Follow your head (and your heart?)


Material Information

Follow your head (and your heart?) cardiac and motor indices of anticipation in 15-month-old infants in a two-alternative, cued location memory task
Physical Description:
vii, 114 leaves : ill. ; 29 cm.
Garner, Elizabeth E
Publication Date:


Subjects / Keywords:
Psychology thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Psychology -- UF   ( lcsh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1997.
Includes bibliographical references (leaves 112-113).
Statement of Responsibility:
by Elizabeth E. Garner.
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 - 028020434
oclc - 37851400
System ID:

This item is only available as the following downloads:

Full Text







I would like to extend my sincere gratitude to several individuals who helped to

make the completion of this dissertation a reality. First, I would like to acknowledge the

numerous parents and infants in the local community who generously contributed their

time and effort in order to participate in the research.

In addition, I would like to thank Glen, Ilene, Dawn, Stacy, Maylie, Evan, Hailey

and many other undergraduates who spent hours of their time viewing videotapes and

coding motor behaviors, a difficult and tedious task which was crucial to data analysis on

this project. In particular, I would like to thank Jana Axelrad, a budding scientist of her

own, whose vigilance on this task helped to ensure the integrity and accuracy of the motor

data. I wish her good fortune in her future research.

I would like to extend a special thank you to Amy Boswell, my right-hand

throughout the course of this research, for the enormous amount of time and effort she

contributed in assisting me with infant recruitment, testing and data processing. Your

competence, encouragement, light-hearted spirit and sense of humor proved to be vital in

the success of this project and actually made the countless hours of infant testing

enjoyable! I truly could not have done it without her.

Of course none of this would be possible without the intellectual guidance of my

academic advisor, Dr. Keith Berg, whose invaluable insight, time and wisdom have largely

influenced the quality of my work. Keith continues to serve as a vital role model in my

professional development as I strive to emulate the high degree of scientific integrity,

creativity and competence that he invariably maintains. In addition, I would like to thank

him for his patience, optimism, and encouragement, as well as his general tolerance of my

anxiety level near completion of this project (keeping the lab refrigerator stocked with diet

Dr. Peppers was a plus too!). Thank you to my doctoral committee for your time and

input on the project as well as numerous others whose ideas and discussion helped to

generate solutions for methodological and procedural challenges.

Finally, I would like to thank Michael, our children (Brittany and Karen) and other

extended family members for their patience, love and support as I subjected them to the

psychological "fall out" from the enormous demands of this project.



ACKNOW LEDGM ENTS ................................................... ..................... ii

A B STR A C T ......................................... ............................ ................... vi

IN TR O D U CTION ................... .......................................................... ... 1

REVIEW OF LITERATURE.................................................. .....................4

METHODS -- STUDY 1................... ...................12

Participants.............. .........................................................12
Stimuli...................... ........................... ..............12
Apparatus............... .........................................................14
Data Collection............ ..................................................... 16
Procedure............... .........................................................18

RESULTS -- STUDY 1................. ... ...................23

Heart Rate.. ............. ..........................................................23
M otor Data.................................................................................... 25
Head Position..... ..... .....................................................25
Reaching Behavior...... ......... ....................29

DISCUSSION -- STUDY 1..........................................39

H eart R ate...................................................... ................................. 39
M otor B ehavior..................................................................................... 41

M ETHODS -- STUDY 2...................................................... ....................45

Participants .................................................................47
Stim uli...... ........................................................................................... 47
D esig n ................................... ................................................ .... 4 9
D ata Collection................................................................ ............ 50
Procedure.................................. ....... ..................... .. 53


RESULTS -- STUDY 2.............. ..................................................56

H heart R ate.................................................. .......... ............. 57
Anticipatory HR and Accuracy ................................... ....................59
M otor Behavior.................... ...... ............. ... ....... .................... 62
H ead Position............................ .................................................... 62
Search Behavior..... .................... .................... 67
Anticipatory Motor Behavior and Accuracy.................................69

DISCUSSION -- STUDY 2................................................ ....................86

Generalized Anticipation (Heart Rate)..................... ....................86
Differential Anticipation (Mot6r Behaviors)..........................................95
Anticipation as a Distinct and Functional process....................................98

CO N CLU SION .............................. ................................ ....................... 106

A P PE N D IX .......................................................................... .................... 109

REFERENCES...................... ....................................... 112

BIOGRAPHICAL SKETCH................................................. ...................114

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


Elizabeth E. Garner

August, 1997

Chairman: Dr. W. Keith Berg
Major Department: Psychology

The ability to anticipate the spatial location of a hidden object was investigated in

15-month-old infants in a cued-location spatial memory delay task. Of interest was

whether, prior to solving the task, infants would exhibit anticipation as indexed by

decelerations in heart rate (as they prepared to search for the hidden object) or in the form

of overt head turns and/or reaches toward the anticipated location of the object. Two

studies were conducted to investigate these possibilities: the first involved a pilot study

designed to develop testing procedures for eliciting and coding reaching behaviors

associated with object retrieval. The results suggested that infants exhibited HR

decelerations and were capable of reaching for the anticipated object as it was presented

to them. They did not, however, exhibit anticipatory behaviors prior to the object's

appearance. The procedure was modified in a second study to better elicit anticipation as

infants searched for the hidden object in the cued-location memory task. An attractive toy

was hidden in one of two opaque containers (to the right or left of the infant). The infant

was distracted from the hiding location of the object and then allowed to search one of the

two containers. The results indicated that, prior to the infants' search, there were

pronounced heart rate decelerations and head turns toward the "target" (correct)

container. The latter, in particular, provided strong evidence of differential anticipation of

toy location. In general, the infants' searches were accurate significantly more often than

not, indicating that they did, in fact, retain the spatial location of the hidden toy.

Furthermore, the orientation of the infants' head just prior to retrieval was a significant

predictor of their overall accuracy suggesting that the overt anticipatory behaviors

themselves may have been functional or strategic in facilitating performance on the

memory task.


One of our most important cognitive facilities is the ability to efficiently focus

attention toward future events, what Haith has called future-oriented behavior (Haith,

Wentworth & Canfield, 1992). The old adage, "forewarned is forearmed" epitomizes the

notion that the ability to plan or anticipate a crucial event allows us to more effectively and

efficiently deal with that event, on both cognitive as well as behavioral levels.

Future-oriented cognitive processes are an important area of study in their own

right but the development of these abilities is particularly intriguing because they seem to

be at the root of many exciting, emerging fields of research involving information

processing in infants and children. One such area of cognitive development involves what

Bidell and Fisher (1994) define as "on-line planning", or the ability to store a mental

representation of a series of action sequences in working memory in order to anticipate

and guide performance on a given task. For example, when presented with a game such as

"The Tower of Hanoi", children must plan a sequence of moves, in advance, in order to

stack a series of blocks on pegs in a particular arrangement with a minimum number of

moves. Similarly, the simple retrieval of a hidden object executed by an infant requires

memory of the object's location as well as goal-directed searching toward that anticipated

location where the object can be retrieved. There are, in fact, several examples which

could easily illustrate the necessity of the role of future-oriented thinking in effective

problem-solving in a variety of tasks which assess some form of cognitive development. In

general, when engaged in any sort of task or game which requires strategy (for example,

chess), the ability to think ahead before executing a response provides for the most

adaptive response. Thus, future-oriented thinking is significant, theoretically, because it

seems to be a vital component--either explicitly or implicitly--in information processing

models which strive to outline the mechanisms underlying the development of strategy

learning and problem solving in children.

As Bidell and Fisher (1994) note, however, there is a relative dearth of research on

the planning process itself which is surprising given its relevance to so many areas of

cognitive development. Furthermore, as limited as this area of research is in children, even

less is known about anticipation and problem solving in infancy. This is ironic given that

even Piaget (1952) and Uzbiris and Hunt (1975) suggest that is during mid to late infancy

that anticipation may first emerge. Thus, it seems that the examination of the anticipation

process in its relatively simple state during infancy has much potential to expand what we

know about infant cognition generally as well as to provide insight into the subsequent,

and more sophisticated, forms of planning which emerge later in development.

Examination of anticipation during infancy is also worthwhile because of the

implications for neurological development. Bidell and Fisher (1994) note that there is an

accumulating body of evidence from neurologically-oriented research with both primates

and human infants which implicates the frontal cortex--an area known to undergo

significant development during infancy in terms of synaptic density--in the developing

ability to process information in working memory. Most relevant to the execution of

future-oriented cognitive processes such as "on-line planning" or anticipation, however, is

Goldman-Rakic's (1987) research with rhesus monkeys in which she demonstrated that

firing in specific cells in the prefrontal cortex was associated with the monkey's

performance on a hidden object task. If the cells did not fire, for example, the monkeys

could not remember the location of an object hidden under a particular cloth, suggesting

that these cells may play a vital role in the ability to store the representation of the object

location and to anticipate where to retrieve it. Thus, examination of the anticipation

process in conjunction with similar simple problem solving in infants not only sheds light

on the antecedents of more elaborate forms of problem solving in childhood but may,

indirectly, provide a glimpse into the development of the neural substrates that support

future-oriented processes in humans.

The present study attempts to examine this issue by assessing anticipation in 14

and 15-month-old infants, as they prepare to solve a simple problem--a forced-choice,

paired stimulus, cued-location delay task. Specifically, anticipation will be measured by

both covert (eg, non-observable) physiological responses, such as moment to moment

changes in heart rate, as well as overt (e.g., observable) behaviors, such as head turns and

searching behavior toward the correct location of a hidden object, which occur as infants

prepare to solve the hidden object task. In so doing, various aspects of the anticipation

process can be examined as infants actively prepare to solve a simple problem--

remembering the spatial location of a hidden object and acting upon that information to

retrieve the hidden object.


The process of anticipation is interesting in its own right and has been studied

extensively under much simpler conditions in both children and adults. For example, given

a warning stimulus, adult subjects readily anticipate the arrival of a cued event presented

after a predictable delay period. This form of anticipation is typically measured using a

"fixed-foreperiod" or "S1-S2" paradigm. In general this involves the initial presentation of

a warning stimulus (Sl) which is followed, after a fixed interval, by a second stimulus (S2)

which is of critical relevance to the subject. S2 may be inherently interesting or

reinforcing, or may serve as a cue to initiate an explicit, instructed response (for example,

a speeded motor response).

During the S1-S2 delay (usually 4 10 s), or foreperiod, participants may do

several things which reflect anticipation of S2's arrival. These include behavioral

indicators, such as head turns toward the expected location of a stimulus as well as several

physiological responses which occur during the delay interval or foreperiod.

Heart Rate (HR) change is one physiological response that is well suited for

measuring anticipation in the foreperiod primarily because it is sensitive to the focus of

attention and can track attention in a continuous, ongoing fashion. Bohlin and Kjellberg

(1979) have reviewed studies of adults tested in the fixed-foreperiod paradigm and have

described the typical HR response which occurs during the anticipatory, S1-S2 interval.

As these authors indicate, the complete triphasic response consists of an initial

deceleration (Dl) followed by an acceleration (A) and, finally, a second deceleration (D2).

Although two decelerations occur, their functional significance differs greatly. D1 is an

orienting response to the onset of S It occurs after S is presented and is more apparent

on the early presentations of S1. In contrast, D2 reaches its nadir or peak just prior to the

arrival of S2, and may take several trials to appear. It is the D2 response, then, that has

been hypothesized to indicate anticipation of S2 (Bohlin & Kjellberg, 1979; Coles &

Duncan-Johnson, 1975; Walters & Porges, 1976). The cognitive correlate of the middle

acceleration is less clear although it has been hypothesized to reflect various information

processing or metabolic demands of the task (Coles & Duncan-Johnson, 1975; Walters &

Porges, 1976; Chase, Graham & Graham, 1968).

This paradigm has been used with infants and children as well and robust

anticipatory decelerations have been documented in infants as young as 7 months of age

(Donohue & Berg, 1991). However, the conditions under which these infants were tested

were relatively simple--with the exception of two trials an animated toy predictably

followed a tone. However, in an attempt to demonstrate that anticipation was, in fact, due

to attention focused on upcoming S2 per se (and not simply sensitization or other non-

anticipatory behavior), Donohue and Berg (1991) then tested same-aged infants in a more

complex S1-S2 paradigm, parallel to that typically used in differential conditioning.

In the first of two sessions, a tone (SI) was paired with a toy (S2), 6 seconds later

(S1+ trials). In a subsequent session the same sort of Sl+ trials were intermixed with

trials on which a different tone was presented alone (Sl- trials). No toy followed on S1-

trials. Although the HR patterns were less clear than those which typically develop during

10" ISIs, the HR nevertheless decelerated prior to S2, indicative of anticipation. More

importantly, it appeared to do so in a selective manner evidenced by larger HR

decelerations in the second session to S1+, than to SI-, trials. Thus, 7-month-old infants

appeared to be able to differentiate cues containing information about whether an event

was coming or not. Essentially, the Donohue and Berg (1991) data suggest that infants

can anticipate under cued, GO--NO GO, conditions.

