provides additional support

for the notion that imagining response propositions is associated

with greater physiological responding. Anderson and Borkovec note,

however, that the two script conditions did not provide clear

differentiation between reported images and warn investigators that

it is important to insure that the manipulation is successful in

providing different conditions for each group. Repeated

presentations of the scenes led to a steady decline in heart rate

responding for the stimulus group and a more variable decline in

responding for the stimulus plus response group.

Carroll, Marzillier, and Merian (1982) raised a number of

questions about the effects of response propositions on

physiological responses to imagery. Specifically, these researchers

questioned the necessity of response training, the specificity of

response propositions for each of the response systems, and the

effect of response propositions for relaxing imagery.

Non-phobic adults were used as subjects in this study. Half the

subjects imagined scenes containing stimulus propositions; the other

half imagined scenes containing response propositions. Imagery

training was not given. Three scenes for each group were

arousing, exciting, or anxiety-provoking, while three scenes were

either relaxing or tranquil. The authors do not specify the exact

content of each scene. Each scene presented to the response group

emphasized responses in only one response system: either cardiac,

respiratory, or electrodermal activity. Each scene was imagined


The response group showed greater heart rate and respiration

responses (but not skin resistance) to the arousing scenes than did

the stimulus group. This finding is consistent with previous

research. The groups did not differ in responses to relaxing

scenes. Thus, response propositions do not seem to have an effect

on relaxation.

An interesting result in the study was that there was a lack of

specificity in the types of physiological responding occurring to

scripts emphasizing a particular response system. Thus, heart rate

increases were produced to scripts focusing on cardiac responses

as well as to those emphasizing respiratory activity. Similar

findings were obtained for respiratory and electrodermal responses.

Thus, as was noted by Lang et al. (1980; Lang et al., 1983) the

content of the script itself is likely to play a role in determining

which response systems are involved in responses to imagery, by

accessing the subjects' own propositional network about such

events. This network, then, may have an influence beyond that of

the specific response propositions included in the script.

Unfortunately, Carroll et al. (1982) did not elaborate upon the

content of their scripts, making it difficult to examine the effects

of specific contents.

Lang, Levin, Miller and Kozak (1983) conducted an extensive

investigation of aspects of the bio-informational theory in speech

anxious and snake phobic subjects. The study addressed two

important issues. First, the study examined the specificity of

responses of phobic subjects. That is, whether phobics respond to

all fearful contents in the same way or discriminate between their

own fear content and other fear contents. The second question

considered in the study was the similarity of response physiologies

for actual tasks and imagery of the tasks.

Snake phobic and speech anxious subjects participated in the

study. Each group underwent physiological assessment during

imagery and during actual exposure to the two feared situations.

Snake phobic subjects showed visceral responses to scenes depicting

both snake and speech content. The largest heart rate change

occurred to the snake scene. Positive skin conductance responses

were obtained to the snake scene as well as to a scene describing

physical exercise. Speech anxious subjects, however, showed heart

rate increases only to public speaking scenes. Additionally, these

subjects showed less skin conductance habituation to the speech

scenes than to the snake scenes. Physiological responses during in

vivo exposure mirrored these results. None of these comparisons

was statistically significant.

Responsivity to imagery was assessed a second time, after

subjects had been exposed to fear stimuli in vivo. It was assumed

that subjects would have stronger memories of both their own and

the opposite fear stimuli following exposure. The pattern of results

obtained was the same as that for pre-exposure imagery; again,

none of the comparisons were statistically significant.

In a second study, Lang et al. (1983) investigated the effects

of stimulus and response training on a new sample of speech

anxious and snake phobic subjects. Half of the subjects in each

fear group were given stimulus training, while the remainder of

each group received response training. All subjects were tested on

scripts that contained both stimulus and response propositions.

An analysis of the effect of scene content yielded a finding

consistent with those obtained in Study 1. The snake phobic

subjects showed the highest heart rate response to snake scenes,

but showed a heart rate response to the speech scene comparable

to that seen in the speech anxious subjects. This was a significant


Imagery training led to effects similar to those described in

earlier studies (Lang et al., 1980). Response trained subjects

showed the inverted V pattern, while stimulus trained subjects

showed minimal responsivity to imagery. Patterns of physiological

reactivity for response trained subjects were the same as those

obtained in Study 1. That is, snake phobics showed increased

physiological responsivity to both fear scenes, while speech

phobics responded only to their relevant fear scene. Stimulus

trained subjects showed the same pattern of responses; however,

the differences between group and scene content were greatly

accentuated in response trained subjects.

Results of the Lang et al. (1983) studies provide support for

earlier findings on the effects of stimulus and response training on

imagery. These studies also indicate that there is some specificity

in the responses of phobic subjects. Speech anxious subjects did

not respond in a phobic manner during snake imagery. The fact

that snake phobic subjects responded with increased physiology to

the speech scenes most likely represents the task demands of

giving a speech. This is consistent with the study's findings of

similarity of responding during imaginal and in vivo exposure.

The findings of this study support the hypothesis that response

information is stored within a propositional network. Even though

the majority of the snake phobic subjects did not report speech

anxiety, their propositional network for giving speeches still

contains inform ation about physiological responding that

accompanies this task. This raises the question of what differences

exist between propositional networks for phobics and non-phobics.

Lang et al. (1983) suggest that phobia networks are higher in

coherence than other networks. That is, the links between nodes in

a phobia network may be stronger. According to the authors, these

"emotion prototypes" require fewer matches in order to be

activated. This is supported by the finding of response sterotypy

in this study. That is, the subjects reporting the highest degree

of fear on the snake questionnaire were more likely to score high

on performance, other verbal and physiological measures of fear,

both in exposure and imagery.

Responsivity to Internally Induced Imagery

It is important to consider the impact that internally induced

thoughts and images can have on phobic patients. Phobics often

ruminate about fear producing situations. If phobia networks are

especially coherent, as is predicted by the bio-inform national

theory, then it is likely that even thoughts about feared situations

can evoke the fear network and produce the full pattern of

physiological responses to the stimulus. Physiological changes in

response to such thoughts and images are likely to reinforce

perceptions of the situation as a fearful one. Further, this

ruminative thinking and the associated fear reaction may result in

continued avoidance of the phobic situation, thus preventing the

phobic individual from obtaining information about the situation

that could serve to disconfirm his or her notions about its danger

(May, 1977b)

Schwartz (1971) noted that in studies such as those described

thus far, it is not possible to determine whether observed

responses were a function of the external stimulus, the

hypothesized mental processing, or both. Schwartz developed a

methodology that allows precise measurement of responses to

internal stimuli generated by the subject, that are under his or

her control. A series of tones is presented to the subject at

regular intervals. The subject is asked to repeat a series of

numbers (i.e., 1, 2, 3, 4), one per tone. He or she is given the

option of repeating the series of numbers one, two or three times.

After the number series is completed the subject generates one of

two thought sequences in synchrony with the tones. Thought

sequences are memorized prior to the start of the experiment. Some

studies using the Schwartz paradigm instruct subjects to form

images to each thought sequence; again, in synchrony with the

tones. Subjects press a button to signal the end of a trial. Data

can be analyzed by evaluating responses occurring after the last

number sequence. Thus, by choosing how many times to repeat the

number sequence, the subject controls the onset of the experimental

thought sequences.

Using this time-locked imagery, Schwartz (1971) found that

subjects showed heart rate accelerations when thinking of affective

stimuli (words such as sex, rape and death) but not to neutral

stimuli (letters, such as A, B, C). A problem with this study is

that the word stimuli were more likely to be meaningful and

perhaps to produce more vivid imagery than would simple letter

stimuli. May and Johnson (1973), using the same paradigm, found

that subjects showed greater heart rate and respiration when

imagining arousing words than when imagining neutral words.

Relaxing words (such as peaceful and tranquil) did not produce

the predicted heart rate decelerations.

The responses of phobic and non-phobic individuals were

compared in a study by May (1977b). Using the time-locked

procedure, subjects were told to think of a number series, then a

sentence (snake versus non-snake content), and then to form an

image to that sentence. Phobic subjects showed greater heart rate

responses to both the sentence and to the image, with the greatest

increases occurring to the snake stimuli. There was a significant

heart rate habituation over trials, although phobic subjects did

not return to baseline. Phobic subjects also showed greater

respiration amplitude when producing an image of a snake.

Finally, phobics showed greater skin conductance level increases

and more frequent ski- conductance responses than non-phobics

during snake imagery. This study indicates that phobic subjects do

generate autonomic responses to internally controlled phobic

thoughts. Thesa subjects also reported greater image clarity and

greater emotional reactions to the imagery than non-phobics. Some

subjects commented that it was difficult to "turn off" the images.

May (1977a) compared autonomic responsivity to three types of

stimulus presentation modes. Using Schwartz's (1971) time-locked

procedure, phobic subjects either saw a slide, heard a sentence,

or produced a thought depicting phobic or non-phobic content.

Subjects then generated images to the stimuli. Heart rate

responding during imagery was greatest for subjects exposed to

either internal or visual stimuli. While all stimuli produced some

habituation over trials, subjects never returned to

baseline--particularly subjects in the internal and visual groups.