The proposed study is designed to take this investigation one step further by

examining whether infants can anticipate under two-alternative, forced-choice conditions.

Like the Donohue and Berg (1991) study, two different S1 stimuli are employed, but

unlike their paradigm, here each S is paired with a salient, meaningful event -- retrieval of

a hidden toy in the spatial location cued by S1. The infant will be required not only to

anticipate that S2 is coming and when it is due to arrive, but, based on the the cue given,

where S2 will be presented. For example, cue A predicts location A while cue B predicts

location B.

In GO--NO GO, differential anticipation tasks such as that used by Donohue and

Berg (1991) HR can readily distinguish GO versus NO GO cued trials because the HR

decelerations will only occur if the infant anticipates that S2 will be presented. However,

in the proposed paradigm, whether S2 will occur should not be an issue since S2 will

follow both cues. Which of two alternative locations it will appear at is the distinguishing

factor. Unfortunately, HR alone is not able to discriminate anticipation of S2's arrival in

one location versus another. Rather, HR is limited to the temporal qualities of the

stimulus events. Thus, additional response measures will be needed that are sensitive to

spatial information.

Head orientation and reaching behaviors are both good candidates because each

has the potential to extract the much needed information regarding where the infant

expects the stimulus event to occur. Employing these measures in an anticipation context

is not new. For example, Haith et al. (1992) examined infants' head turns toward the

expected spatial location of a stimulus event which alternated from the left to right visual

field, but the delay intervals utilized were much shorter than those implemented by

traditional fixed-foreperiod anticipation tasks. Furthermore, Haith's paradigm does not

elicit anticipation based on differential cues (or Sis) per se. Rather infants simply learn

the pattern of the alternating stimulus presentation sequence.

With regard to reaching behaviors, research in this area has demonstrated, for

example, that infants will modify the orientation of their hand in preparation for grasping a

horizontal versus a vertical dowel, an ability which appears to improve with age

(VonHofsten & Fazel-Zandy, 1984; Lockman, Ashmead & Bushnell, 1984). Not only did

the ability to modify the reach improve markedly from 18 to 34 weeks of age in the infants

in this study, but the point at which the modifications were made also developed with age.

For example, whereas the 5-month-olds adjusted their hands only after having made

contact with the dowel, by 9 months of age, infants adjusted the orientation of their hand

early, prior to contact, suggesting better anticipation of the target object's characteristics.

As a result, the older infants seemed to be able to use anticipation to more effectively

control their motor behavior.

Although anticipatory changes in motor behavior are clearly in place by 15 months

of age, the above paradigms are limited in that all anticipatory responses are in relation to

an object that is perceptually available. That is, unlike in the S1-S2 paradigm (in which S2

is not visible during the interim delay) anticipation was not subject to any demands on

memory or based on any type of mental representation of the anticipated stimulus. Rather,

the adjustments were made in the presence of the target stimulus, suggesting that this

behavior may involve nothing more than simple hand-eye coordination, a less cognitively-

oriented ability.

A closer parallel to the S1-S2 paradigm utilized by Donohue and Berg (1991)

would have to require motoric adjustments made while preparing to grasp an unseen

object. This arrangement would yield more compelling evidence that, as with HR, motor

behavior can also be used as an index of anticipation of a mentally stored, future event.

Clifton, Rochat, Letovsky and Perris (1991) arranged a paradigm which met this criteria

using 6 1/2-month-old infants. Using a within subject design, one of two tones,

equidistant from the infant, sounded as a small or a large, object approached. The infants

were first given 8 trials with the lights on allowing them to see the object as they were

preparing to grasp it, as with the VonHosten and Fazel-Zandy (1984) studies. These trials

were followed, however, by several trials carried out in the dark. Unlike light trials, under

dark conditions the infants had only the differential sound to guide the manner in which

they prepared to grasp either the small or large object.

The results indicated that under both light and dark conditions, the infants used the

tones as cues to modify their grasp in anticipation of the object's size. For example, after

hearing the tone associated with the large object, infants reached with both arms, with

their hands open wide. Alternatively, after hearing the small object tone, they tended to

reach with only one hand, with their hands slightly closed. On light trials, this adjustment

is similar to the modifications in hand orientation described earlier since the small or large

target object was visible as the tones were sounded. Similar anticipatory modifications

made in the dark, however, can not simply be based on perceptually-guided preparation.

Rather, Clifton argued that this behavior was more clearly dependent on the infants' mental

representation of the anticipated object's size. Essentially, the type of responding executed

in VonHofsten and Fazel-Zandy's paradigm can be thought of as a "perceptually-based"

form of anticipation, in which adjustments in behavior are guided by a perceptually

available stimulus. In contrast, Clifton et al.'s scenario more clearly relies on a stored

mental representation of the anticipated stimulus--a more sophisticated, cognitive, ability.

Given the dependence on mental representations, this latter form of anticipation will be

referred to as "memory-based" anticipation and will include those adjustments in behavior

which are made without perceptual access to the anticipated stimulus event.

The proposed studies will expand on the work done by Clifton et al. (1991) in

three important ways. First, although Clifton et al. (1991) used object size to differentiate

the two anticipated stimuli, they were always presented in the same location. Thus,

relatively minor adjustments were required to grasp the object of interest. Here, "S2" will

be differentiated by spatial location, requiring gross adjustments in the trajectory of the

infants' reach. Second, the SI (sound) and S2 (approaching toy) were presented

simultaneously by Clifton et al (1991) alleviating the additional memory demands imposed

by a temporal delay between cue and event. Here, the presentation of discrete stimuli will

allow the examination of anticipation across a longer delay; conditions which impose a

greater demand on memory. Finally, the addition of head turn behaviors as well as

moment to moment changes in physiological measures during the task allows a more well

rounded examination of both internal as well as overt manifestations of the anticipation


In addition, the relationship between each of these two systems (ie, cardiac and

motoric) and the infants' performance on the task will be of interest. Berg and Donohue

(1991) argue that an important element of anticipation is its degree of "functionality," that

is, whether the ability to anticipate or plan for something serves any adaptive purpose.

The motor behaviors are of particular interest in determining whether the ability to exhibit

differential anticipate of the object's location may be used as a strategy which facilitates

memory for the location and ultimate retrieval of the object. Thus, whether infants with

larger HR decelerations or anticipatory head turns and/or reaches are better able to

execute the proper motoric responses to ultimately retrieve the toy may touch on the

functionality of anticipation in this task

In conclusion, the proposed studies are designed to expand upon the infant

anticipation literature in several ways. First, the use of longer ISIs as well as the addition

of HR measures expands upon paradigms typically employed in cued-location tasks by

allows examination of the ability to utilize spatial cues over longer delays. Second,

although cardiac components of anticipation have been demonstrated in simple, paired-

stimulus paradigms, the present study will attempt to elicit anticipation under more


complex stimulus conditions requiring discriminate learning as well as spatial memory.

Finally, the addition of motor behaviors not only allows a more well-rounded examination

of anticipation but may provide insight into the functionality of anticipation in a rather

sophisticated, cognitive task.

Two studies are proposed in an attempt to fulfill the aforementioned goals. The

first will serve primarily as a pilot study in which both testing and coding procedures can

be developed to ensure anticipatory reaches can be obtained under simple conditions. In

contrast, it is the second study which will involve the discriminate learning spatial task and

is designed to yield new information on the unexplored, aforementioned facets of memory-

based anticipation.



All parents in the city of Gainesville with infants at least 7 months of age were

identified via birth certificate records and were recruited for participation via letters

describing the research study. Included in the letter was a business reply card interested

parents could return indicating their willingness to participate in the study. The final

sample of infants consisted of twenty-four, 9-month-old infants (Mean age = 9 months, 3

weeks), who were primarily Caucasian and from middle-income homes. For various

reasons not all infants who were initially tested (N = 57) contributed data to all response

measures. Specifically, whereas there were 24 infants with usable HR data, fewer infants

contributed data to the motor analyses due in large part to greater difficulties in coding (N

= 16 for reaching behaviors and N = 14 for head position).


Several significant stimulus events occurred within each 26 sec trial (see Figure

la). These events will be described in two sets, those used by the experimenter and those

presented to the infants. The first set of stimuli involve those used solely to provide the

experimenters with a series of temporal cues to aid them in presenting the various stimulus

events to the infant at the correct time. All of the experimenter's cues were presented via

headphones and were not audible to the infant. The first cue consisted of a 1 sec tone,

presented at second 5, which signaled the experimenter to jingle the bell and raise the

warning flag. The second cue, a 1 sec tone presented at second 11, prompted the

experimenter to begin rolling out the carriage with the toy attached. The third stimulus

was a 6 sec, broad-band white noise stimulus secss 12 to 18), intended to pace the

experimenter as she rolled the carriage out to the end of the track and back such that it

reached the end of the track mid stimulus and was pulled all the way back by the noise's

offset. These stimuli, especially the latter, were crucial in maintaining temporal

consistency over trials given that the apparatus was manually operated. For example, if

the experimenter noted a discrepancy between the offset of the white noise and the

position of the containers on the track, that particular trial could be excluded.

The second set of stimulus events was directed to the infant. Two main stimulus

events were presented. The 14s warning stimulus (conceptually similar to a "S ")

consisted of a 6 X 6 in square piece of foamboard with a checkerboard (6 x 6, 1" checks)

pattern attached to it. This panel was attached to the end of a 9.5 wooden stick so that,

when cued, the experimenter could rotate the stick and foamboard to a vertical position,

exposing the entire checkerboard panel just above the curtain, directly across from the

infant, at the back of the table (see Figure 2a). To capture the infant's attention, the

raising of the flag was accompanied by the brief sounding of a jingle bell, on the same side

that S2 would be presented. This auditory component of SI ended, approximately 1 sec

later, once the checkerboard was in place.

The S2 event consisted of the presentation of one of several bright, colorful, rattle

toys. The toys were attached with velcro to a 3.5" x 1.5" flat piece of wood. These

"carriages" were each attached to a track extending from behind the back curtain to within

12 of the front end of the table. The carriages were designed so that the experimenter

could attached a toy to one of them and could move the toy down one of the two tracks

toward the baby.


The apparatus consisted primarily of a 28" high table (36" deep x 42" wide)

surrounded on three sides by curtains (see Figure 2a). The 8" high back curtain extended

the entire width of the table, placed 26" from the table's front edge. The two side curtains,

extending from the back curtains to the front edge of the table, were considerably taller

(19") to reduce the amount of peripheral distractions in the room. A semicircular notch

(14" wide X 6.5" deep) was carved out of the front edge of the table at midline to allow

the infant to fit snugly against the table. This both reduced the amount of time the infant

spent looking under the table as well as allowed better contact with the toys on the top of

the table.

Two tracks, constructed from modified, traverse curtain rods, were placed on the

table's surface. They extended, at a 49 degree angle, from behind the back curtain (one on

the right side, one on the left), such that the two tracks converged as they approached the

front end of the table (see figure 2a). The wooden carriage was affixed to the track such

that it could be moved by a series of ropes and pulleys toward and away from the infant,

allowing the toy to be moved smoothly from behind the curtain, to within reach of the

infant, and back again.


The only between subjects factor was the S2 presentation side (for half of the

subjects S2 was presented on the right and for half, on the left). Three dependent

measures were of interest: 1) HR 2) head position and 3) the infant's attempts to touch

and retrieve S2. Twenty trials were presented and were blocked into 5 blocks of 4 trials

each for all response measures.

In general, trials were excluded if the infant's state was unacceptable (excessive

fussiness or crying), if the infant was distracted for than a few seconds, or if there were

experimenter errors involving timing (if a stimulus event was off by more than 2 sees) or

the apparatus. In addition, trials were excluded from HR analyses only if there was

excessive fluctuation in the interbeat intervals or the data were otherwise contaminated

during recording. Similarly, most trials excluded from motor analyses were a result of

videotaping errors (for example, one parent blocked the view of the infant's head) or cases

in which the parent interfered with the infant's reaches. In general, if more than two trials

on any block were excluded, the data from that infant was not included in the study.

However, for the HR analyses, if less than two trials per block were deemed "bad", then

the data for lost trials were replaced by the average of the data in the trial before and after

the "bad" trial, provided the replacement trials were in the same trial block. If the "bad"

trial was the first or last trial of a particular block, then the data was replaced by the

nearest good trial in that same block. Since motor responses were expected to be less

stable from trial to trial, bad trials were not replaced with data from other trials. Rather,

"bad" trials were simply omitted from analysis and percentages of motor responses were

taken from the total of "good" trials only.

For the HR analyses, this design generated two within subject variables: trial

Blocks (1 5) and Seconds. Although 26 sec of HR were recorded, only two segments of

this data were of interest: the 6 s prior to the emergence of the toy from behind the

curtain (S1 Alone Interval) and an additional 6 sec during which the infant had an

opportunity to retrieve the toy (S2 Visible interval).

Head position was the primary anticipatory motor response of interest. Much like

HR, head position was recorded every half second throughout the trial but, for purposes

of analysis, is restricted to the analogous two, 6 sec periods of time. The two within

subject variables of interest in this analysis were Blocks and Seconds.

Finally two sets of analyses involving touching and retrieving behaviors were

conducted. The first examined the overall prevalence of these behaviors, over trial blocks,

whereas the second analysis was focused on the percentage of touches or retrievals which

occurred early (before the carriage reached the end of the track) across blocks. Each

dependent measure (either percentage of touches/retrievals or the percentage of early

touches/retrievals) was determined for each trial block and subjected to an ANOVA in

which the single within subject variable, Block, was considered.

Data Collection

Both the temporal cues for stimulus presentation and the collection of

electrophysiological data were done on-line. The electrocardiogram was detected from

two Sensormedics Ag/AgCI electrodes in a modified Lead II arrangement, with active

leads attached to two locations on the sternum, above and below the level of the heart.

Electrocardiogram data was amplified by a Coulbourn Instruments S75-01 hi-gain

bioamplifier. The signal then triggered a peak detector on the R wave of each heart beat.

The time between peaks was recorded on-line, measured to the nearest millisecond by a

computer using a 80286 processor. After testing, the interbeat intervals were assessed for

errors and converted into a weighted average HR for each second of the trial

All motor data were recorded from a video camera, directed at an overhead

mirror, designed to capture an aerial view of the infant and the top of the table. In

addition, a time/data stamp, superimposed on the video image, kept a running clock (to

the nearest .01 sec) such that the real time that motor responses occurred could be coded.

A piece of colorful tape was affixed along the midline of the infant's head to serve as an

aid to determine head orientation from above. Raters then placed a transparency, which

had lines drawn on it to divide the video image of the table top into five zones (see Figure

2a), over the video monitor. The video tape was paused every 0.5 sec, and a code which

corresponded to the zone that the infants' head was oriented to at that half second was

recorded in order to track changes in head position over time. Of particular interest was

whether, and when, the infants would be oriented toward the "Target Zone" (the zone

with the track carrying the toy), the "Opposite Zone" (the zone with the other track) or

the Middle Zone (where SI was located).

Two aspects of the infant's behavior (touches and retrievals of the toy) were

scored for each trial: whether or not a touch or retrieval occurred on each trial and, if

there was a response, the time at which it occurred relative to the carriages' approach. In

addition, the exact time that the experimenter presented each of the stimulus events was

scored independently from videotape so that any trials on which the timing fluctuated by

more than 2 sees could be excluded from analysis.

In general the experimenter's timing was fairly accurate but variable enough to

warrant adjustments in the motor analyses in particular. For example, the average time

that the S1 flag was raised was 6.12 (intended time was second 6.0) with a average

standard deviation of .58 sec. The toy emerged, on average, at second 12.75 (intended

time was second 12.0), with an average standard deviation of .45 sec. The carriage should

have reached the end of the track at second 15 and did so, on average, at second 14.94,

(std = .60). The trajectory of the carriage, as it rolled back, was slightly less accurate with

the carriage starting to move back at second 16.61, on average (std = .85) and being fully

withdrawn at second 19 (std = 1.05), an event which should have occurred at second 18.

Therefore, all head position data were analyzed relative to the point in time, coded

independently from videotape and rounded to the nearest half second, that S2 was first

visible on each trial. It should be noted, however, that although raters were trained,

informally, to code data in a reliable manner, no formal measure of the percentage of

agreement was calculated-a methodological limitation which was improved upon in the

second study.


After obtaining informed consent from one or both parents, HR electrodes were

placed on the infants' chest. In addition, a strip of colored tape was attached down the

midline of the infants' head. The parent was seated in front of the table, holding the infant

in his/her lap. Parents were instructed not to assist or hinder (e.g., by blocking their arms)

their children in any way and, in general, to remain passive and uninvolved during the

session. Parents were allowed, however, to praise infants when they successfully retrieved

the toy and their help was sometimes elicited in returning retrieved toys to the



After receiving these instructions, subjects were given a series of "practice" trials

in which the presentation of Sl was immediately followed by an approaching S2, which

remained at the end of track until the infant retrieved it. Thus, there was no delay interval

between the warning flag and the time at which the carriage rolled out on these trials. The

purpose of these trials was to ensure that infants had enough experience with the

apparatus to know that the toy could by detached from the carriage. Although three

successful retrievals were required before beginning the session, most infants quickly

retrieved the toy on the first presentation. No data were recorded on practice trials.

Testing Session

Two experimenters conducted the testing. The first monitored the infant via

videotape and was responsible for initiating trials from the computer when the infants were

deemed reasonably calm and facing the front of the table. Following this, the second

experimenter was cued from the computer via headphones to present S1. Six seconds

later the second cue signaled the experimenter to roll out the carriage with the toy

attached The entire duration of the carriage's approach and withdrawal was approximately

6 s (See Figure Ia) regardless of whether the infant grasped the toy from it or not.


If the infant successfully retrieved the toy, parents were instructed that they could

verbally praise the infant (the experimenter did this regardless so that all infants would

have verbal reinforcement), allow them to explore the toy for a few seconds and then

return it to the experimenter behind the curtain. If the infant did not retrieve the toy, it

simply remained attached to the carriage and was returned to behind the curtain. In either

case, once the carriage was completely behind the curtain, the warning flag was taken

down. At this point, a new toy was chosen and placed on the carriage and the

experimenter waited for cues to begin the next trial.

Study 1

appieahsDi usiabdr- n

Sl Flau Raised

Conceptual Prestimulus Period S1 Alone Interval S2 Visible Interval

S) Cue

t t
Toy Toy at
Emerges end of track

0 6 12 15 18 20
Seconds Post Trial Onset


Study 2

Conmainers presered|

Sr .r.n .nh Co..n'. Fsae Uoa, n

To-, llat dl,.r I Search & retrieval possible
_. j_ -------->

S Centering Hiding Delay lAppiroah Retrievall

Experimenter attracts Screen Is Containers Containers at
baby' attenton to center pulled down Emerge end of .rack

0 6 9 12 15 20
Seconds Post Trial Onset

Timing Diagram for Study l(a) and 2(b)







I xperimenter

Tracks carrying toy to baby
Zone 3



Schematic Diagram of Apparatus: Study 1 (a) and 2 (b)


Based on previous research, cardiac indices of anticipation were expected to

occur, given that infants were utilizing the warning flag as a cue for the presentation of the

toy. Whether or not anticipatory heart rate responses were accompanied by analogous

changes in the motor system is of particular interest. Each response system, as well as the

infant's reaching behaviors, will be considered separately below.

Heart Rate

Two anticipation intervals are of interest: the 6 seconds between the warning flag

(S1) and the point when the toy could first be seen but not retrieved (Sl Alone Interval)

and the 6 seconds following this point, during which infants were preparing to and actually

did retrieve the toy (S2-Visible interval). As in previous research, anticipatory HR during

the first interval would provide information on the infant's ability to anticipate the

presentation of the toy before it was perceptually available (memory-based anticipation).

The S2 Visible Interval, in contrast, was not designed to capture true anticipation, but it

was of interest nonetheless to discern whether any visually-guided anticipatory behaviors

occurred as S2 approached (perceptually-based anticipation). Thus, HR change scores

were calculated for each interval using the second prior to the beginning of each interval

of interest. Data were then grouped into five blocks of four trials each and analyzed in the

two intervals separately. In both, these data were examined with regard to the between

subject factor, S2 Orientation, to determine whether side biases were present, as well as

the two within subject variables, trial Blocks and Seconds.

Since there was no indication that S2 orientation side contributed to differences in

HR response, this factor was dropped and subjects from each group were pooled together

for the following analyses. Typically, a triphasic HR curve emerges during the anticipation

interval between the cue (S1) stimulus and the awaited stimulus event (S2) consisting of a

DI, a D2 and, sometimes, an intermediate acceleration (A). Although such a response

was expected during the S1 Alone Interval, no evidence of HR deceleration (either D1 or

D2) was present in the S -Alone interval at all, as reflected by the absence of significant

Seconds effects in this analysis. As depicted in Figure 3, there appears to be a slight

decrement in HR. However, not only was this response statistically insignificant, it also

failed to show the typical habituation pattern over blocks. This is unusual given that

infants usually at least orient to S1, producing a Dl response that quickly habituates over

trial blocks. In fact, there appears to be larger early decelerations on later blocks, an

effect inconsistent with orienting-related Dl responses. In any case, these sporadic effects

failed to reach significance and are probably not meaningful.

It is not until the point at which the toy begins to approach that a significant

change in HR occurs as evidenced by several significant Block by Seconds interactions in

an overall analysis of data in the S2-Visible Interval alone (see Figure 3). It is at this point

that HR seems to turn sharply downward, especially on blocks 4 and 5 when one would

expect an anticipation response to be well established. Separate analyses of individual

blocks indicated that, in fact, whereas the first trial block had no significant linear,

quadratic or cubic trends, the HR patterns for blocks 2 through 5 were primarily linear,

Fs(1,23) = 5.92, 6.06, 12.85, and 7.92, ps < .02, respectively, reflecting the larger,

sustained decelerations characteristic of the latter trial blocks in the seconds just prior to

the infant's retrieval of the toy. HR continued to decelerate during the period while most

infants were exploring and manipulating the toy--if they retrieved it--probably reflecting an

orienting response to the stimulus itself

As with the SI Alone Interval, the HR patterns in the S2 Visible interval also failed

to exhibit the typical 3 component curve. That is, it is difficult to distinguish a clear Dl

and D2, especially late in the session. Rather, a strong, uninterrupted linear deceleration

characterizes the infant's response as the toy approaches.

In summary, there was no evidence of a true anticipation response given the lack

of decelerations in the S1 Alone interval. Significant decelerations did develop during the

second interval and became increasingly stronger and more linear over blocks, suggesting

that anticipation was not fully engaged until the approaching stimulus became visible.

Possible explanations for this unexpected response pattern will be offered in the


Motor Data

Head Position

For each half-second, the percentage of trials, in each block, that the infants' heads

were oriented to the Target, Opposite or Middle zones was determined. The most

compelling evidence of anticipatory motor behavior would be obtained if the infants'

heads became increasingly likely, over Blocks and Seconds, to be oriented toward the

Target zone, indicating that they learned to expect the toy in this location in the seconds

just prior to its emergence from behind the curtain.

In order to determine whether this occurred, the percentage of head turns to each

of the three zones was examined from 6 seconds to 1 second prior to the S2 Visible point

on each trial. The last three half-seconds of the SI Alone interval (sec -1, -.5 and the S2

Visible point itself) were omitted from analysis due to the possibility that this time period

may be contaminated by one important, potential confound--noise created by the carriage

as it began to push through the curtain before it became visible. It is not until 1 sec before

the toy emerges that there is a sharp increase in head turns toward target and, in fact, the

percentage of head turns toward Target does not surpass 50% until one half-second

before the toy is visible-- a small enough amount of time that noise accompanying the

carriages approach could contaminate the head turn response. For this reason, increased

looking toward Target during the SI Alone interval during this time period could be

somewhat suspect. Thus, the first set of SI Alone analyses was restricted to seconds -6 to

-1 to minimize the possibility of confusing apparent anticipatory motor responses with

head turns which are merely orienting responses to the noise. Head turns during these last

three half-seconds were analyzed separately in a Zone X Block X Seconds ANOVA.

Given that HR did not respond until the 6 seconds after this point, an additional set

of ANOVAs was conducted which examined the head turn data during the S2 Visible

interval, resulting in three primary sets of head turn ANOVAs: 1) Zone X Block X

Seconds secss -6 to -1 in the SI Alone Interval), 2) Zone X Block X Seconds secss -1, -.5

and 0, in the S1 Alone Interval), and 3) Zone X Blocks X Seconds (in the S2 Visible


As shown in Figure 4, infants are primarily fixated at midline after the presentation

of S1. This response wanes over the next several seconds, becoming offset by small

increases in head turns to the Target and Opposite zones. The different patterns of

responding in the three zones were statistically significant as well, as indicated by a three-

way Zone X Block X Seconds interaction, F(1,15) = 110.66, p <.0001, among numerous

other significant effects for Zone and Seconds.

An additional ANOVA, which omitted the Middle zone, was conducted so that

head turns in the Target and Opposite zones could be compared directly during the first 6

seconds of the S1 Alone Interval. As in the overall analysis of all three zones, the three-

way interaction was upheld in a Zone X cubic Block X quadratic Seconds effect, F (1,15)

= 9,05, p < .01. Thus, even within these two zones, the shifts in head position appear to be

somewhat different.