Respiration amplitude was greatest for those subjects forming

images to visual stimuli. Respiration rate did not differ between

groups. Electrodermal activity appeared to be most responsive to

the visual mode; subjects in the visual group showed greater skin

conductance level increases and more rapid habituation than did

subjects in the other groups.

Internally controlled thoughts, then, have been shown to

produce autonomic responses comparable to those produced to visual

representations of feared stimuli. These data support the notion

discussed earlier that imagery and ruminative thoughts can play

an important role in the maintenance of phobic disorders. May's

(1977b) finding that some of his phobic subjects could not "turn

off" the images is an interesting example of this effect. A likely

explanation for this phenomenon is the cohesiveness of

propositional networks in phobics, as was suggested by Lang et

al. (1983).

Imagery Ability in Children

As noted earlier, many imaginal treatments used with adults

are used with children with little or no modification. Reviews of

the child treatment literature (Graziano, DeGiovanni & Garcia,

1979; Hatzenbuehler & Schroeder, 1978) have noted that the

majority of reports are case studies. The few well controlled

studies do not provide strong support for the effectiveness of

imaginal treatments with children. It is important to consider

developmental factors that can influence the ability of children to

benefit from such therapies. The discussion now turns to a

consideration of children's ability to use imagery.

Piaget and Inhelder (1971) noted that mental images and

symbolic schemata are the basis for conceptual thought. They

describe two types of imagery: reproductive imagery and

anticipatory imagery. Reproductive imagery refers to an

internalized representation of an overt sensory or motor event,

while anticipatory imagery is the manipulation of these mental

images. Anticipatory imagery ability develops at 7 to 8 years of

age. Piaget and Inhelder observed an improvement in children's

memory for changes in physical stimuli at approximately age seven

and suggested that this improvement was a function of storage of

this information in the form of images.

Much of the work on children's ability to use imagery is found

in the literature on learning and memory. One issue in this

literature is whether children's mental representations are pictorial

in nature, with imagery ability developing later, or whether the

reverse is true. The investigators typically hold a view of mental

images as pictures in the mind but it will be seen that many of

the results can be re-interpreted in terms of a propositional

network theory of memory.

Investigations of the effects of imagery as an aid to paired

associate learning generally have shown that younger children do

not benefit from the use of imagery to the same degree seen in

older children and adults. For example, Paivio and Yuille (1966)

found that children's memory for paired associates was not

enhanced when the items were concrete rather than abstract,

although the concrete items were assumed to be more readily

imaginable (Paivio, 1970).

Paivio and Yarmey (1966) compared the effectiveness of pictures

and concrete noun labels on performance of college students on a

paired associate learning task. They found that picture-word pairs

resulted in better performance than picture-picture pairs. Dilley

and Paivio (1968) noted that the superiority of picture-word pairs

was even greater in 4- to 6-year old children. Paivio and his

colleagues suggested that the picture-word pairs were superior to

the picture-picture pairs because the pictures required decoding

before the child could make a response. They concluded that

children less than 7 or 8 years old, who have not yet developed

the ability to use anticipatory imagery, may have difficulty

making transformations between words and images, and vice versa.

Rohwer (1970) posed an alternative explanation for Paivio's

findings. He hypothesized that pictures are easier to remember, but

only when verbal labels are stored with them. He assumed that

younger children are unable to simultaneously store visual

information and verbal information, and that this ability increases

with age. He thus predicted that the superiority of pictures to

words in enhancing paired associate learning would increase with


Rohwer (1968; cited in Rohwer, 1970) studied kindergarten, first

and third grade children and found that picture pair performance

was better than word pair performance; this difference was

increased with grade level. Performance on picture-aural

presentation pairs was compared with performance on

picture-picture pairs. Supplying the verbal label yielded superior

performance; however the magnitude of this superiority decreased

with increasing grade level. This supports the notion that the

children are increasingly able to provide their own verbal label

as they get older. A similar study was conducted on a group of

older children: third and sixth graders (Rohwer, Lynch, Levin &

Suzuki, 1967). Pairs of words were presented either as pictures or

as printed words. All subjects heard the words read aloud. For

both grades, better performance was obtained for the picture pairs

than for the word pairs. The superiority of the pictures was

greater in the third grade group.

Rohwer (1970) concluded that if imaginal representations of

items are more likely to be elicited with pictorial than verbal

items, then imagery facilitates learning in children. He suggested

three possible explanations for the developmental trends observed:

1) the probability that imagery will be evoked is lower in younger

children than in older children; 2) the capacity for obtaining some

benefit from imaginal representations develops later than that for

verbal representations, and 3) the ability of imaginal storage to

facilitate learning is dependent on concurrent storage of a verbal

label for that image, and that this ability is more likely to be

seen in older children. Note that Rohwer's data contradict the

notion that the ability for pictorial representation develops earlier

than the ability to learn from verbal representations of


Rohwer and his colleagues also studied the influence of

elaboration of images on paired associate learning in children. In

the Rohwer, Lynch, Levin and Suzuki (1967) study described

above, subjects heard one of three different verbalizations along

with the pictures. These verbalizations differed on the type of

connective used to link members of the noun pair. Thus, subjects

heard either a conjunction (i.e., the shoe and the chair), a

preposition (i.e., the shoe under the chair), or a verb (i.e., the

shoe taps the chair). The verb condition produced the most correct

responses, with prepositions being the next most effective. Rohwer

(1970) suggests that the three kinds of connectives evoke different

types of imagery and reasons that action imagery (verb) is more

memorable than static imagery and that locational static images

(preposition) are more memorable than coincidental static

(conjunction) images. For the third graders, there was a linear

relationship between type of connective and learning. For the sixth

grade subjects, the preposition facilitated learning to the same

degree as the verb connective. The researchers suggested that

older children have a lower threshold for the use of facilitory

processes than do younger children.

Rohwer, Lynch, Suzuki and Levin (1967) compared the effects of

verbal and pictorial elaboration. Elaborations used in both

conditions corresponded to those used in the study just described.

In the verbal mode, verb connectives produced the best

performance, while in the depiction mode, the greatest facilitation

of learning occurred with action pictures. There were no

developmental differences in the relative efficacy of the verbal and

visual response modes.

A closer study of these trends (Rohwer, Lynch, Levin & Suzuki,

1968) examined the effects of only four verbal-visual conditions

used in the previous study. These were naming-coincidental,

naming-action, verb-coincidental, and verb-action (the coincidental

condition refers to the two items pictured together, but not

interacting). For al' three grade levels, the elaboration conditions

produced better recall than the naming-coincidental condition. A

grade by condition interaction was obtained. For kindergarten and

and first grade children, the verb-coincidental condition produced

more facilitation than the naming-action condition. The reverse was

true for the third and sixth graders. The authors suggested that

the older children are better able to make use of the imagery

evoked by the action depiction. A possible explanation for this is

that younger children do not store an appropriate verbal label

with the action image. Thus, younger children obtain more benefit

from hearing a sentence with the picture than do the older

children. This explanation is consistent with the data.

The studies just described provide support for Rohwer's (1970)

contention that while action imagery can facilitate performance, the

ability to obtain full benefit from this imagery develops later than

the ability to benefit from verbal elaboration. Rohwer suggests

that this developmental trend is due to the fact that the well

organized linguistic system develops earlier than the less

organized imaginal mode. This view is counter to the more widely

accepted notion that the capacity for visual representation develops

earlier than that for verbal representation.

Reese (1970) notes that imagery facilitates learning in older

children and adults, but that less faciliatation is observed in

younger (i.e., preschool) children. He suggests several possible

explanations for this observed developmental trend. The first is

that facilitation depends on covert verbalization of the image and

that young children are less able to do this. However, Reese notes

that this view contradicts the notion that young children's thought

is primarily iconic, and that one would expect sentences to

interfere with retention. A second explanation is that visual

memory is less effective than verbal memory in young children.

This view also contradicts the belief that young children's thought

is primarily imagery based. A third explanation is that young

children ignore pictured interactions between objects and code the

information as if no elaboration was presented. Thus, younger

children may have difficulty with the production of mediators.

Another alternative explanation is that young children can

encode information visually, but that they are unable to decode it

verbally. The work by Paivio and his colleagues provides some

support for this notion. However, some work cited suggests that

young children have difficulty producing mediators, rather than

that they have problems decoding the mediator. Further, Reese

points out that in mediation studies, in which a deficiency is

observed, the mediator is conditioned to the stimulus and response;

yet in paired associate studies, the mediator is not conditioned,

but suggested by the image. Control subjects have an equal

opportunity to learn the links, but perform less well than imagery

groups. A fifth alternative proposed by Reese is that the materials

used to evoke images are deficient in detail, and thus are not

adequate to evoke imagery in younger children, or at least not to

produce images that are sufficiently vivid to enhance memory. The

final alternative presented does not hold that imagery is less

effective with young children, but rather, that young children fail

to "read" the pictorial materials used to evoke images. The child

may form images of the stimulus and response items, but even a

picture of the objects interacting fails to produce an image of this

interaction. Reese suggests that a sentence can produce imagery

since the elements and their interaction are explicitly named.

Reese concludes that facilitation of memory does not result from

imagery per se, but from integrated imagery, which provides

contextual meaning for the object to be remembered.