Separate analysis of each block indicated that this complex interaction seems to

stem mainly from the presence of a Zone X Seconds interaction on blocks 2 and 3. In

addition, there were subtle differences in the type of Seconds effects across in the

individual blocks. For example, whereas linear Seconds trends were present on all five

trial blocks, F(1,15)= 13.42, 18.01, 15.43, 9.09, and 13.55, respectively, 's < .05, the

first three blocks had significant higher order seconds trends as well, suggesting that the

response pattern may have become less complex over blocks (see Figures 5, 6 and 7 for

individual block data in each zone). Furthermore, follow-up analyses of each zone

separately suggested that, while linear Seconds trends were present for both Target,

F(1,15)= 21.38, < .005, and Opposite, F(1,15)= 9.65, p < .01, zones, the higher order

Seconds trends were more characteristic of the Target zone in particular. Most notable

was a significant quartic Seconds trend, F(1,15) = 6.91, p < .05, in the Target zone.

In summary, the largest effect in the first 6 seconds of the SI Alone Interval is the

apparent trade-offbetween looking toward the Middle zone versus the other two

alternatives as the time of the toy's appearance approaches. The smaller interactions

between Target and Opposite zones simply reflect the greater complexity of responding in

the Target zone and are probably not very meaningful

In the half-seconds just prior to the toy's appearance, in contrast, a very dramatic

difference between these two zones emerged, reflected by a significant linear Block X

linear Seconds interaction, F(1,15) = 20.75, p < .001 during this time period. Although

there is not much change in the Middle zone (looking simply continues to decrease) there

is a notable divergence in the percentages of head turns toward the Target versus the

Opposite zone as the likelihood of looking toward Target increases abruptly while the

alternative response wanes (see Figure 4). To examine this pattern more closely, another

ANOVA was conducted which was restricted to the Target and Opposite zones only. In

addition to significant Seconds, F(1,15) = 12.61, < .005, and Zone, F(l,15) = 25,47, p <

.0001, effects, the interaction between the two variables was significant as well, F(1,15)=

40.37, p < .0001, suggesting that, in fact, head turns shift toward the Target, and away

from the Opposite, zone just prior to the toy's appearance. However, the theoretical

significance of this response must be interpreted with caution until the possibility of noise

confounds can be ruled out.

Once the toy begins to approach (the S2 Visible Interval), the infants' heads are

clearly directed to the Target zone, almost exclusively, as they prepare to retrieve the toy.

Analysis of head turns in all three zones during this interval supports what is shown in

Figure 4, with a significant Zone X Quadratic Seconds, F(1,13) = 41.72, p <.0001,

interaction. An ANOVA restricted to Target and Opposite zones only continued to

support this same interaction, F(1,13) = 41.92,

p < .0001, which resulted from the presence of quadratic Seconds trend in the Target,

F(1,13) = 46.92, p < .0001, but not Opposite, zone (see Figure 4). It should be noted that

two subjects were dropped from the S2 Visible Interval ANOVAs due to missing data

points on second 6.

Reaching Behavior

Two categories of data were of interest here: 1) the infant's success in either

touching or retrieving the toy from the carriage and 2) the time at which these behaviors

occurred. Regarding the latter, the extent to which the infant made contact with the toy

early (before the carriage reached the end of the table) was of particular importance since

it could be interpreted as a type of motor anticipation.

In general, infants were quite effective in making contact with the toy (see Figure

8). It appears that infants were more likely to simply touch the toy without actually

detaching it from the carriage as indicated by the greater percentage of good trials with

"Touches" (M = 82%) than "Retrievals" (M = 67%). The difference between Touches


and Retrievals was statistically significant as well, in a Response X Block ANOVA which

resulted in a main effect of Response, F(1,15) = 28.22, < .0001. However, it should be

noted that the percent of touches is probably inflated since retrievals could include the

initial "touch" before the infant ultimately pulled the toy from the carriage. The number of

times an infant merely touched without making further contact was not coded in this

study--an aspect of the methodological which was improved upon in study 2. In any case,

there were no significant Block effects, indicating that neither response changed

substantially over trials.

The second set of reaching analyses focused on the timing of the infants' reaching.

On average, infants touched and then retrieved the toy from the carriage at second 15 (SD

S.78 secs) and second 16.15 (SD = 1.02, respectively. Of greater interest, however, was

the percent of trials in each block with early Touches and Retrievals was determined and

subjected to a Response X Block ANOVA. As shown in Figure 9, the frequency of early

contacts of either type (Mean % of early Touches = .33, early Retrievals = .07) dropped

dramatically from the overall number of responses, with early Retrievals representing less

than 10 percent of all good trials. The relative dearth of early retrievals versus Touches

was verified statistically, via a main effect of Response, F(1,14) = 17.45, p < .001. In

addition, a significant Block X Response interaction, F(1,14) = 6.37, p < .05, depicts the

significant change across Blocks in the percentage of Touches, F(1,14) = 6.37, p < .05,

but not Retrievals. However, this Block effect is most likely reflecting the

disproportionate number of early Touches in Block 2 rather than a linear increase in early

responding across the testing session. In summary, both the head turn and reaching data

seem to indicate that infants are responsive once the toy is visible and this behavior

appears to be present on the very first trial block--HR decelerates, infants orient their head

to the approaching carriage and the infant execute reaches which are culminate in efficient

toy retrieval shortly after its approach.

9-mo-old HR (N = 25)
During the S1 Alone Interval
2 --

-2 .... .... BLK 2

-8 BLK 4
5.5 6.5 7.5 8.5 9.5 10.5 11.5 BLK
Seconds During S1 Alone Interval


9-mo-old HR (N = 25)
During the S2 Visible Interval

-2 ,- BLK 2
-4 BLK 3

56 -6 -0- _
-8 I BLK 4
S11.5 12.5 13.5 14.5 15.5 16.5 17.5 BLK5
Seconds During S2 Visible Interval

9-mo-old Mean HR change scores, across infants, during the
S 1 Alone (a) and S2 Visible (b) Intevals, on trial blocks 1 to 5

9-mo-old Head Position: Mean percentage of trials,
across all infants and trial blocks, head is in each zone

9-mo-old HEAD POSITION (N = 16)
w 1
ig S1 Alone S2 Visible
S0.4 Opposite
z 0.2 Midline

Ir 0
S -6 -4 -2 0 2 4 6

9-mo-old Head Position: Mean percentage of trials,
across all infants, in each trial block, that head is in Target Zone

9-mo-old HEAD POSITION (N = 16)
% Head Turns to Target Zone

I 0.8

m S1 Alone S2 Visible
: 0.4 BLK 3

: 0.2 _, BLK 4

-6 -4 -2 0 2 4 6 BLK5

9-mo-old Head Position: Mean percentage of trials,
across all infants, in each trial block, that head is in Opposite Zone

9-mo-old HEAD POSITION (N = 16)
% Head Turns to Opposite Zone

z 1 ._
o 0.8
__ BLK2
SS1 Alone S2 Visible
0.4 BLK 3

Q ---
S0.2 BLK 4

0 BLK 5
-6 -4 -2 0 2 4 6

9-mo-old HEAD POSITION (N = 16)
% Head Turns to Middle Zone

S0.8 BLK 1
o 0.8

S0.6 BLK2

0.4 BLK 3
< S1 Alone S2 Visible
0.2 BLK 4

0 ++ !!+++ BLK 5
-6 -4 -2 0 2 4 6

9-mo-old Head Position: Mean percentage of trials,
across all infants, in each trial block, that head is in Middle Zone


9-mo-old Motor Responses: Mean percentage oftrials,
across all infants, with Touches and Retrievals on trials blocks 1-5

9-mo-old MOTOR RESPONSES N = 16

) 1

0 0.8

0 Touch
* 0.4
3 Retrieval

T1 2 3 4 5
Trial Block

9-mo-old MOTOR RESPONSES N = 15

S 1
| 0.8 .--------___

| Touch
w Retrieval

T 0
1 2 3 4 5
Trial Block

9-mo-old Motor responses: Mean number of early contacts with toy,
across all infants, for each trial block


Heart Rate

The progressive HR decelerations in the 9-mo-old subjects seem to indicate

anticipation during the S2 Visible interval as retrieving the toy became imminent. Not

only did the HR decelerate across seconds but these decelerations became increasingly

large and more linear across blocks, suggesting that the infants were learning something

about the sequence of events over trial blocks.

The primary question revolves around what they were learning. Counter to

expectation, the infants did not demonstrate any cardiac evidence of anticipation during

the SI Alone interval, the foreperiod originally intended to elicit anticipation of the toy's

appearance. Rather, no anticipatory response was initiated at all until the stimulus event

intended to be S2 (the toy's appearance). Infants this age should have been capable of

using mental representations to anticipate an unseen S2 (Berg and Donohue, 1991) leaving

the question of why they failed to do so here unanswered.

One explanation is that the paradigm was in some way unable to elicit the

response. Classic anticipation paradigms always include two critical components: 1) a

clear cue or S1, which is reliably associated with 2) an interesting event. The present

paradigm did include these stimuli, but, in an attempt to elicit anticipatory motor behavior,

the sequence of events was necessarily more complex than the presentation of these two

stimuli alone. For example the toys moved from one end of the table to the

other to ensure they would be out of reach until the end of the S1 Alone interval. A basic

assumption which guided the design of this procedure was that the most salient stimuli for

the infant would be the flag (intended to be S1) and the toy's initial appearance (intended

to be S2). Infants were expected to begin reaching toward the curtain occluding the toy,

in anticipation of it's impending appearance. Thus, the subsequent approach of the toy

was designed to only to serve as an extension of it's initial appearance, allowing the

infants to finally grasp the toy they were awaiting.

However, if the most interesting event for the infant was not the toy's initial

appearance, as intended, but the opportunity to actually grasp and explore the toy, then

the latter effectively becomes S2; that is, the most significant event or goal. Consequently,

the cue which best predicts this event is the toy's first appearance. Not only is this event

closer in time to toy retrieval than the flag, but it is a more ecologically valid cue as well

since infants see the very object that they anticipate grasping. Thus, the infant may have

perceived the toy's initial appearance as "SI" the most salient predictor of the time when

they would be able to actually handle the toy (the infant's perception of the real "S2").

If this is what occurred it would help to explain some facets of the HR data. First,

no significant HR decelerations occur at all until the toy begins to approach. Although it

is difficult to discern a distinct Dl and D2, there are strong decelerations in the S2 Visible

Interval nonetheless and, more importantly, they become progressively more linear in

shape and greater in magnitude over trial blocks. This pattern is significant, theoretically,

because it is more typical of a D2--a HR decelerations which results from anticipation--

than a Dl (an orienting response), lending support to the notion that, at least during the

first half of the interval, infants may have been anticipating grasping the toy (Mean

retrieval time was at sec 15, exactly in the middle of the S2 Visible Interval). Although

there were not two distinct decelerations during the interval, this may simply be the result

of using a shorter ISI, which can interfere with the typical triphasic response pattern (Berg

and Donohue, 1991). Furthermore, the infants could have been learning something about

the trajectory of the approaching toy as well. Thus, the ability to anticipate the time that

the toy will be in reach may also contribute to the initial portion of the progressive

decelerations, reflecting the infants' greater temporal precision over blocks.

Motor Behavior

As with HR, infants were expected to become increasingly likely, over blocks, to

look toward the Target zone during the S Alone Interval, indicating that they were

anticipating the toy's appearance. However, although there is some increase in looking

toward Target during the early portion of this interval, the probability is still below chance

level with infants looking toward midline more than anywhere else.

The likelihood of looking toward Target does increase sharply in the second just

prior to the toy's appearance but, as previously mentioned, it is uncertain whether these

data were contaminated by unintended noise as the carriage began to move through the

curtain. In any case, the infants do not make clear head turns to Target until after the toy

becomes visible.

Where the head turn data differ from HR is in the extent to which the one can

interpret the response as anticipation rather than orienting. Whereas decelerations during

this second interval increased in magnitude over blocks, suggesting that infants learned

something about the object's trajectory, there is a notable absence of block effects in the

head turn data. Rather, the tendency to turn toward Target remains stable across the

testing session, suggesting that this response may be more characteristic of an orienting

response to the approaching toy--a response which may serve a different sort of adaptive,

albeit non-anticipatory--function.

One explanation for the dearth of motor responding during the first interval is that

there simply was not enough incentive for the infant to look toward the Target zone until

the toy appeared, cueing them that they would soon be able to physically handle it.

Looking toward the side of the curtain which obscures the toy during the S1 Alone

Interval provides little or no advantage over waiting until the carriage approaches in terms

of their ability to grasp it. Preparatory head turns toward Target, even in the last second

of the carriage's approach, may still allow the infant enough time initiate an adaptive

process that results in reaching to the correct location before the carriage is withdrawn.