Reese's (1970) point regarding the inability of certain materials

to evoke imagery in children supports the contention of many

clinical researchers (Elliot & Ozolins, 1983; Graziano, DeGiovanni,

& Garcia, 1979; Harris, 1983). A study by Wolff and Levin (1972)

shows that children may need more powerful, or more elaborate,

stimuli in order to be able to use imagery. They studied the role

of overt motor activity in the formulation of mental images.

Kindergarten and third grade children were presented with pairs of

actual toys in a paired-associate learning task. Four learning

conditions were investigated. In the first, the control condition,

children were simply instructed to remember which toys go

together. In the imagery condition, children were told to form an.

interactive image of the toys. In the third condition, the child

watched the experimenter manipulate the toys in an interactive

way, while in the fourth condition, the child was encouraged to

manipulate the toys on his own. Memory was tested by recognition.

The children were asked to select the response toy from an array

of several toys. This method avoided the necessity of having the

child decode a visual image to a verbal response. Wolff and Levin

found that the two manipulation conditions produced significantly

better performance than the imagery and control conditions. For

kindergarten students, the manipulation conditions did not differ

significantly from one another. For the third grade subjects,

performance in the two manipulation conditions and the imagery

condition was significantly greater than the control condition.

These three conditions were not significantly different. The

findings provide support for the notion that dynamic imagery does

not develop until the age of seven.

A second study of kindergarten and first grade children showed

that it was the actual manipulation of the objects and not just

observation of the interaction that produced enhanced memory. The

authors concluded that up until the age of five, the formation of a

dynamic image depends on concomitant motor output, that

duplicates the form of the percept. With development, formation of

images becomes less dependent on overt motor activity. Although

results of other studies described indicate that children are able

to generate imagery upon instruction, with the aid of pictures or

even without external cues, these findings emphasize the need for

researchers and clinicians alike to attend to children's special

abilities when using imagery.

Kosslyn (1976) examined the question of whether children use

imagery to retrieve semantic information. First graders, fourth

graders and college students participated in a pair verification

task. For example, a subject might be asked to verify whether "a

mouse has whiskers" was true. Kosslyn looked at two dimensions of

the properties to be verified: size of the detail and the association

strength. He hypothesized that if imagery was used, a longer

reaction time would be found for smaller details than for large

details. Further, with imagery, it was expected that the strength

of the association would be related to reaction time. Kosslyn

assumed that adults would use non-imaginal methods, while

children would use imagery to complete the task. Thus, no

difference was expected for children with or without imagery

instructions, while adults were expected to be slower when forced

to use less effective imagery. Subjects completed two blocks of pair

verifications, first without imagery and then with instructions to

use imagery. The true properties were either high association/low

area or low association/high area. Area refers to the size of the

imagined object, while association refers to the strength of the

relationship between a property and a noun (i.e., whiskers and


Kosslyn's hypotheses were borne out by the data: reaction times

decreased with age. Imagery instruction did not influence

performance of the younger children, while adults performed more

slowly. In addition, in the no imagery condition, high

association/low area properties had faster reaction times, while low

association/high area properties were responded to more quickly in

the imagery condition. Kosslyn inquired to the strategies subjects

used in the no imagery condition. The number of subjects reporting

the spontaneous use of imagery decreased with age. Those first

graders reporting that they did not use imagery showed a similar

effect of item type as was seen with the adults, that is; high

association/low area properties were verified more quickly. The

results of this study suggest that adults who do use imagery

spontaneously do it more quickly than children. Perhaps the adults

use the imagery more efficiently, imaging, for example, only the

relevant parts of the object. Furthermore, the fact that first

graders did not respond most quickly to the high association items

as did the fourth graders and adults suggests that there are some

qualitative changes in memory with development.

Prawat and Kerasotes (1979) extended Kosslyn's work with a

few methodological changes. Different subjects participated in the

imagery and no imagery conditions, and pictures were used in the

imagery condition. In addition, Prawat and Kerasotes defined two

types of meaning: perceptual and functional meaning. It was

expected that if children used imagery to retrieve semantic

information, perceptua1 propertiess would be responded to more

quickly, since thes3 should be more easily imaged. This hypothesis

is based on the view of images as pictures in the mind. An

interaction was expected between type of meaning and saliency

(association) such that for perceptual features large size/low

saliency properties would have faster reaction times, while for the

functional properties, small size/high saliency items would be

responded to more quickly. Second grade students served as

subjects in this study.

There was a main effect of property type, with perceptual items

having faster reaction times, supporting the view that children do

use imagery. However, children in the imagery condition did not

retrieve semantic information more quickly. The authors suggest

two possible explanations for this; either subjects cannot make

effective use of the induced imagery, or control subjects

spontaneously used imagery in an effective manner. Additional

comparisons showed that the imagery group took longer than the

control group to respond to perceptual items; however, there were

no differences between groups on reaction time for large versus

small items. The authors suggest that induced imagery may

interfere with spontaneous imagery. The authors also found that

high saliency items were responded to more quickly than low

saliency items, while there was no such difference for the

perceptual items.

The studies discussed in this section suggest that children are

able to use imagery to enhance memory, to a limited extent. Young

children may have difficulty with the transformations required to

form images from verbal information and to decode images to

language. Young children may not be as likely to use imagery

spontaneously, and stimulus materials may need to be richer in

detail than those typically used with adults. The findings obtained

in the studies reviewed can be discussed in terms of a

propositional theory of memory. It seems that memory networks in

children may be less well organized or more immature than those

in adults. Wolff and Levin's (1972) finding that overt motor

activity is important to image formation suggests that more

powerful stimuli are needed to evoke the image network or to form

associative links. However, research is needed to determine what

"more powerful stimuli" really means. It could be that children

need to have stimuli that are as realistic or as detailed as

possible, such as actual objects or films. It may mean that

children may be better able to imagine objects or situations for

which they already have a propositional memory network. Or, it

could mean that to-be-imagined stimuli need to be more emotionally

salient for children. Viewing the findings of Kosslyn (1976) and

Prawat and Kerasotes (1979) in terms of network theory suggest

that the more rapid performance of adult subjects could be a

function of the better organization and stronger associative links

in their memory networks. The fact that high association stimuli

were responded to more quickly in the Kosslyn (1976) study

supports the notion that the spread of activation along strong

associative links in the memory network facilitated responding.

That some children also responded more quickly to high association

items suggests that propositional memory networks are present in

children. Subjects were first graders, approaching Piaget's stage

of concrete operations. Perhaps some of the subjects had already

reached this stage, and concomitantly had developed the capacity

for better organized propositional networks of memory.

Affect and Memory

Thus far, this paper has examined relationships between

imagery and memory in order to understand developmental

considerations that may be relevant to the use of imagery in

treatment of children. Since these treatments can require that the

child retrieve memories of emotionally laden situations, it is

important to consider the influence of affect on memory.

Bower (1981) noted that an adult subject's affective state can

serve as a contextual cue that can aid recall of material learned

while in that state. He used hypnosis to vary the mood of his

subjects during learning and recall of two word lists. He found

that recall was improved when the mood at learning and recall

were congruent, and that recall was impaired if the moods at each

point were opposites. Using a propositional theory of memory

(Anderson & Bower, 1974), Bower suggested that when a subject

stores information in memory, that affect can serve as a context,

and that contextual information is stored in propositional nodes.

When a subject attempts to recall information when in the same

mood as during learning, activation from the affect nodes spreads

throughout the network, summing with activation from the context

nodes, thus facilitating recall.

Similar results have been obtained with children. Bartlett,

Burleson and Santrock (1982) conducted an experiment to determine

whether memory traces for verbal stimuli included information

about the subjects' emotional state during learning. They

hypothesized that if this was so, then the subjects' emotional state

could serve as an effective cue for the retrieval of information.

Subjects were 5- and 8-year-old children. Subjects learned two

lists of words and recalled them immediately and again following a

ten minute delay. Prior to learning each list, either a happy or

sad mood was induced by having the child think of a personal

experience appropriate to the mood. The opposite mood was induced

prior to learning of the second list. For recall, half the subjects

in each condition then experienced a happy mood, the other half

experienced a sad mood. Thus, there were four conditions of happy

(H) and sad (S) moods; H-S-H, H-S-S, S-H-H, S-H-S.

In the first experiment a relaxation exercise preceded all other

procedures. No state dependent learning effect was obtained.

Relaxation was omitted in the second experiment. In this

experiment, a clear state dependent effect was observed, with

greater recall occurring for the list learned in the mood equivalent

to the mood at recall. This effect occurred for both age groups.

This affect dependence was asymmetrical; a happy mood at test

produced greater recall than a sad mood for a list learned while

happy; no such effect was obtained for lists learned in a sad

mood. Bartlett et al. (1982) concluded that affect can serve as an

effective retrieval cue, but noted that relaxation can preclude this

state dependent effect. They suggested that the perception of

emotional arousal is a prerequisite for the experience of emotion

and that relaxation seems to decrease the intensity of the emotion.