Thus, although the head turns to Target are not truly anticipatory (in that they do not rely

on mental representations of an unseen, future event), they may still be adaptive

nonetheless in preparing the infant to retrieve that object in the most efficient manner


If looking toward the correct location does, at least, facilitate reaching, then one

would expect a high number of early touches and retrievals during the approach phase--an

expectation which has modest support. Although the percentage of early contacts was

relatively low (infants were only able to touch the carriage before it reached the end of the

track on approximately 35% of the trials), they were fairly successful in making contact

with the toy before it was fully withdrawn. For example, infants at least touched the

carriage on 80% of the trials and were able to fully detach the toy from the carriage at

least 60% of the time before it was withdrawn suggesting that these were highly motivated

responses. In any case, the motor data continue to lend strong support to the notion that

it was the retrieval of the toy which was of utmost interest, an event which was apparently

cued by the toy's appearance at the onset of the S2 Visible interval.

The theoretical limitation of this scenario, of course, is that it does not require the

infants to make preparatory motor adjustments in the absence of the anticipated object. In

fact, the motor behaviors which occurred during the toy's approach are not all that

different from the adjustments in hand orientation made by VonHofsten and Fazel-Zandy's

(1984) infants as they prepared to grasp horizontal versus vertical dowels. Although the

responses elicited in study 1 are adaptive in that they enabled the infants to achieve their

goal--grasping the toy before it was withdrawn--they do not rely upon the mental

representation of a future event. In short, the motor data provide strong evidence of

perceptually-based, but not memory-based, anticipation.

Taken together the cardiac and motor data both suggest that it was the toy's

appearance which served as the most salient cue, prompting strong HR decelerations and

head turns as the infants prepared to make contact with the approach toy. There is some

evidence that, although "S2" is perceptually available, the HR and motor behaviors which

occurred during this time may still serve an adaptive purpose: the progressively large HR

decelerations may reflect anticipation of the carriage's trajectory, enabling infants to

better coordinate their behaviors with the temporal parameters of the events over trial


blocks while the head turns, in contrast, probably do not reflect true anticipation since the

object they are responding to is visible and the response does not increase over blocks

Nonetheless, the infants appear to be exhibiting a strong orienting response to the toy they

were preparing to grasp an adaptive behavior which enables them to retrieve the toy

with great efficiency.


Several modifications in procedure were made in the second study. Most of these

were designated a priori, such as the utilization of differential cues and distracters in study

2. However, some of the changes were based strictly on the results from study 1, in an

attempt to revise methods which were ineffectual. For example, data from study 1 suggest

that the approaching carriage and toy was especially capable of eliciting decelerations and

head turns. However, these findings are limited in that the toy was visible during this

approach phase. Since the infant could see the toy as they were preparing to make contact

with it, we do not really have evidence of anticipation based on the mental representation

of S2. Rather, any anticipatory behaviors which occurred in study 1 may be little more

than covert and overt manifestations of an orienting response to the approaching toy.

Likewise, preparatory adjustments in reaching were, essentially, visually guided

adjustments more typical of VonHofsten and Fazel-Zandy's (1984) subjects, who modified

their grasps for different orientations of a dowel. Thus, the responses found in study I do

involve anticipation but they are changes in behavior which are probably based more on

visual feedback than on a mental representation of an expected, but as yet unseen object.

In short, there seems to be evidence of "perceptually-based" but not "memory-based"

anticipation in study 1.

The procedure in Study 2 was designed to offset this shortcoming by occluding the

approaching toy in one of two opaque containers. Thus, this second paradigm

capitalizes on the apparent salience of the approaching carriage while still allowing

examination of anticipatory responses to an unseen S2--the toy hidden inside one of the


One of the most significant methodological differences between the two studies is

that the procedure in study 2 allows examination of differential motor anticipation since

the toy they are preparing to retrieve can be hidden in either of the two simultaneously

approaching containers. Thus, the infant is presented with a forced choice in terms of

which container he prepares to search. In this manner, the extent to which infants can

utilize a differential cue to prepare to search in a particular location can be examined

during a stimulus event which seems optimal for eliciting anticipation--the presentation of

approaching objects.

In order to successfully implement this change, the resulting procedure became,

necessarily, more complex. The most notable aspect of this complexity is the greater

number of stimulus events which were required in order to ensure that infants encoded and

acted upon the differential cue. As a result, the paradigm in study 2 departs from the

traditional fixed-foreperiod S1-S2 paradigm but in so doing facilitates the examination of

differential motor behaviors. Finally, study 2 evaluates the relationship between

anticipatory behaviors on the one hand, and the infant's spatial accuracy (the infant's

ability to accurately remember which container the toy was hidden in and search in that

one) on the other, providing an opportunity to examine whether anticipation is functional

in promoting performance on this task.


All recruitment procedures were identical to those employed in study 1 with one

exception. Initial pilot testing of infants aged 12 to 15 months suggested that only the

older subjects (14 and 15-mo-olds) were capable of searching the containers well. Thus, it

was this age group, rather than 9-month-olds, which was recruited to participate in study

2. The mean age of the subjects included in the final sample was 14 months and 2.5

weeks. Seventeen subjects contributed usable HR and head position data; of these two

subjects showed very little search behavior, resulting in an N of 15 for analyses of search



Several significant stimulus events occurred within each 26 sec trial (see Figure

lb). As in Study 1, these events will be described in two sets, those used by the

experimenter as temporal cues and those presented to the infants. The first of the

experimenter's cues consisted of a 1 sec tone, presented at second 5, which signaled the

experimenter to begin hiding the toy. The second cue, a 1 sec tone presented at second

11, prompted the experimenter to begin pulling down a screen in front of the two

containers. The third stimulus was a 3 sec, broad-band white noise stimulus secss 12 to

15), intended to pace the experimenter as she rolled out the containers such that they

reached the end of the track at the noise's offset.

The second set of stimulus events were directed to the infant. The first of these

(see Figure Ib) involved placing the toy the infant would search, (selected from a group of

colorful, three dimensional, figurines) on the experimenter's hand in the middle of the table

to center the infant's attention to midline at the beginning of the trial. Six seconds later,

this same toy was hidden in one of the two plastic, opaque (5" X 2.5" X 2") containers.

These containers had a hinged lid with a large, red knob at the front which allowing infants

to easily grasp and open the container.

Nine seconds into the trial, an attractive, centrally located, distracter stimulus was

introduced which served two purposes: 1) to redirect the infant's gaze from the hiding

location to the middle of the table and 2) to temporarily occlude both containers from

sight. The distracter stimulus consisted of a (7" X 8") multi-colored picture of a clown's

face drawn on the center of a vinyl screen (38" across X 16" tall). At the beginning of

each trial this screen was rolled up and suspended 16" above the far end of the table. Nine

seconds into the trial the screen was pulled down just in front of the containers, occluding

them from the infant's sight while simultaneously revealing the clown face. At second 12,

this screen was momentarily lifted just enough to allow the containers to pass through and

roll out along a track toward the infant's end of the table (these tracks were identical to

those described in study 1). Hiding the toy in one of the two containers provides the

differential, spatial cue for where the toy can be found (the toy is always in the same

container it was hidden in). However, this or any of the subsequent stimulus events could

serve as a trigger (the closest parallel to "Sl" in traditional paradigms) for the anticipation

response as infants await the upcoming event of interest (the conceptual equivalent of

"S2") --the point in time when the infant retrieves the toy from the inside of one of the


As shown in the mid-section of Figure 2, this series of events generated five

conceptually distinct intervals: 1) the Centering Interval (from sec 0 to 6 as the

experimenter attracts the infant to midline at the beginning of the trial), 2) the Hiding

Interval (sec 6 to 9, as the toy is placed in one of the containers and the subject encodes

the differential cue), 3) the Delay Interval (sec 9 to 12, while the containers remain

behind the screen), 4) the Approach Interval (sec 12 to 15, as the containers are rolled

along the track to the infant) and 5) the Retrieval Interval (beginning when the infant

initiates the search and extended until the infant begins to manipulates the toy or until the

end of the trial).


Infants were presented with 20 trials in which they were to find the hidden toy.

Unlike in study 1, however, the toy could be hidden on either the left or right side. Thus,

for each subject in this study, a single presentation side no longer exists as a between

subjects factor. Rather, side of presentation varied from trial to trial according to one of

two randomized presentation order sequences, counterbalanced between subjects, in order

to rule out potential trial order effects. Each presentation order was randomized with the

restriction that there were no more than two consecutive trials with the toy hidden on the

same side, and that there were an equal number of right and left trials. All scores were

arranged into 5 blocks of 4 trials each for data analysis.

The same three dependent measures from study 1 are still of interest here: 1)

orthogonal trends in HR across blocks and seconds, 2) changes in the infants' head

position across zones, blocks and seconds, and 3) the time at which the infant attempted

to retrieve the toys, across blocks. The latter was of interest since early searches

represent a form of overt anticipation. Study 2 examined two additional measures study 1

did not address: 4) the infant's search accuracy -- that is, whether the first container they

searched was the one with the toy inside it, and whether the percentage of correct

response increased across blocks, and 5) whether cardiac and/or motor indices of

anticipation were related to spatial accuracy in the delay task.

Data Collection

Heart Rate

All computer controlled stimulus timing and HR data collection procedures were

identical to those described in study 1. HR was measured starting 6 seconds prior to when

the toy was hidden and continued until the end of the trial (See figure Ib).

Motor Behavior

Head Position

As in study 1, the zone to which the infants' head was oriented was coded by

raters every half-second, throughout the duration of the trial. Second raters coded 30% of

these same subjects and an inter-rater reliability of 92% was obtained. All other coding

procedures involving head position were identical to those described for study I with one

exception. In study 1, infant's head positions were determined from videotape by placing a

transparency with the five zones drawn on it over the video monitor. Although useful, this

method was somewhat cumbersome and, due to occasional inconsistencies in video

camera position, it was difficult to align the template perfectly with the image on the

monitor each time. To improve upon these procedures in study 2, the zoning lines were

drawn on the table itself, eliminating the need to align an external map with the video

image. In addition, whereas the Target Zone was always located on the same side of the

table for a particular infant in study 1, in the present study, the spatial location of the

Target Zone changed from trial to trial with the hiding location of the toy. Thus, for every

trial, the Target Zone was simply the area surrounding the track with the container which

concealed the toy on that particular trial.

Search Behavior

The percentages of"touches" (defined in this study as contact with the container

which did not result in the infant opening it) and "retrievals" (defined as contact which

resulting in the container being opened) were determined and examined separately. The

defining criteria of a "touch" was altered from the first study to reflect only those contacts

which did not lead to a search, in an effort to allow a more direct comparison between the

two types of contact. Using this criterion, however, it became clear that, in fact, infants

rarely touched the containers without further exploration (the Mean % of trials with

touches only was .133) This behavior was so underrepresented in the data, in fact, that

this response measure was dropped from Study 2. Thus, all reaching behavior in study 2

refers to searches only.

The coding of search behaviors was similar to that in study 1 in that it differed

from head position in two important ways. First, whereas head position was coded

continuously at each half second, search behaviors were considered to be single, discrete

responses. In order for the behavior to be considered a "search", two criteria were

observed: 1) the infant's hand had to make physical contact with one of the containers

and 2) following this contact, the lid of this same container had to be at least partially

raised. If no contact was made at all on that trial, a code for "no response" was entered

instead. If a response did occur, the time at which it was initiated was noted for each trial.

The time for the onset of a search was coded as the time (to the nearest .01 sec) at which

the lid was first visibly displaced from the rest of the container. Finally, the raters coded

the infant's accuracy--that is, whether responses involved the correct or incorrect

container. Only the first search was considered.

The difference in the search onset times coded by two different raters were

calculated (the Mean of the absolute differences in time was .56 seconds) and deemed

reasonably consistent. With the accuracy data, reliability measures between two raters

revolved around their consistency in determining which container the infant first searched

(raters were found to be in agreement 95% of the time).

Stimulus Timing

As in study 1, because the apparatus was manually operated, an additional set of

ratings was done on each subject to evaluate the experimenter's timing accuracy and

consistency (both within and between subjects). To achieve this, the time (to the nearest

.01 sec) at which the following critical stimulus events occurred was coded from

videotape for each trial: 1) the exact point in time, during the Hiding Interval, that the toy

was placed in the container, (intended time was 6"; the experimenter's Mean time =

7.33", std = .38"), 2) the time the screen was pulled down (intended time was 9", the

experimenter's Mean time = 9.33", std = 26), 3) the time the containers emerged from

behind the screen (intended time was 12", experimenter's Mean time = 12.27", std =

.15"), and 4) the time the containers reached the end of the track (intended time was 15";

experimenter's Mean time = 15.13", std = .30"). With the exception of the hiding time,

the experimenter's timing was within .33 sec of the intended time and was reasonably

consistent across trials and subjects. Thus, reference points for HR change scores and

head position were not adjusted to compensate for fluctuations in as they were in study 1.


All initial subject recruitment, consent and electrode placement procedures were

identical to those described in study 1.