Nasby and Yando (1980) conducted a similar study; they

examined the influence of mood on the recall of effectively valent

information. Subjects in this study were fifth graders who were at

least 10 years old. Moods were induced in the same manner as in

the Bartlett et al. (1982) study. A happy or sad mood was induced

prior to list learning. Either the same or opposite mood was

induced for the free recall test condition. Only one word list was

used; these words were slightly positive, highly positive, slightly

negative and highly negative. Mood at retrieval had no effect on

recall; however mood at encoding did affect performance for girls

only. Happy mood at encoding facilitated learning of slightly and

highly positive words, and disrupted learning of slightly negative

words. Sad mood at encoding disrupted learning of positive

material, but did not facilitate learning of negative words. This

asymmetrical effect is similar to that seen in the Bartlett et al.

(1982) study. Nasby and Yando (1980) noted that studies in the

literature have reported that mood at retrieval (rather than at

encoding) influences recall in adults, and they suggest that there

may be an age related shift in ability to regulate the influence of

mood on encoding and retention of affective materials.

In another study, Nasby and Yando (1982) repeated the same

design, but also varied the frequency of word usage, and position

of the word on the list. Fifth graders (mean age = 10.75 years)

were used as subjects. As in the previous study, a selective

encoding effect was obtained for the medium and high frequency

words only. A happy mood at encoding resulted in the recall of

more high and medium frequency positive words. A sad mood at

encoding disrupted recall of positive words. No selective encoding

of negative material was observed. Unlike the previous study, a

selective retrieval effect was also found. Happy mood at retrieval

facilitated the recall of positive words, while sad mood did not

disrupt recall of positive material. No selective retrieval occurred

for negative material. Congruence of mood at encoding and

retrieval did not produce superior recall to incongruent moods;

thus there was no affective state dependent effect. Note that

studies with adults (Gilligan & Bower, 1984) obtained state

dependent effects only when two word lists were used. So, while

moods do have some impact on memory in children, the effects seen

in this study do not mirror those obtained with adults.

A second experiment reported by Nasby and Yando (1982)

examined the influence of an angry mood on recall. Angry mood at

encoding disrupted recall of positive words, but facilitated recall

of highly negative wor'-3. Anger at retrieval did not have any

effect on performance. No state dependent effect was obtained with

the angry mood.

The authors note that the pattern of results obtained differs for

children and adults. For both, positive mood facilitates the recall

of positive material. For children, negative mood inhibits the

recall of positive information, while this is not the case for

adults. Nasby and Yando suggest that the ability to regulate mood

states could be an important quality that changes with

development. Thus, adults may be better able to maintain positive

moods and avoid or eliminate negative moods. They point out that

this ability to regulate the effect of moods can have an important

influence on the learning of overt behavior. These findings also

support the contention that memory networks in children are less

well organized than those in adults.

The studies just described focused on affect-dependent memory

for isolated word lists. Bartlett and Santrock (1979) examined the

influence of mood on episodic memory. This approach can yield

more information relevant to clinical issues. Preschoolers (mean

age 5 years 7 months) were tested for memory of a list of words

presented in the context of a story. Pictures were presented to aid

recall. A 2 X 2 factorial design was used, with happy vs. sad

mood at input and happy vs. sad mood at recall. Mood was

induced at input by having the experimenter tell a story that was

either happy or sad. The same words were incorporated into both

happy and sad stories. Mood at recall was induced by having the

child look at a series of happy or sad pictures. As a manipulation

check, children were asked to point to one of two faces (smiling or

frowning) to indicate how each story made him feel. At both input

and retrieval, the experimenter acted in a manner appropriate to

the mood being induced. A variety of retention tests were given.

The free recall data indicated that subjects who learned words

while in a happy mood had better recall when tested in a happy

mood than when tested in a sad mood. There was no significant

difference between test moods for subjects who learned words while

in a sad mood. These results indicate that a change in affect can

influence children's ability to generate appropriate retrieval cues.

As in the other studies discussed, an asymmetric effect was


Children's Responsivity to Imagined Stimuli

Studies reviewed thus far have shown that imaginal treatments

can be very effective in the treatment of adult phobic disorders,

and that it is important to consider the patient's physiological

responses during imaginal treatment. This area is poorly

researched with children (Johnson & Melamed, 1979). A review of

the literature on imagery and memory in children reveals that

while children do have an ability to use imagery to enhance their

memory, they do not do this as effectively as adults. They may

have difficulty with transformations between verbal and imaginal

materials (Paivio, 1970), they may not use imagery spontaneously

(Kosslyn, 1976) or they may need very detailed stimuli in order to

evoke imagery (Wolff & Levin, 1972).

When using imagery to treat child patients developmental

factors must be taken in to consideration. For example, the

therapist cannot assume that children create images in the manner

in which they were instructed. Children may elaborate upon the

images, and these elaborations themselves may be important. The

meaning assigned to a given image must be considered as well, as

this may vary from child to child. For example, one child asked

to imagine a teacher at school might image a positive situation,

with a benevolent teacher providing positive reinforcement. A

second child, however, may imagine himself being punished by the

teacher; this child is likely to find the imagery experience to be

quite unpleasant.

Rosenstiel and Scott (1977) recommend that imagery scenes be

tailored to the age of the child. They also suggest that involving

the child's existing fantasies into an imagery procedure may

decrease the complexity of the scenes and may help to maintain the

child's interest during the procedure. Having the child report

specific details of the procedure is expected to enhance image

clarity for the child. Finally, Rosenstiel and Scott (1977) suggest

that children may have difficulty describing their behavior and

images. Thus, attending to non-verbal cues such as fidgeting, skin

flushes, or heart rate change may provide the therapist with

useful information about the effects of the scene on the child

(i.e., changes in anxiety level).

All of the treatment studies described thus far have examined

physiological responsivity of adult subjects. Only two studies in

the literature have considered children's physiological responses to

imagery. It is important to investigate this issue in children for

two reasons. First, children present to treatment centers with a

variety of phobic concerns. Although imagery is used in a number

of treatments for children, there are few data on the effectiveness

of these approaches (Hatzenbuehler & Schroeder, 1978). Second, an

understanding of developmental changes in the phenomena described

in this paper may yield important information on the etiology of

phobic disorders and on the structure of phobic disorders in


Tal and Miklich (1976) studied 10- to 15-year old children with

chronic asthma. The children were asked to imagine neutral,

fearful, and anger-arousing situations. Increased heart rate and

decreased expiratory flow rates during imagery occurred in the

fear and anger sessions. Neutral images resulted in slowed heart

rate and increased expiratory flow rates. The heart rate increases

suggest that the imagined scenes (fear and anger) did produce

arousal, which was associated with pulmonary function.

Hermecz and Melamed (1984) used Lang's paradigm (Lang, 1977,

1979; Lang et al., 1980) to study imagery in 6- to 12-year-old

dental patients. It was expected that children would not be able

to generate emotions as effectively as adults and that they may

need additional aids, such as a modeling film, to generate

appropriate emotional imagery (Lang, 1977).

Children were given either stimulus or response imagery

training. Imagery training included two trials on each of two

action oriented scenes: riding a bicycle and flying a kite.

Following imagery training, the child viewed a film about dental

treatment. Each group saw the same videotape; however, the

soundtracks differed according to the type of imagery training

given. Thus, the sound track for stimulus subjects focused on

descriptive aspects of dental treatment, while the response film

focused on expected physiological responses to treatment. After film

viewing the child was taken to a dental operator, where

restorative treatment was performed.

Results indicated that there was a borderline interaction

between condition and scene, with response subjects showing

greater heart rate increases to the first practice action scene. On

trial two of scene one, response subjects showed an increase in

heart rate during imagery with decreased heart rate during the

subsequent recovery period. This pattern is consistent with that

seen in adult patients. Similar results were obtained with the

respiration data. Response subjects indicated greater image clarity

across scene types, while clarity decreased across scene types for

stimulus subjects. While there were no significant differences in

physiological responsivity during film viewing, response subjects

showed differential responding across scene contents. Stimulus

subjects responded in the same way across scenes. In the

operator, response subjects were more disruptive than stimulus

subjects during dental treatment. For the response subjects, there

was a significant correlation between overall disruption and

self-reported dental fear. These children thus showed concordance

between self-report and behavior following response training. Thus,

children do appear to be able to generate physiological responses

to imagined scenes. Support was obtained for the notion that

response training is more effective in eliciting these responses

than is stimulus training.

Children and the Bio-informational Theory

The studies described in the previous two sections support the

notion that children do store emotional information in memory. As

was the case with the adults studied by Bower and his colleagues

(Bower, 1981; Gilligan & Bower, 1984), affective information

appears to function as a contextual cue that can influence memory

performance in children. However, the results of studies described

in this paper suggest that there are some differences in the

imagery ability of adults and children, as well as in the

emotional memory functions of these two age groups. For example,

adults may be able to perform a paired associate learning task by

forming an interactiv-- image; children, however, need to

manipulate actual objects in order to form such an image.

Klingman, Melamed, Cuthbert and Hermecz (1984) obtained similar

results in an emotionally laden situation. They found that children

about to undergo dental treatment benefitted more from a modeling

film that allowed them to actually practice coping skills than from

a film that only told them what techniques they could use to cope

with treatment. Thus, children may need to engage in motor

behaviors in order to store some types of information in their

emotional networks.

Some researchers (Gilligan & Bower, 1984; Kagan, 1984) suggest

that infants are born with an innate emotional structure. They

note however, that the fact that there are changes in the structure

and function of the nervous system throughout development makes it

unlikely that emotional functions in children and adults are alike.

Rather, the emotional systems of children are modified throughout

development via the processes of learning and acculturation.

Nevertheless, there is research that suggests that emotions and

memory are linked together even in infancy. It has been widely

reported (Lewis & Rosenblum, 1974) that infants develop a fear of

strangers at approximately eight months. It is generally held that

certain cognitive changes are required for the development of this

fear; the infant needs to develop a memory schema for familiar

people, and then to be able to compare the stranger with his

schema for familiar people (Schaffer, 1974). A discrepancy between

the stranger and the schema leads to fearful reactions. This

phenomenon is important to the bio-informational theory for two

reasons. First, it shows that emotional reactions are linked to

memory even at a very young age. Second, the fact that infants

develop this fear at a fairly predictable time suggests that

maturational factors are involved in the development of emotions.

Many of the studies on imagery and memory described in this

paper found that changes in the abilities of children occurred at

approximately age seven or eight, concurrent with movement into

Piaget's stage of concrete operations. It is likely then, that

research on children' s emotions will find some qualitative

differences between emotions in children and adults; these

differences are likely to be diminished throughout childhood.

With regard to the bio-informational theory, several differences

between children and adults can be postulated. First, there is

likely to be less information in the emotional networks of children,

by virtue of the fact that they have fewer experiences that can be

stored in memory. It may be that subtle nuances of various

situations are absent in the networks of children, and this

information may not be understood. Second, the links between

propositions in children's memories may be less well organized,

and the associations between propositions may be less coherent

than is the case in adults. Children may need stronger stimuli

than do adults in order to access the emotional networks. The

studies by Wolff and Levin (1972) and Klingman et al. (1984)

suggest that motoric information is integrally bound to the other

information stored in children's memories. Finally, children may

not be able to use imagery as effectively as adults. Imagery

ability for emotional stimuli is likely to vary with age, as is the

case for imagery ability in memory tasks. There is evidence (Tal &

Miklich, 1976) that older children can generate emotional images.

The results of the Hermecz and Melamed study suggest that

children may be able to benefit from imagery training. That study

did not do a physiological assessment of emotional imagery;

however, response training did influence children's responses to a

film about dental treatment.

Statement of the Problem

The literature reviewed in this paper suggests that imagery

ability in children follows clear developmental trends. Young

children are often not able to use imagery in memory tasks;

further they may not be able to transform information from the

verbal mode to the visual (or imaginal) mode, and vice versa.

These findings have important implications for the use of imagery

in child treatment. Specifically, therapists need to evaluate the

efficacy of these treatments and to attempt to specify the ways in

which children actually apply imagery instructions.

It has also been seen that physiological responses during

imagery play a significant role in the efficacy of treatments such

as systematic desensitization. However, with only two exceptions,

this research has not been applied to children.

The purpose of the study presented here was to extend the work

done by Hermecz and Melamed (1984) on emotional imagery in

children. This study demonstrated that children can generate

physiological responses to action scenes. It is not clear, however,

whether children can show differential responding across different

scene contents, or whether the results obtained are simply artifacts

of the imagery task demands. In addition, subjects in the Hermecz

and Melamed (1984) study were not selected for fearfulness.

The present study compared the responsivity of high and low

dentally fearful children to three imagery contents: dental fear,

school fear, and neutral. Children were given either stimulus or

response imagery training. The study was designed to address

three major issues:

1) Can children imagine scenes from structured stimuli (text)

and generate physiological responses appropriate to scene content?

Do they show differential responsivity to the different scene


2) Do fearful and non-fearful children show different patterns

of responsivity?

3) Do stimulus and response training have a differential effect

on responsivity?

In addition to these issues, the study also examined changes in

self-reported and observer rated fear as a result of exposure to

the imaginal materials. The relationship between reported imagery

ability and responsivity was investigated in an exploratory


Hypothesized results were as follows:

1) Fear scenes would evoke greater heart rate responses than

would neutral scenes.

2) Responses to dental fear scenes would be greater in the

fearful versus non-fearful children.

3) Responses to fear scenes would be greater in response

trained versus stimulus trained subjects.

4) There would we an interaction between fear level and

training, such that high fear response trained subjects would show


greater response magnitudes than high fear-stimulus trained


Although skin conductance was assessed during the study, no

specific predictions were made about skin conductance responsibity,

as some studies reported that skin conductance increases during

fearful imagery (i.e., Grossberg & Wilson, 1968), while other

studies reported that skin conductance decreases were observed

during imagery (Lang et al., 1983).



The basic study design included a 2 (fear level) x 2 (imagery

training) x 3 (scene content) design. Fear level (high vs. low

dental fear) and imagery training (stimulus vs. response) were

between subjects variables. Scene content served as a within

subjects variable. The scene contents (Appendix A) were dental

fear (examination and injection), school/social fear (going to the

principal, speaking in class) and neutral scenes (sitting in the

living room, sitting in a lawn chair). The neutral scenes were

designed to be without affective content or physiological response

demands. The school scenes were included for the purpose of

comparison. Since children were selected for dental fear, it was

expected that children would show differential responsivity to

dental content. Examination of responsivity to the school scenes

provided information about whether dentally fearful children are

generally responsive to fearful items, or whether they can

discriminate between contents that are fearful for them and those

that are not. Further, it was expected that some of the children

who were not dentally fearful would show increased responsivity to

the school related scenes. However, no group differences in

responses to the school fear scenes were expected a priori.


Subjects were 24 children ranging in age from 6 to 12 years

old selected by their scores on the Children's Fear Survey

Schedule--Dental Subscale (Melamed, Hawes, Heiby & Glick, 1975;

Scherer & Nakamura, 1968; see Appendix B). This measure has been

shown to be useful as a screening instrument to identify children

high in dental fear (Cuthbert & Melamed, 1982). This questionnaire

was distributed to children in a variety of community organizations

and schools. Children scoring one standard deviation above or

below the mean for their age and sex were asked to participate in

additional screening activities to determine eligibility for the

study. Children with obvious physical or mental handicaps or

cardiovascular problems were excluded from the study.


Heart rate was measured using two Beckman Standard Size

Ag-AgCl electrodes placed on opposite sides of the lower rib cage.

Electrolyte gel was placed inside the electrodes. These were

attached to a Coulbourn Instruments Hi-Gain Bioamplifier/Coupler

Model S75-01. Skin conductance was measured with two Coulbourn

standard Ag-AgCl electrodes placed on the hypothenar eminence of

the left hand. These electrodes were filled with Johnson & Johnson

K-Y Jelly and were connected to a Coulbourn Instruments Skin

Conductance Coupler, Model S71-22. Data collection was controlled

by an MDB microcomputer. Ratings of emotion were made using the





I Lid



Self-Assessment Mannequin (SAM; Greenwald, 1987; Lang, 1980;

Figure 1). Visual presentations of the SAM figure were presented

by computer on an Amdek Video-300 in the subject room. All tape

recordings were played via speaker using a Eumig cassette player.


Each subject participated in three sessions: an initial

screening session and two experimental sessions. All dental

treatment in this study was provided by faculty dentists from the

Department of Pediatric Dentistry at the University of Florida.


Parents of children scoring in the appropriate range on the

Dental Subscale were contacted and asked to bring their child to

the dental clinic for a screening examination. After informed

consent was obtained (Appendix C), each child was given a brief

examination to determine whether he or she had at least two

cavities that could be filled as part of the study. Children

meeting this criterion were then given the Peabody Picture

Vocabulary Test (Dunn & Dunn, 1981). Children scoring below 85

were excluded from the study. Next, bitewing x-rays were taken.

Finally, parents completed the Subject Data Form (Appendix D),

which provided demographic information as well as information

about the previous dental experience of the child. Parents were

asked to rate their child's fear and cooperation during past dental

examinations and dental injections.

Session 1

Upon arrival at the clinic, each child was administered the

full version of the Children's Fear Survey Schedule (Melamed,

Weinstein, Hawes & Katin-Borland, 1975; Scherer & Nakamura, 1968;

Appendix E). This provided a measure of general fearfulness as

well as dental fearfulness; the Dental Subscale is embedded within

this measure. This measure also includes items about school and

social situations. A subscore based on these items was compared

with the children's physiological responsivity to the school scenes.

Next, children saw the dentist to have a filling done. This was a

standard procedure for all children, involving an anesthetic

injection, placement of a rubber dam, and drilling. When treatment

was completed, the dentist rated the child's cooperation and

fearfulness using two ten-point scales (Appendix F). Dentists were

blind to each subject's fear level and group assignment. Dental

treatment was videotaped. These tapes were later scored for the

child's disruptive behaviors with the Behavior Profile Rating Scale

(Melamed, Hawes, Heiby & Glick, 1975; Appendix G).