Four pretrials were carried out before the regular testing session, in 9 of the 15

participants, in an attempt to compare infants' performance on the two-alternative delay

task to performance on more traditional object permanence delay tasks such as those used

by Diamond (1991). Infants were presented with two pretrials in which one of the

attractive toys was hidden in one of two (either to the right or left of midline) small,

opaque tubs. Both tubs were then covered with a felt cloth. On the first two pretrials (no

delay) the infant was allowed to search immediately after the tubs were covered. The toy

was then hidden in the other tub in the same manner. In the delay condition (pretrials 3

and 4), after the experimenter hid the toy, she made eye contact with the infant and

counted to six before allowing the infant to search, as in Diamond's (1991) procedure (see

the Appendix for further detail).

Infants were then given at least two additional "practice" trials intended to serve

two purposes. First, it was necessary to acquaint the infant with the basic contingencies of

the procedure. In particular, the experimenter attempted to demonstrate that the hiding

location was, indeed, a valid cue for where the toy could be later found. Second, it was

important, even in these older infants, to ensure that they knew how to raise the lid to

retrieve the toy. The infants first watched the experimenter hide the toy on one side of

the table. Instead of using the distracter screen, the containers were simply rolled out,

without a delay, and the infant was then allowed to practice retrieving the toy. If infants

searched on the wrong side, the experimenter captured their attention and redirected them

to the correct container. They then were allowed, by trial and error at first, to explore the

container and attempt to open it. If the infant could not learn how to lift the lid the

experimenter demonstrated this for them, closed the lid, and prompted the infant to try

again. The infant was allowed to continue practicing with the lid until they could

successfully open it on their own. After successfully retrieving the toy from this container,

the same procedure was repeated with the other container.

Testing Session

The testing was always conducted by two experimenters. The first monitored the

infant from a video monitor in the adjacent room and was responsible for rating the

infant's state and for initiating trials when the infants were deemed reasonably calm,

attentive and facing the front of the table. The second experimenter was seated across the

table, facing the infant from behind the rolled up screen.

At the beginning of a trial, both containers were open and resting at the

experimenter's end of the table, out of reach of the infant. Once cued to hide the toy the

experimenter captured the infant's attention by making eye contact and placed the toy

inside one of the containers. After this 3 sec Hiding Interval, the experimenter was

signaled again to pull the screen down in front of the containers. At the end of the Delay

Interval (sec 12), the screen was temporarily lifted and both containers were rolled out

toward the infant, reaching them by second 15.

During approach, the infant was carefully monitored by the experimenter via a

second video monitor underneath the table. The infant was allowed one search at any

point during or after the containers approached. If the first container the infant searched

was the one with the toy inside both containers were left within reach until the end of the

trial to allow the infant enough time to retrieve and explore the toy. If, however, the

infant's first response was incorrect, both containers were immediately withdrawn (even if

they were in mid-approach) and the trial was terminated to prevent the infant from

searching in the other container. If the infant failed to open either container by second

20, the trial was terminated and both containers were withdrawn. Although failure to

retrieve a toy on an incorrect trial was potentially frustrating for infants, these

contingencies were deemed necessary in order to motivate accurate reaches.

After the containers were withdrawn and any toys the infant retrieved were

collected, the experimenter raised the screen to its resting position, positioned the

containers at the far end of the table, selected a new toy and waited for the cues to begin

the next trial.


Two levels of anticipation of retrieval are of interest-- generalized and differential.

Evidence of the former would be obtained if there were HR decelerations during either the

Delay or Approach Intervals, indicating that the infants had learned the temporal

contingencies. In other words, anticipatory HR during either interval would suggest that

infants learned when "S2" (retrieving the toy) could occur. However, HR decelerations

alone are not sufficient to infer that infants utilized the cues differentially to anticipate

where S2 might be. Therefore, evidence of latter, more sophisticated, type of anticipation

is dependent on the outcome of the motor behavior analyses with the clearest evidence

coming from a greater prevalence of anticipatory head turns toward the Target Zone

during either the Delay or Approach Intervals. These results, coupled with clear cardiac

decelerations during either interval, would provide strong evidence that infants can utilize

differential cues to anticipate both when and where an object of interest will be presented.

Finally, evidence for the functionality of anticipation would be obtained if there

was a significant relationship between the degree of anticipation and spatial accuracy,

suggesting that infants may be using anticipatory behaviors as a type of strategy for

solving the spatial problem.

Heart Rate

As shown in Figure 10, there is not much notable change in HR prior to the onset

of the Delay Interval. Examination of the average HR per second scores (the time, in

milliseconds, from the "R" wave in one cardiac cycle to the next) suggested that there was

very little fluctuation across infants or across trial blocks in these scores. Although HR

did became slightly elevated from block one to five (Mean bpm = 125 on block 1 and 128

bpm on block 5), this difference actually only represents a 2% change in heart rate and is

probably not meaningful. Thus, it is reasonable to assume that HR at the trial onset was

fairly stable throughout the session and between subjects allowing more straightforward

interpretation of changes in HR which occurred after the Centering Interval since these

changes are not likely the result of mere shifts in HR baseline.

The first such HR change occurs at or just before the screen is pulled down, in the

form of a marked deceleration, especially on early trial blocks. The response on these

blocks appears to reach a nadir at the beginning of the approach interval remain low

throughout the container's approach As the infants actually retrieve the toy secss 14.5 -

19.5), however, the HR takes another sharp turn downward, producing what appears to

be a second wave of even larger decelerations which persist across all five trial blocks.

HR change scores, using second 8.5 (just prior to the Delay Interval) as a

reference point, were generated for all seconds across the trial. Given that there were

several stimulus events which could have elicited changes in HR, the data were subjected

to analysis in four separate sets of ANOVAs: one for each of the four conceptual intervals

(Centering, Hiding, Delay and Approach) of interest. In each set, changes in HR were

examined with regard to the within subject variables, Blocks and the Seconds of that

respective interval.

Although there does not appear to be much HR change during the Centering

Interval, the ANOVA restricted to this time period produced a significant quadratic Block

X quartic Seconds effect, F(1,16)= 10.01, p < .01, nonetheless suggesting that there may

be more significant responding on some blocks than other. Analysis of each block

separately confirmed this: while there were no significant seconds effects on blocks 1

through 4, the same quartic seconds effect emerged on block 5,F(1,16) = 13.30, p <

.005, reflecting the greater fluctuations in HR on this particular trial block--an effect which

has little theoretical meaning or relevance.

In contrast, there was very little variability from block to block during the Hiding

Interval, as indicated by the absence of Block effects in this ANOVA. There was,

however, a significant quadratic Seconds effect, F(1,16) = 8.47, p < .05, which most likely

reflects the small increase, and then decrease, in HR on all five trial blocks in this interval.

The HR becomes more responsive again in the Delay Interval as the distracter

screen is placed in front of the containers, resulting in a significant linear Block X linear

Seconds interaction, F(1,16) = 7.64, p < .05. As shown in Figure 10, this interaction

stems from the presence of a linear Seconds effects on block 1 and 2, Fs(1,16) = 8.12,

7.25, p's < .05, respectively, which disappears on subsequent blocks as the decelerations

wane in strength.

At the onset of the Approach Interval, HR remains low and stable, with the

exception of block 5, resulting in a lack of Seconds effects in this interval. The magnitude

of deceleration going into the Approach Interval, however, is smaller with each block,

however, resulting in progressively less deceleration on each block throughout this

interval, which results in a main effect of linear Blocks, F(1,16) = 10.73, p < .005.

Anticipatory HR and Accuracy

Finally, HR was examined in relation to the infants' searching behaviors in an

attempt to assess whether the decelerations prior to toy retrieval were statistically related

to accuracy on the spatial memory component of the task by conducting two sets of

analyses. The first involved an ANOVA in which, HR change scores, across the Delay

and Approach Intervals (seconds 8.5 to 14.5), were examined on trials in which the

infant's first searches were correct versus incorrect. In addition, the data were divided

into two blocks of 10 trials each to explore whether the relationship between HR and

accuracy changed from the first to the last half of the session. There were insufficient

trials of each type--correct and incorrect--, when the trials were divided into five blocks as

usual. In addition, the relationship between HR, block and accuracy was examined in the

Delay and Approach intervals separately yielding an Accuracy X Block X Seconds

ANOVA for each interval. Four subjects were excluded from this analysis because they

did not have both usable HR and search behavior data.

The results of the ANOVA conducted on data in the Delay Interval indicated that

there was a significant main effect of Block (see Figure 11, sections a and b), F(1,12)=

9.51, p < .01, as well as a Block X linear Seconds interaction, F(1,12) = 6.97, p < .05.

Additional analysis of each block separately suggest that the interaction results from the

presence of significant linear, F(1,12) = 9.05, p < .05, and quadratic, F(1,12) = 5.80, p <

.05, Seconds effects in the first, but not second, trial block. However, there were no

effects of accuracy at all during this interval, suggesting there was little or no relationship

between HR and accuracy early in the Delay Interval.

In contrast, accuracy does appear to influence HR as the containers approach, as

indicated by a three-way, Accuracy X linear Block X cubic Seconds interaction, F(1,12)=

5.33, p < .05, in the Approach Interval ANOVA. Separate analysis of each block

individually indicated, as expected, the presence of a significant Accuracy X cubic

Seconds, F(1,12) = 9.96, p < .01, effect in the first, but not second, trial block (see Figure

11). Thus HR appears to be related to the infant's accuracy, but only during the

Approach Interval, and only in the first half of the session.

The second set of analyses involved a Linear Regression in which HR at second

14.5 (chosen because HR was generally at its peak deceleration level at this point yet

before the earliest time that the infants actually retrieved the toy) as a predicted variable

for the infants' overall accuracy to determine whether the cardiac decelerations would

predict performance as well. However, although HR and accuracy were related in the

Approach Interval in the ANOVA (assessing all seconds of the interval) the regression

analysis (using only sec 14.5) did not reach significance, suggesting that the magnitude of

HR deceleration, at second 14.5, was not a strong predicted by itselt of the overall

performance on the memory task. Possible interpretations of this discrepancy will be

addressed in the discussion.

One final issue of interest was whether there would be any individual differences,

across infants, in the overall degree of HR deceleration (across all trials) during the Delay

and Approach Intervals and, if so, whether the magnitude of this responding is related to

their overall accuracy. To address this issue a given infant was sorted into one of two

groups--a "good" versus "poor" decelerater--according to the percentage of their trials on

which decelerations occurred. Whether or not a deceleration occurred was decided

separately, on each trial, for both the Delay and Approach interval using a modified

version of the criteria used by John Richard's (1987) to identify sustained decelerations.

For each interval on a given trial, a deceleration was said to occur if each of the seconds of

recorded HR in that interval (3 consectuative seconds in all) was below a predetermined

baseline average (calculated as the mean HR change score across all seconds in the

Centering Interval).

Once the presence or absence of decelerations in each interval was determined for

every trial, two overall percentage scores were calculated for a given infant: the

percentage of trials with decelerations in the Delay and the percentage of trials with

decelerations in the Approach Interval. For each interval, scores across infants were

arranged such that a median split of percentiles could be established. All infants with

percentages greater than or equal to the median score were deemed "good" decelerators

whereas those scoring below median were considered to be "poor" decelerators.

The mean search accuracy was then determined within each group of infants.

Although the search accuracy of the "good" decelerators (as determined by HR in the

Delay Interval) was slightly superior (Mean percent of correct trials = .62) to that of

thier "poor" decelerator counterparts to the "poor" decelerators (Mean = .55), the

difference was not statistically significant. Similarly, infants classified as "good"

decelerators on the basis of their HR in the Approach Interval were accurate on 59% of

the trials compared to 57% in the "poor" decelerator group--a modest difference in

performance which failed to reach statistical significance. Thus, although trial by trial

differences in cardiac decelerations appear to be related to accuracy within a particular

subject, the between subject differences in the overall magnitude of HR deceleration did

not appear to impact on performance in a significant manner.

Motor Behavior

Head Position

As shown in Figure 12, there are numerous shifts in head position across seconds

as the various events unfold. As expected, while the experimenter was purposely

centering the infant, percentages of head turns to the Middle Zone were quite high in

contrast to those to the Target and Opposite Zones. Once the experimenter attracted the

infant's gaze toward the hiding location secss 6 9), an increasingly high percentage of

turns toward the Target Zone followed, with a resultant decrease in head turns toward the

Middle Zone. In effect, one response appears to replace the other as looking shifts from

the Middle, to the Target, zone. As intended, the clown's face on the screen, which

appeared at the beginning of the Delay Interval, attracted attention back to the Middle

Zone, as indicated by another reverse in the trends--percentages increase again to the

Middle Zone with an accompanying decease to the Target Zone. From the beginning of

the trial until the end of the Delay Interval, head turns to the Opposite zone remain low

throughout. During the Approach Interval, the trends reverse again, with the expected

decline in looking toward Midline. For the first time, however, this trend is accompanied

by increases in head turns to both the Target and Opposite Zone as infants decided where

to search.

In summary, there are numerous, but predictable, shifts in head position across the

various conceptual intervals. For sake of clarity, head position was analyzed separately in

each of these intervals except for the Centering Interval. The latter was omitted from

analysis since infants were oriented toward midline almost exclusively for the first 6

seconds of the trial, and because are were not any notable shifts in head position until the

onset of the Hiding Interval, the first interval to be examined statistically.

Hiding Interval Analyses

A Zone X Block X Seconds ANOVA during the Hiding Interval secss 6-9)

verified a substantial shift in head position between the Middle and Target Zone (see

Figure 11), as indicated by a main effect of Zone, F(, 16) = 73.38, p < .001 as well as a

Zone X quadratic Seconds, F(1,16) = 14.33, p <.01, interaction. Although a modest

effect of Block emerged as well, F(1,16) = 4.78, p < .05, inspection of individual blocks

suggests that the effect does not seem to reflect any meaningful changes in head position

(See Figures 13, 14 and 15). Coupled with the fact that this factor did not interact with

either Zones of Seconds, the block effect was deemed trivial and data were collapsed

across the factor for all subsequent analyses in the interval. Of particular interest, was the

separate analysis of each zone individually to discern the nature of the aforementioned

Zone X quadratic Seconds interaction. Thus, an ANOVA, collapsed across Block, was

conducted for each of the three zones. The results of these analyses suggest that the

interaction is due to the presence of strong quadratic Seconds trends in the Target,

F(1,16) = 77.60, p < .0001, and Middle, F(1,16) = 125.52, p < .0001, but not Opposite,

zone. Of course, the quadratic trends in the former two zones are opposite in direction as


Delay Interval Analyses

A Zone X Blocks X Seconds ANOVA, identical to that used to examine head

position in the Hiding Interval, was conducted for data in the Delay Interval secss 9-12).

As with the Hiding Interval, a linear Block effect, F(1,16) = 5.26, p < .05, was significant

here as well. Although it appears that this effect may be driven by the more notable

changes across block in the Middle zone data, Block did not interact with Zone,

suggesting that, as with the Hiding Interval, this effect is probably only the result of very

subtle, and inconsistent, fluctuations across blocks and is probably not very meaningful

(see Figures 13, 14 and 15). Thus, as before, the data were collapsed across Blocks in a

Zone X Seconds ANOVA so that changes in these two factors could be examined more

directly. The most compelling result from this analysis involved a main effect of Zone,

F(1,16) = 182.00, p < .0001, (see Figure 12)

In addition, Zone interacted with quadratic Seconds, F(1,16) = 54.00,

p < .0001, in this interval as well. As in the Hiding Interval, the interaction is most likely

a reflection of the quadratic nature of the data in the Target and Middle, but not Opposite,

zones. The separate analysis of each zone supported this assertion revealing strong

quadratic Seconds effects in the Target and Middle zones, Fs(1,16) = 38.26, 48.57, ps <

.0001, respectively, while failing to produce any significant trends across seconds

whatsoever in the Opposite zone.

A direct comparison of Target and Opposite Zones was of particular interest since

this contrast would yield the most information about differential anticipation. In order to

address whether infants were more likely to turn their heads to Target than the Opposite

Zone during the Delay Interval, another ANOVA, also collapsed across trial block and

excluding the Middle zone, was conducted to examine relationship between these two

zones and Seconds. As expected, there were interactions between the two factors

including a Zone X linear, quadratic and cubic Seconds, Fs(1,16) = 14.32, 13.70 and 7.80,

ps < .05, respectively, as well as numerous trends for Seconds alone. Analysis of each

zone alone indicated that, whereas both Target and Opposite zones each had significant

quadratic seconds effects, the Target zone had linear and cubic trends as well, suggesting

a greater level of complexity in head turns to the Target zone. When examined in the

context of the entire trial, it appears that this interaction is most likely the result of greater

decreases in turns toward the Target Zone (as infants shift toward the Middle zone), an

effect which is probably a continuation of a trend initiated in the Hiding Period (See Figure

12). The amount of decrease in the Target and Opposite Zones differs, presumably,

because infants were not as likely to be looking at the Opposite Zone during the Hiding

period to begin with. In either case, the trends discovered in the Delay Interval clearly do

not indicate anticipatory head turns to either Zone since they do not increase over


Approach Interval Analyses

During the Approach Interval secss 12 to 15), the Zones X Blocks X Seconds

ANOVA produced the same linear Block effect, F(1,16) = 44.32, p < .001, as some subtle

fluctuation across trial block remains (see Figures 13, 14 and 15). When the data were

collapsed across block, this second ANOVA revealed a significant Zone X linear Seconds,

F(1,16) = 91.6, p < .001, interaction as well as a main effect of Zone, F(1,16) = 31.08, p <

.001. Further analysis of individual zones in this interval, suggests that this interaction

may stem from the different levels of complexity in the Seconds effects in each of the three

zones. For example, whereas Target, Opposite and Middle zone analyses all produced

strong linear Seconds effects, Fs(1,16) = 105.66, 41.75, and 117.01, ps < .0001,

respectively, the latter two were characterized by higher order trends as well, suggesting a

more variable pattern across seconds in these two zones, (See Figure 12). More

importantly, however, is the fact that the trends across Seconds in the Target and Middle

zone are moving in the opposite direction from that in the Middle zone, yielding an

interaction which reflects yet another shift in head position.

The most important question of all, however, was the extent to which the Zone X

Seconds interaction in the Approach Interval analysis could also be driven by differences

between the Target and Opposite Zones in alone. As in the Delay Interval, an ANOVA

was conducted which directly compared these two zones against Seconds during the

Approach Interval. Indeed, the significant interaction between Zone and linear Seconds,

F(1,16) = 8.64, p < .009, remained, suggesting that there were significant differences in

the rate of increasing head turns toward Target and Opposite zones. Thus, although

infants increase their looking to both zones, they are even more likely to turn their heads

toward Target as the containers approached.

Search Behavior

In general, infants were quite capable of searching the containers and did so

frequently with the exception of two participants (the Mean % of trials with searches,

excluding these two infants, was .989). Data from these two participants, deemed

"outliers", were excluded from all of the following search behavior analysis because their

frequency of searching was so low relative to the other infants' (see Figure 16). In any

case, the infant behavior analyses in study 2 were broken into two sets: the 1) Time of

Search (defined as the point in time when the infant visibly displaced the hinged lid on the

container) and 2) Search Accuracy (whether the first container the infant searched was

the correct or not).

Time of Search

The actual time that the containers reached the end of the track and the time of

the infant's first contact with a container were determined (from videotape) for each trial,

for each subject. In general, infants were able to execute a search quickly and, on average,

did so at second 16.03, just one second after the containers typically reached the end of

the tracks. On many of these trials, however, infants actually opened the lids on one of the

containers during mid-approach, as it was moving toward them down the track (when

infants did search early, they did so, on average, .37 seconds before the container reached

the end of the track). These responses were categorized as "early" searches--those which

occurred before the containers reached the end of track and were determined on a trial by

trial basis by subtracting the search time from the time that the containers actually reached

the end of the track on that particular trial since there was some variability in when this

occurred. These scores were then used to determine, for each infant, the overall

percentage of early searches for each trial block (see Figure 17).

An ANOVA was then conducted to examine the relationship between two within

subject factors, Percentage of Early Searches and Blocks, and revealed a significant effect

of Linear Blocks, F(1,14) = 9.05, p < .01. Figure 17 suggests that this Block effect is

probably due to the increase in early searches between Block 1 and 2, specifically. In

order to test this assertion, a series oft tests, comparing successive pairs of blocks,

verified that, in fact, the only significant difference was between Block 1 and 2, t = -2.93,


Finally, the relationship between the time of the search and the infants' search

accuracy was examined by comparing the percentage of trials with early searches on

correct versus incorrect trials. Although infants were somewhat more likely to respond

early on correct (39% of trials, on average, were early) than incorrect (35% of trials, on

average, were early), the difference was slight and failed to reach statistical significance.

Search Accuracy.

The more interesting, second set of analyses, revolved around the infant's

knowledge of the spatial components of the task. To assess this, it was determined for

each trial whether the infant's first search was correct (they chose the container the toy

was hidden in), incorrect, or if no search occurred at all. On average, subjects were

correct (M = 62% of all trials with usable motor data) more often than incorrect M =

37% of all good trials)_regardless of which side of the table the toy was hidden on. Thus,

there were no significant differences between trials in which the correct responding was on


the right or the left side. Trials with no response were rare, accounting for only .7% of all

good trials, on average (See Figure 18). An Accuracy X Block ANOVA examined the

percentage of responses in each of the aforementioned accuracy categories across the five

trial blocks to determine whether 1) they were correct more often than incorrect, and 2)

whether their performance improved across trial blocks. The results of this ANOVA

supported the first assertion with a main effect of Accuracy, F(1,14) = 356.76,

E < .0001, verifying that infants were, in fact, correct a significantly greater portion of

time. However, as can be seen in Figure 19, there were no significant effects of Block.

Rather, the tendency to choose the correct container more often than the incorrect one

was acquired early and remained stable across the testing session.

Anticipatory Motor Behavior and Accuracy

As with HR, the relationship between motor behaviors prior to searching and the

infants' subsequent accuracy was of interest to discern whether infants who anticipated the

location of the toy were better able to solve the spatial task than infants who did not

anticipate. To answer whether anticipatory motor behaviors contributed to task success,

three analyses were conducted: 1) an ANOVA, in which the percentage of trials with

head turns oriented toward particular zones during each 0.5 sec of the Approach Interval

was compared on Correct versus Incorrect search trials, 2) an analysis comparing the

total percentage of seconds in the Approach Interval, which infants spent oriented toward

particular zones, on Correct and Incorrect trials, and 3) a Linear Regression, in which the

overall percentage of head turns toward the Target zone late in the Approach Interval was

tested as a significant predictor of overall search accuracy.

The first ANOVA is similar to the HR x Accuracy ANOVA with two exceptions.

First, whereas the HR data were subdivided into two trial blocks, the head position data

were not since, unlike with HR, there was little evidence that head position changed

significantly over trial blocks. Thus, collapsing across this factor permitted a larger

sample of trials in the "Correct" and "Incorrect" groups for each infant. Second, the

dependent measure for each second of data in the HR x Accuracy ANOVA was simply a

single HR change score. With head position, for each half second of data, the dependent

measure is the difference between the percentage of head turns toward what will be

referred to as the "Selected" versus the "Unselected" zone. These terms are imposed in

an effort to simplify the description of the first two analyses and the corresponding results.

In each accuracy condition the "Selected" zone simply refers to the zone which the infant

subsequently searched. On Correct trials, for example, the "Selected" zone is the Target

Zone (the "Unselected" zone is the Opposite zone) whereas for Incorrect trials, the

reverse is the case--the "Selected" zone is actually the Opposite Zone and the

"Unselected" zone is the Target Zone. Thus, this comparison, on Correct versus Incorrect

trials, was tested in a Zone (selected versus unselected) x Accuracy (correct versus

incorrect trials) x Seconds secss 12 14.5 in the Approach Interval) ANOVA.

As depicted in Figures 20a and 20b, infants were more likely to look toward the

Selected zone regardless of accuracy. One nuance of this pattern of responding, however,

is that this pattern appears stronger still on the Correct trials in particular. For example,

on Correct trials, infants become increasingly likely (on up to 65% of the correct trials at

the end of the Approach Interval) to turn toward the Selected zone. On incorrect trials,

however, infants seem less certain, exhibiting this preference on only 50% of the trials, at a

maximum, during the same time interval. When subjected to the aforementioned

ANOVA, however, this trend just missed reaching the level of significance, F(1,13)=

4.17, E = .06, suggesting that the difference a fairly small one. Nonetheless, there is a

consistently stronger, albeit statistically insignificant, preference for looking at the to-be-

selected container on Correct versus Incorrect trials-a trend which will be considered

further in the discussion.

One limitation of the .5 by .5 second evaluation, is that this dependent measure

only captures the percentage of trials, at a particular half second, with head turns toward a

particular zone. If infants were looking back and forth quite a bit, it would be difficult to

discern their overall preference from such a discrete sampling technique. Thus, a second

analysis was conducted which compared the total number of seconds during the Approach

Interval spent turned toward either the Selected or Unselected, within each trial type.

The data shown in Figure 21 basically mirror the second by second trend in Figure

20 in that infants' overall preference appears to be to look toward the Selected zone on

both Correct and Incorrect trials. It is clear, however, that when the data are collapsed

across seconds, the extent of this preference is even greater on Correct trials, with infants

spending relatively more time in the selected zone (Mean % of time turned toward the

selected zone = 33% versus 8% for the unselected zone)--a result which was statistically

significant in a paired sample t test, t = 3.77, p < .05 (2 tail). On trials with Incorrect

searches, this difference was smaller (Mean % toward the selected zone = 28% versus

14% toward the unselected zone) and failed to reach significance in a similar t test. Thus,

although the .5 second x .5 second trends on Correct and Incorrect trials were not quite

significantly different from one another, the overall amount of time spent turned toward

the respective zones was. This is significant because it bolsters the initial assertion that the

motor behaviors preceding accurate versus inaccurate searching are, in fact, different: the

infants' overall preference for the zone they subsequently search is greater preceding

searches which are ultimately correct than on those which prove incorrect.