Following dental treatment, the child was taken to the

laboratory for imagery training. First, the child completed the

Sheehan (1967) version of the Betts (1909) Questionnaire Upon

Mental Imagery (QMI; Appendix H). Some modifications were made

in the use of the QMI for this study. Children rated the vividness

of imagery to a variety of items by pointing to one of a series of

photographs of keys. The photographs varied in the degree of

focus, from very clear to extremely blurry. If a word on the

questionnaire was unclear to the child, a definition was provided

from a standard list. The QMI has been shown to be predictive of

physiological responsivity and reported image vividness in adults

(Miller et al., 1981), but the scale has not been used with

children. Thus, test-retest reliability was determined on a separate

sample of children. Reliability on a sample of 11 children was

r=.83, p<.001. There was an average of 27.8 days between testing

sessions for the reliability study, although for most children,

sessions were one to two weeks apart.

Next, the child completed the Measure of Communication

Apprehension (MLCA; G-:-'ison & Garrison, 1979; Appendix I). This

measure of public speaking anxiety was used in an exploratory

fashion, in order to determine whether there was a relationship

between reported public speaking anxiety and physiological

responsivity to the school related scenes during imagery. The MECA

has been shown to be both reliable and valid (Garrison &

Garrison, 1979) and to be sensitive to changes in public speaking

anxiety resulting from behavior therapy (Harris & Brown, 1982). As

only half of the children in this study completed this

questionnaire, results will not be discussed here.

After completion cf these measures, the child was given a

relaxation exercise (Koeppen, 1974; Appendix J). This exercise was

designed for use with children, and involves tension-relaxation

cycles for eight muscle groups. Relaxation training was given

during Session 1 to familiarize the child with this procedure since

it was used during the second session. During this exercise, the

child was seated in a reclining chair. The overhead light was

turned off, and a dim floor lamp was turned on.

After the relaxation exercise, the child was given imagery

training. Half of the children in each dental fear group were

given a "response" imagery set, while the remaining children were

trained so as to provide a "stimulus" imagery set. Children in

each fear group were matched for sex, age, race and dentist and

then randomly assigned to a condition. Two action scenes were

presented twice to each child. One scene described flying a kite,

the other, riding a bicycle. Action scenes were used because they

have been shown to generate physiological responsivity when they

contain response propositions, yet they are not effectively


Stimulus group

Each scene was tape recorded and presented to the child via a

speaker. The child was instructed to close his or her eyes and to

imagine the scene as it was being read. The stimulus oriented

instructions encouraged the child to create a detailed mental

picture of the situation, including as many descriptive aspects as

possible. The child continued to imagine the scene after the

presentation for an additional 20 seconds. The child was then told

to relax his or her muscles (20 seconds). Next, the child was

asked a series of questions about the imagery experience. These

focused on the stimulus aspects of the situation (i.e., Could you

see the kite dancing in the wind?). Verbal praise was provided

each time the child reported having imagined a stimulus aspect of

the scene. The same imagery script was then repeated, followed by

two trials for the second script. These imagery training procedures

are listed in Appendix K.

Response group

The procedures used for the response group were nearly

identical to those used for the stimulus group (see Appendix L).

However, the scenes required that the child focus on his/her own

physiological responses during the imagined scene rather than on

stimulus aspects. Each scene contained one response proposition

from each of the following systems: heart rate, sweating,

breathing, muscle tension and eye movement. Questions presented to

the child concerned the response aspects of the imagery (i.e., Did

you feel the sweat dripping down your face?). Verbal praise was

given contingent upon the child's report of response propositions

in imagery. As was the case for the stimulus group, each scene

was repeated twice.

Session 2

Session 2 took place approximately one week after Session 1 in

which all subjects had undergone routine restorative dental

treatment. In the laboratory, children were seated in the reclining

chair. Following electrode placement the lights in the room were

dimmed. Next, the imagery procedure was explained to the child

(Appendix M). The child was questioned to insure complete

understanding. In addition, an experimenter remained in the room

with the child to answer questions and to encourage maximal

cooperation with the procedures. Before beginning physiological

assessment the child completed the same relaxation exercise given

in Session 1. Children were rewarded with a sticker for cooperation

during electrode placement and again following cooperation with

the imagery procedure.

The imagery assessment consisted of seven scenes; a practice

neutral (data for this scene were discarded) and two scenes each

of dental fear (dental exam and dental injection), school fear

(blackboard and principal) and neutral (lawn chair and living.

room) content. Children's emotional responses to these scenes were

evaluated in a pilot study on a separate sample of children. This

study is described in Appendix N. The scripts for all subjects

were identical, and all contained response propositions. Twelve

scene sequences were determined at the start of the study. Each

child in each of the four groups had a different scene order. The

scene orders were constructed so that each of the six scenes

appeared in a given ordinal position two times.

Data was collected during a pre-image baseline phase (rest

period). Each child was then instructed to imagine each scene as

it was read (read period). When the tape finished, the child

continued to imagine the scene until a tone was presented (image

period). When the child heard the tone, he/she then stopped

imagining, and relaxed all of his/her muscles (recovery period).

When a second tone was presented, the child opened his/her eyes

and made ratings of his/her feelings during imagery. This was

accomplished using the Self-Assessment Mannequin (SAM; Lang,

1980). SAM was presented by the computer, and appeared on a

video monitor on a counter near the subject. Three affective

dimensions were rated: pleasure, arousal and dominance. The child

used a joystick to adjust the picture of SAM until it best

represented his/her feelings. In addition, the child rated the

vividness of his/her image by moving an arrow along a rating

line. Children were shown the array of pictures used with the QMI

to help them make this rating. When the ratings were completed,

the child closed his/her eyes and waited for the presentation of

the next scene.

After completion of the imagery assessment procedure the child

was returned to the dental clinic for treatment of a second filling.

The same dentist performed both treatments for a given child.

Again, the session was videotaped. The dentist provided ratings of

the child's cooperation and fearfulness using the ten-point scales.

Finally, the Dental Subscale of the Children's Fear Survey

Schedule was administered to the child.

Data Scoring

Physiological Measures

Physiological measures were taken during each of the imagery

scenes, with separate values obtained for each phase of the image

presentation. This consisted of a 30 second rest period (baseline),

a 30 second read period, a 30 second image period, and a 30

second recovery period. Up to 50 seconds were provided for the

actual reading of the scene; however, data were collected only

during the final 30 seconds of this 50 second period. Difference

scores were computed in two ways. One method was to subtract the

value of the rest period baseline score from the value for the

period under consideration. For most of the analyses to be

described the read period was used as baseline with change scores

computed in the same manner.

Self-Report Measures

Items on the Children's Fear Survey Schedule and the Dental

Subscale are rated on a five point scale, with a score of 1

representing low fear, and a score of 5 representing high fear.

Scores on the full CFSS range from 50 to 250. Scores on the Dental

Subscale range from 15 to 75. The score for the question on

injection fear was used separately in correlational analyses. Items

on the QMI are rated on a seven-point scale, with a score of 1

representing maximum image clarity, and a score of 7

representing poor imagery. Scores range from 35 to 245, with lower

scores indicative of better imagery ability. Items on the MECA are

rated on a 1 to 5 scale, with a rating of 1 meaning "very happy/I

like it a lot" and a rating of 5 meaning "very unhappy/I really

don't like it". Scores on this measure range from 20 to 100, with

higher scores indicating greater public speaking anxiety. SAM

ratings and vividness ratings are scored by the computer using a

scale ranging from 1 (low pleasure, arousal or dominance) to 29

(high pleasure, arousal or dominance).

Observational Measures

Children's behavior during dental treatment was scored by

observers using the Behavior Profile Rating Scale (Melamed et al.,

1975). This scale consists of a variety of behaviors reported by

dentists to be disruptive to dental treatment. Each item is

weighted by the degree of disruptiveness it represents. Weights

range from 1 (i.e., inappropriate mouth closing) to 5 (i.e.,

leaving chair). Scores are based on the average score for the time

period in question, ie overall session, injection, rubber dam

placement and drilling. Independent observers, who were blind to

subject group, used an interval sampling procedure to record

behavior. Two minute intervals were used to record the occurrence

or non-occurence of a given behavior. Times were marked on each

videotape using a time-date generator. This helped to maximize

accuracy by allowing the observer to review a difficult portion of

the tape. Raters scored eight practice videotapes before beginning

to make actual ratings. Further, regular discussions were held

with the raters to discuss problems in order to insure accuracy.

Table 1 lists the Spearman-Brown reliability coefficients for the

scores on the Beha -ior Profile Rating Scale. All reliabilities were

in the acceptable range, with the exception of BPRS for the rubber

dam and drilling during the second treatment session.


Session 1 Spearman-Brown
Correlation Coefficient

Overall Disruptiveness .8947*

Injection Disruptiveness .9030*

Rubber Dam Disruptiveness .9257*

Drilling Disruptivenss .9501*

Mean, Session 1 .9184*

Session 2

Overall Disruptiveness

Injection Disruptiveness

Rubber Dam Disruptiveness

Drilling Disruptiveness

Mean, Session 2

Spearm an-Brown
Correlation Coefficient







Data Analysis

The data of this study were analyzed by Analysis of Variance.

All of these ANOVAs included Fear and Training as factors. Fear X

Training X Scene ANOVAs were computed for the SAM ratings.

ANOVAs for the physiological data (heart rate and skin

conductance) also included the factor of imagery period. Thus, the

analyses of the physiological data were Fear X Training X Scene X

Period. These ANOVAs were also repeated for each image period

individually. Thus, a Fear X Training X Scene ANOVA was run for

the read, image and recover periods for both heart rate and skin

conductance. The ANOVAs for the physiological data were repeated

a second time; this time the read period heart rate was used as

the baseline (instead of rest period heart rate). Since resting

heart rate was found to be correlated with age, Analysis of

Covariance was used for all analyses involving heart rate. Data

for the dental subscale, BPRS and dentist ratings were analyzed

with Fear X Training X Session ANOVAs. Comparisons between means

were made using the least significant difference test (LSD test).


Results of this study indicate that children do differentiate

between fearful and neutral imagery, both through their verbal

report of emotion and through their physiological responses.

Following a discussion of a pilot study of the scene contents,

demographic data and data relevant to each of the hypotheses

tested in the study will be presented. First data regarding the

main effects of scene content will be discussed. The effects of fear

level and of imagery training will then be considered.

Study I: Pilot Study of Scene Contents

In order to insure that children viewed the scenes used for

physiological assessment in accordance with the experimenter's a

priori designations as fearful or neutral, a pilot study was

conducted on separate sample of 22 children. There were seven

first graders, eight third graders and seven fifth graders. The

method and complete results of this study are described in

Appendix N. Children were asked to rate their emotional responses,

using SAM, to each of the scenes used in the main study. They

were also asked to rate the fearfulness of each scene. In addition,

children rated the similarity of each scene both to past

experiences and to situations that could happen in the future.

Results validated the manipulation of providing phobic dental and

social (school) scenes as compared to neutral scenes.


There was a main effect of scene on pleasure ratings

(F(5,95)=14.65, p<.0001). The ranking of scenes from most to least

unpleasant, is as follows: principal, dental injection, dental

exam, blackboard, living room and lawn chair. The blackboard

scene was rated the least unpleasant of all four fear scenes. This

scene was rated as significantly more pleasant than the dental

injection scene (LSD p<.05) and the principal scene (LSD p<.01).

The principal scene was rated as significantly more unpleasant

than all other scenes, including the dental fear scenes (LSD



Again there was a main effect of scene (F(5,95)=2.46, p<.04).

The ranking of scenes from most to least arousing is: dental

injection, lawn chair, blackboard, dental exam, principal and

living room. The dental injection scene was rated as significantly

more arousing than the dental exam (LSD p<.05), principal (LSD

p<.05) and living room scenes (LSD p<.01).


Dominance ratings also yielded a main effect for scene

(F(5,95)=5.93, p<.0001). The ranking of scenes from lowest to

highest dominance is as follows: principal, dental injection, dental

exam, living room, lawn chair, and blackboard. The blackboard

scene received significantly greater dominance ratings than did the

dental exam (LSD p<.05), dental injection (LSD p<.05) and

principal (LSD p<.01) fear scenes. The principal (LSD p<.01),

dental exam (LSD p<.01) and dental injection (LSD p<.05) scenes

were rated as significantly lower in dominance than the two

neutral scenes.


There was a significant main effect of scene on fear ratings

(F(5,95)=9.92, p<.00001). The ranking of scenes from most to least

fearful is: principal, dental injection, blackboard, dental exam,

living room and lawn chair. The principal and dental injection

scenes were rated as being significantly more fearful than all

other scenes (LSD p<.01).

Similarity to Past Experience

The ranking of scenes from most to least similar to past

experience is as follows: dental exam, lawn chair, dental

injection, blackboard, living room and principal (main effect of

Scene: F(5,95)=2.79, p<.02). The dental exam scene was

significantly more similar to past experience than the living room

(LSD p<.05) and principal (LSD p<.01) scenes. The principal scene

was significantly less similar to past experience than were the

dental injection (LSD p<.05), lawn chair (LSD p<.01) and dental

exam (LSD p<.01) scenes.

Similarity to Possible Future Experiences

Analyses of Variance on the children's ratings of the likelihood

that each of the situations could happen to them in the future

yielded no significant main effect of scene.

Study II: Tests of Hypotheses

Subject Demographic Data

Table 2 shows demographic data for subjects in each of the

four groups in the study. The high dental fear children had

significantly higher scores on the CFSS-DS across all three

measurement points than did the low dental fear children

(F(1,19)=44.04, p<.00001, LSD p<.01). There were no significant

differences in fear scores between stimulus and response trained

subjects within each fear group.

High fear children in the study were significantly older than

the low fear children 'F(1,20)=6.96, p<.02; 9.75 years vs. 8.25

years, LSD p<.05). It was difficult to control for this, as subjects

were self-selected to fear groups by their report of dental fear.

Further, this age effect is likely to represent the true nature of

this population, as there is a suggestion in the literature that

older children tend to report greater dental fear (Cuthbert &

Melamed, 1982; Winer, 1982).

Parent ratings were consistent with the children' own report of

fear. Thus the high fear children were rated as being

significantly more fearful than the low fear children during

previous dental examinations (F(1,17)=14.17, p<.002, LSD p<.01)

and previous dental injections (F(1,9)=11.43, p<.008, LSD p<.01).

Similarly, the high fear group was rated by their parents as

being significantly less cooperative than their low fear


Stimulus Response

Stimulus Response

AGE (years)


Pre-Treatment 1

Post-Treatment 2







48.57 35.17
(12.39) (10.05)

8.36 8.08
(1.90) (0.87)

17.50 25.40
(2.35) (3.78)

32.40 30.80
(11.59) (12.59)

25.80 26.49
(8.53) (11.48)




*Means for each group are listed.
in parentheses below each mean.

Standard deviations are listed

counterparts during these past examinations (F(1,17)=8.60, p<.009,

LSD p<.01) and dental injections (F(1,9)=4.66, p<.06, LSD p<.05).

As was the case with the child's report of dental fear, parent

ratings of fear and cooperation were also correlated with age.

Thus older children were rated as being more fearful during both

examinations and injections (r=.5469, p<.02 and r=.7951, p<.01

respectively), and as less cooperative during injections (r=-.7758,

p<.01). The dentists' ratings of fear and cooperation were

unrelated to the children's fear level. This is an interesting

result, which suggests that dentists' perceptions of their child

patients' fearfulness is not always in accordance with the child's

own self-report of fear.

The high dentally fearful children also reported greater general

fearfulness as measured by the CFSS (F(1,19)=9.63, p<.006, LSD

p<.01). The high fear and low fear groups did not differ on the

number of children who had previous experience with dental

treatment. The groups did not differ in scores on the Questionnaire

Upon Mental Imagery or on a subset of questions from the CFSS

measuring social anxiety. There were no differences between groups

on the number of days between experimental sessions (chi-squared

test did not reach significance).

Imaginal Responses to Affective and Neutral Content

Analyses of the main effects of scene content revealed that the

subjects did discriminate between fearful and neutral contents.

Fear scenes were rated as more unpleasant, and as lower in

perceived dominance, than were neutral scenes. Further, the

subjects showed a differential pattern of heart rate responding to

the two types of scenes. Specifically, children showed heart rate

acceleration during imagery of fear scenes and deceleration during

imagery of neutral scenes. Skin conductance responses also showed

differentiation between scene contents, with higher read period

skin conductance levels occurring to the fearful scenes. Tables of

means for affective ratings, heart rate and skin conductance can

be found in Appendix 0. Correlations between affective ratings and

physiology are also included.

Affective ratings

Pleasure. All four fear scenes were rated as significantly less

pleasant than the two neutral scenes (main effect of Scene:

(F(5,100)=23.49, p<.00001; LSD p<.01). The principal scene was

rated as more unpleasant than all other scenes (LSD p<.01). The

dental examination, blackboard and dental injection scenes (listed

in descending order of pleasantness) received intermediate ratings

on the pleasure dimension. There was a positive correlation (r=.78,

p<.01) for pleasure ratings on the dental examination and dental

injection scenes. Thus, children giving low pleasure ratings to one

dental scene tended to give low pleasure ratings for the other

dental scene.

Arousal. No significant differences between scenes were obtained

for arousal ratings. It is possible that the children did not

understand the concept; observations suggested that some subjects

may have associated "excited" with "happy". Review of the data

indicates that the pattern of ratings was not consistent with the

designations of scenes as fearful and neutral.

Dominance. As was the case with pleasure ratings, dominance

ratings differed for fearful and neutral scene contents (main effect

of Scene: F(5,100)=9.26, p<.00001). The ranking of scenes on the

dominance dimension closely parallels that for pleasure. Scenes

rated as most unpleasant tended to receive lower ratings of

controllability. Thus, the ranking of scenes from highest to lowest

dominance ratings is lawn chair, living room, blackboard, dental

exam, dental injection and principal. The dental exam, dental

injection and principal scenes were rated as significantly lower in

dominance than the lawn chair and living room scenes (LSD p<.01).

These three fear scenes did not differ significantly from one

another, nor did the two neutral scenes. There were significant

positive correlations between ratings of pleasure and dominance for

three of the scenes: dental injection (r=.4424, p<.05), blackboard,

(r=.5800, p<.01) and principal (r=.4127, p<.05). These correlation

suggest that children do not view pleasure and dominance as

distinct emotions, but rather that they discriminate only between

"good" and "bad". Significant correlations between the ratings

averaged across all six scenes support the notion of a lack of

independence between these emotional dimensions (pleasure &

dominance: r=.4368, p<.05; dominance & arousal: r=.4743, p<.02;

pleasure & arousal: r=.2119, ns).

Vividness. There were no significant effects of scene content on

vividness ratings.

Heart rate

Older children had lower resting heart rates. Negative

correlations between age and rest period heart rates are as

follows: lawn chair r=-.4987 (p<.01), living room r=-.4114 (p<.05),

dental exam r=-.5676 (p<.01), dental injection r=-.5805(p<.01),

blackboard r=-.5362 (p<.01), principal r=-.4993 (p<.02). Therefore,

age was used as a covariate in all analyses involving heart rate.

When faced with an imagery task, adults typically show heart

rate acceleration during the reading of the scene, further

acceleration during the imagery period, and deceleration during

the recovery phase. Figure 2 shows that children in the present

study tended to show heart rate deceleration during the read

period (Scene X Period interaction: F(10,200)=2.14, p<.02). Subjects

showed a significant deceleration during the read period of the

principal scene (LSD p<.01). This deceleration was significantly

greater than that occurring to all other scenes (LSD p<.05 for

difference between principal and dental exam and dental injection;

LSD p<.01 for all other differences). However, as with adult

subjects, within scene analyses showed that significant heart rate

acceleration occurred for the dental injection scenes from the read

to the recovery period (LSD p<.05). Similarly subjects showed heart

rate acceleration from the read to image periods of the blackboard

scene (LSD p<.05) and from read to recover for the principal scene




o (

CU ("






(LSD p<.05). Heart rate deceleration is typically associated with

attention (Lacey, 1967). Thus, these data suggest that unlike

adults, who show affective responses during the read period,

children need time to attend to the stimulus and then to access the

emotional network related to the scene being presented before they

can show any affective response. Children use the read period to

attend to the material and to access the appropriate network;

affective responses can then occur during the image and recovery

periods. For children then, the read period appears to represent

processing that is quite different from that occurring during the

image and recovery periods. Since imagery for children is a dual

process task, all subsequent analyses of physiological responding

will use the read period physiology, rather than rest period, as a


Figure 3 shows that when the read period is used as the

baseline, children showed heart rate increases to the fear scenes

and heart rate decreases to the neutral scenes (main effect of

Scene: F(5,100)=2.84, p<.02). There were no differences between the

image and recovery periods. An examination of the data for the

read and image periods separately reveals that this main effect is

due to the recovery period (main effect of Scene for recovery:

F(5,100)=3.44, p<.007). The heart rate increment during the dental

exam, dental injection and principal scenes was significantly

different from the heart rate decrement occurring to the two

neutral scenes (all LSD p<.05).

o ID 0
ei -










- I

non u

go o o
1, 0 U


Skin conductance

The heart rate data indicated that the imagery task is

comprised of two processes: mental processing and accessing of the

emotional network during read, and affective responding during the

image and recover periods. Thus, skin conductance data were

examined separately for the read period, using rest as the

baseline, and for the image and recovery periods using read as

the baseline. During the read period, subjects did show some effect

of the affective content (marginal main effect of Scene:

F(5,100)=1.97, p<.09). Skin conductance decrease was less for the

fear scenes than for the neutral scenes. Using the read period as

baseline for the image and recovery periods reveals no further

effects of scene on skin conductance. The habituation typically

seen across imagery periods did not occur. This suggests that the

affective content of the scenes did have an effect on physiology.

Although there was no elevation in skin conductance levels during

image and recovery, the absence of a decrease suggests that the

affective nature of the scenes may have prevented the habituation

that is often observed. This is consistent with the results of other

research (Lang et al., 1983).

Effects of Fear Level and Imagery Training

In addition to the hypothesis that children would show

differential responding to fearful and neutral imagery, it was

hypothesized that physiological responses to fear scenes would be

greater in fearful than non-fearful children. It was also predicted

that physiological responses would be greater in response trained

children than in stimulus trained children. An interaction between

fear level and imagery training was also expected. That is, it was

predicted that responsiveness would be greatest in high fear

children who had received response training. Dividing the sample

into groups based on fear level or training reduces the sample

sizes; thus, these results must be viewed with caution.

Dental fear

Scores on the Dental Subscale at the three measurement points

(Screening, pre-Treatment 1 and post-Treatment 2) were all

positively intercorrelated (r=.5369, p<.01 and r=.5311, p<.05

between Screening and Treatments 1 and 2 respectively; r=.6016,

p<.05 between Treatments 1 and 2). The high fear children did not

show any significant changes in reported dental fear across the

study. The low fear children, however, showed a significant

increase in reported dental fear from the screening to

pre-Treatment 1 (Session X Fear: F(2,38)=3.84, p<.03; LSD p<.01).

There was a decline in fear level from Treatment 1 to Treatment 2,

however this fear level is not significantly different from that at

the screening or at Treatment 1. The low fear children reported

higher levels of dental fear when actually faced with dental


Affective ratings

Pleasure. High fear children rated the two dental scenes as more

unpleasant than did low fear children, with response trained

children giving lower pleasure ratings than stimulus trained

children within each fear group (Scene X Fear X Training:

F(5,100)=2.16, p<.06). Ratings given by the high fear response

trained children were significantly different from those given by

the low fear stimulus trained children (LSD p<.01). The high fear

response trained subjects rated all four of the fear scenes as

being significantly more unpleasant than the two neutral scenes

(LSD p<.01). The high fear stimulus trained subjects rated the

principal (LSD p<.05), dental injection (LSD p<.01), and dental

exam (LSD p<.01) scenes as significantly lower in pleasure than

the lawn chair scene. Low fear stimulus trained subjects made

some differentiation between fearful and neutral scenes as well,

rating the principal (LSD p<.01), blackboard (LSD p<.05) and

dental injection (LSD p<.05) scenes as more unpleasant than the

two neutral scenes. The differentiation between fearful and neutral

content was less clear for the low fear response trained subjects.

These subjects did rate the principal (LSD p<.01) and dental

injection (LSD p<.05) scenes as lower in pleasure than the lawn

chair scene.

Dominance. A similar effect was obtained for affective ratings of

dominance (Scene X Fear X Training: F(5,100)=2.39, p<.04). Figure

4 shows that the high fear response trained subjects made the

strongest discrimination between fearful and neutral content, so

that ratings for all four fear scenes were significantly lower than

those for the two neutral scenes (LSD p<.01 for all differences).




................... : .. ...*.*. .*.-
... ... ... .. ... ... ... ... ... ... ... ... ....... .. .. .
........... ....................... .. ......

1-1 .. .. .. .. .. .. .. .. .. .
. . .
.. . . . .
.. .. .. . .. . .... ;. *..;



........ ..* II ..........;

. ... ...
.. .
.... ... .... .
.... .. ..





s_ .

Ca 0


C L. 0



- 0

a. La


& i5b







The high fear stimulus trained group also made some

differentiation between fearful and neutral scenes on dominance

ratings. For this group, the ratings for the dental injection and

principal scenes were significantly lower than those given for the

lawn chair scene (LSD p<.05). The high fear, response trained

group reported the least dominance for all four fear scenes. Low

fear subjects reported the highest dominance for each fear scene.

Response training consistently produced lower dominance ratings for

fear scenes in the high fear subjects; however, there was no such

effect within the low fear group.

Heart rate

None of the hypothesized effects of fear level and imagery

training on heart rate were obtained. However, imagery training

was seen to have an effect on heart rate responses when

self-reported imagery ability was taken in to consideration. All

good imagers and the poor imagers with response training showed

heart rate declaration during imagery (across periods and across

scenes) while poor imagers with stimulus training showed heart

rate acceleration (QMI X Training interaction: F(l1,18)=10.10,

p<.005). Although LSD tests did not reveal any significant

differences between the means, these data suggest that response

training serves to make poor imagers respond in a manner similar

to that seen in good imagers.

Skin conductance

Fear level and imagery training interacted to produce an

interesting effect on read period skin conductance (Scene X Fear X

Training: F(5,100)=2.21, p<.06). The subjects in the high fear

response trained group were the only ones to show increased skin

conductance to the reading of the dental fear scenes and decreased

skin conductance to all other scenes (see Figure 5). The skin

conductance increase occurring to the dental exam scene was

significantly different from the decrease occurring to the lawn

chair and living room scenes (LSD p<.05). This effect represents a

more microscopic view of the main effect of scene on read period

skin conductance that was described earlier. Recall that skin

conductance decreases were less to fear scenes than to neutral

scenes. Here it can be seen that when children report a high

degree of dental fear and receive response training, they actually

do show increased skin conductance to the presentation of dental

fear scenes.

Dental operator behavior

There were no differences between fear groups for

disruptiveness during the injection at Treatment 1. However, at

Treatment 2 the high fear subjects showed significantly greater

disruptiveness than the low fear subjects (Session X Fear

interaction: F(l,19)=4.09, p<.06; LSD p<.05). This effect appears to

be due to the fact that the high fear group showed a slight

non-significant increase in disruptiveness across sessions while the

- 00
Q- C5 o 5

ll.i; 'I
*"'- ":






r4 *C
e .




o 0 M v $D 0