One final analysis--a Linear Regression--was conducted to determine whether

Accuracy could be actually be predicted directly by either 1) the percentage of Early

Searches or 2) the percentage of head turns to Target Zone during second 14.5 in the

Approach Interval. As in the HR regression, second 14.5 was chosen because it was late

enough in the Approach Interval to maximize the chances of capturing increasing

probabilities of head turns to the Target container, yet it was before the earliest time that

any infants actually retrieved a toy. Thus, it was the latest point of anticipation during the

Approach Interval that would not be contaminated by head turns to a visible toy.

The results of the two regressions indicated that, although the prevalence of Early

Searches did not have any predictive effect on Accuracy, the percentage of head turns at

second 14.5 did, F (1,13) = 19.13, p < .005, lending further support to the notion that

anticipatory head turns may directly influence or at least predict the infant's success at

obtaining an object of interest.

In summary, of the four analyses examining the relationship between motor

behaviors and search accuracy, only the analysis of the infants' time of search x accuracy

failed to suggest any relationship between behavior during the Approach Interval and

performance. The three head turn x accuracy analyses, in contrast, produced much more


promising results with two of these (t tests on the overall percentage of time turned to

each zone and the linear regression using second 14.5) clearly suggesting that increased

looking toward the selected zone prior to search is related to subsequent accuracy and one

of them nearing the level of significance (the ANOVA assessing the .5 x .5 head turn

data). Thus, although the time at which the infant executed the search does not appear to

influence accuracy, the position of the infant's head prior to search does, suggesting that

these early patterns of looking may be most influential in impacting task performance.

14 & 15-mo-old

HR N=17

Centering Hiding Delay Approach


S1.5 3.5 5.5 7.5 9.5 11.5 13.5 15.5 17.5


14 and 15-mo-old Heart Rate:
Mean HR change scores across infants, on Trial Blocks 1 to 5











14 & 15-mo-old HR N = 13 (a)
(M search time = 16.03)

Deby ...,
2 2


-4 correct

8.5 10.5 12.5 14.5 168.5 18.5


14 & 15-mo-old HR N = 13
e ___N(M search time = 16.03)
Dely A~

2 -2 .

Mean HR change scores (across infants) &

Across trials in the first (a) and the second (b) trial block
% -6'- ------- ----



Mean serach accuracy (across infants) on Correct and Incorrect trials
Across trials in the first (a) and the second (b) trial block


14 and 15-mo-old Head Position: Mean Percentage of trials (across infants and trial blocks)
with head toward Target, Opposite and Middle Zones

14 & 15-mo-old HEAD POSITION (N = 17)
% of all Good Trials head in each Zone
CntriD Inht.m, Hiding Delay Approch (M rh ume = 1603)




S 0 1 2 3 4 5 6 7 8 9 10 11 1213 14 15 16 1718 19 20

S% to Target A- % to Opposite -- % to Middle

14 & 15-mo-old HEAD POSITION N = 17
% Trials head in Target Zone x Block
1 (M sam:h time 16.03)
. 0.4

S 0 1 2 3 4 5 6 7 8 9 1011121314151617181920

-.-BLK1 -- BLK 2 -w- BLK 3
-- BLK 4 BLK 5


14 and 15-mo-old Head Position: Mean Percentage oftrials (across infants and trials)
with head toward Target Zone onTrial Blocks 1 to 5

14 & 15-mo-old HEAD POSITION N = 17
% Trials head in Opposite Zone x Block
) (M search time = 16.03)
C 1
N Centering__ Hiding Delay Approach
z 0.8
= 0.6
I--_ __ _
- 0.4

* 0.2

e= 0
- 0 1 2 3 4 5 6 7 8 9 1011121314151617181920
s Seconds

-- BLK 1 BLK 2 -v- BLK 3
-- BLK 4 -- BLK 5


14 and 15-mo-old Head Position: Mean Percentage of trials (across infants and trials)
with head toward Opposite Zone onTrial Blocks 1 to 5


14 and 15-mo-old Head Position: Mean Percentage of trials (across infants and trials)
with head toward Middle Zone onTrial Blocks 1 to 5

14 & 15-mo-old HEAD POSITION N = 17
% Trials head in Middle Zone x Block

S Centering iding Appa (M search time = 16.03)
N 0.8B 2
r 0.4

0 4
0 1 2 3 4 5 6 7 8 91011121314151617181920

B--BLK1 -i-BLK2 -w-BLK3
-v- BLK4 BLK5


14 and 15-mo-old Search Behavior: Total number of searches for each infant

14 & 15-mo-old Search Behavior
# of Searches for each Participant



o 10

L 5

10 12 17 18 21 22 25 26 28 32 33 34 35 36 37 38 39
Participant #

14 & 15-mo-old Searching Behavior, N = 15
% Trials, per bik, with Early Searches


m 0.8


S0. ________________
0 -

D" 0.4 --- --
< 0.2 ---
1 2 3 4 5


14 and 15-mo-old Search Behavior:
Mean percentage of early searches (across infants) on each Trial Block


14 & 15-mo-old Search Behavior, N = 15
Search Accuracy, Across all Blocks



.2 -------

0 L-
Response Category


14 and 15-mo-old Search Behavior: Mean accuracy (across infants & trial blocks)


14 and 15-mo-old Search Behavior: Mean accuracy (across infants) in each Trial Block

14 & 15-mo-old Search Accuracy, N = 15
on each Trial Block


o 0.6
I 0.4
5 0.2 -

0 .


14 & 15-mo-old Head Position (N = 14) (a)
% Turns to each zone on Correct Trials

1------------------------- -
0.8 ..........-.- .... --.. .. -.. ...... Selected Zone
S--- -- Unselected Zone
0.4 ..... -- -- i _

o 0.2'

12 12.5 13 13.5 14 14.5
Seconds During the Approach Interval

14 & 15-mo-old Head Position (N = 14) (b)
% Turns to each zone on Incorrect Tris

0 nSelected Zone

Unselected Zone


12 12.5 13 13,5 14 14.5
Seconds During the Approach Interval


14 and 15-mo-old Head Position during the Approach Interval and Accuracy:
Mean percentage head turns (across infants) toward Selected & Unselected Zones,
across all Correct (a) and all Incorrect (b) trials

14 & 15-mo-old Head Position, N = 14
% Approach Interval x Zone x Accuracy
N 0.8

S0.6 Selected

E 0.4
Im 1Unsel.
c 0.2

Correct Incorrect

Trials grouped by Search Accuracy

14 and 15-mo-old Head Position during the Approach Interval and Accuracy:
Mean percentage of Approach Interval (across infants) head in Selected & Unselected Zones,
across all Correct and all Incorrect trials


The main objectives of the second study were to 1) examine whether infants could

utilize spatial cues to exhibit generalized and/or differential anticipation over a moderate

delay, and 2) to determine whether differential anticipation, if it did occur, was an

independent, functional process which then facilitated performance on the two-alternative

spatial task. The bulk of the initial discussion will be devoted to the consideration of the

first objective (HR and motor behaviors will be addressed separately). Following, is a

separate section which addresses the second objective exclusively-whether the cardiac

and/or motor data collected provide evidence that anticipation is an independent and

functional process.

Generalized Anticipation (Heart Rate)

The most dramatic changes in HR do not occur until the onset of the Delay

Interval, when there is a sharp deceleration, peaking at the onset of the containers'

approach, which is then maintained throughout the Approach Interval. However,

although HR clearly decelerated across these two intervals, it is difficult to distinguish

between a discrete D1 and D2, two theoretically independent components of the typical

triphasic anticipatory HR response. In addition, the response appears to decay over trial

blocks, counter to expectation, leaving the following questions: 1) What is it the infants

are reacting to in these two intervals? and, 2) Is there sufficient evidence to conclude that

the decelerations reflect anticipation or are the data better fit to alternative cognitive


Although HR is not sensitive to the spatial location of the hidden toy in study 2, it

was included nonetheless since it could be sensitive to the temporal parameters of the

procedure. Thus, it could provide useful information about infants' ability to anticipate,

generally, when critical events are due to occur. However, the temporal parameters of

study 2 were designed to facilitate the anticipatory motor behavior, and because of the

necessary complexity they were less than ideal for eliciting meaningful HR patterns. For

example, in order to maximize anticipatory motor behaviors while providing appropriate

controls, more stimulus events were presented with relatively short intervals between

them. The resulting theoretical complication for HR is that any one of these events could

have served as the equivalent of a warning cue (or S1) making it difficult to know which

events elicited the HR decelerations. Furthermore, the shorter intervals do not allow

much time for multiple components of the HR response to occur. Thus, the complexity of

the design risks raising more questions about the cognitive processes associated with the

HR response than can be thoroughly answered and the interpretations provided should be

considered with caution. However, it is useful to explore the following possibilities


The HR decelerations that occurred across the Delay and Approach Intervals

could be interpreted as a reflection of several cognitive processes. Three specific

interpretations of the data will be considered: 1) that the HR was, essentially, an orienting

(or D1) response to stimulus events presented at the beginning of the respective intervals,

2) that the HR decelerations reflected anticipation of stimulus events occurring at the end

of the intervals and 3) that the HR responses reflected neither of these two traditional

processes but, rather, were a manifestation of cognitive processes other than orienting

and anticipation.

The first possibility--that infants were merely orienting to the novelty of either the

distracter stimulus in the Delay, or the appearance of the containers in the Approach

Interval--is the simplest explanation. There are at least two compelling reasons why this is

appears to be the case. First, the decelerations do not increase in magnitude over trial

blocks in either interval. In fact, the response seems to habituate over blocks, a

characteristic which is more typical ofDls or orienting responses. Perhaps as the novelty

of the respective stimuli waned, infants became less responsive upon their presentation,

accounting for the progressively smaller decelerations across blocks. Secondly, in the

Delay Interval the HR decelerations are accompanied by overt head turns, oriented almost

exclusively, to the Middle zone where the clown face is located suggesting that, at least in

this interval, this is where the infants' attention was focused.

It is possible, however, that despite habituation of the decelerations over blocks

and the head orientation to midline during the delay, infants were anticipating something

at the end of each of these intervals. For example, based on the motor data, it is

reasonable to assume that infants learned enough about the temporal parameters of the

design to anticipate either 1) when the containers would be released, or 2) the exact

point in time when the containers would be within reach (at the end of the Approach

Interval), making toy retrieval possible. In regard to the former, anticipation of the

containers' initial appearance could have been strong enough to influence HR, producing

some element of a D2 in addition to a strong orienting response during the Delay Interval.

The fact that the infants' heads were oriented to midline is not necessarily inconsistent

with an anticipation account of the HR data if infants were anticipating the time that both

containers would emerge but not the location of the hidden toy. Anticipation is an even

stronger explanation of the HR data in the Approach Interval since it is bolstered by the

head turn data which strongly indicates differential anticipation of toy location.

Of course, it is also possible that the HR in both intervals resulted from combined

response in anticipation of events which occurred at the end of the Approach Interval.

Thus, it is possible that the distracter served as a cue for later searching, producing an

initial deceleration during the delay which was sustained throughout the containers'

approach. Thus, the sustained decelerations in the Approach Interval may simply be the

second wave of deceleration in a triphasic HR pattern, (the D1 of which was initiated

early in the Delay Interval), reflecting the infants anticipation of toy retrieval.

Finally, the decelerations which occurred across these intervals may not involve

either orienting or anticipation. Rather, they could be associated with alternative cognitive

processes such as stimulus intake as outlined by Lacey & Lacey (1970) or sustained

attention (Porges; 1980, & Richards; 1985). The latter of these two alternatives, in

particular, will be discussed at length below.

In summary, the evidence in support of an orienting response account of the data is

rather compelling, especially in the Delay Interval where there is the largest amount of HR

change across the interval. It is reasonable to presume that the distracter stimulus,

intended to briefly reorient infants to midline, was, in fact, too interesting, capturing their

attention almost entirely and for a longer amount of time than intended. The head position

data certainly support this shift in attention as infants' heads were positioned in that zone

nearly 80% of the time over much of the Delay Interval. The fact that the decelerations

dissipate over trials also supports the notion that the HR pattern is a Dl, a response

known to habituate over trials.

The salience of the clown face may have been so intense that this D1 was

maintained throughout the Approach Interval as well. However, that the HR in the

Approach Interval is merely a continuation of decelerations in the first interval does not

seem likely given that the head position data clearly indicate that infants are orienting away

from, not toward the clown face in the Middle Zone. It is possible that infants are

responding to the novelty of the moving containers. However, if this were so one would

expect to find a distinct, second wave of decelerations, an outcome which did not occur.

Finally, if the infants were merely reacting to the novelty of moving objects generally, then

they should have been equally interested in either container. However, the motor data

clearly indicate that infants exhibited a preference for the Target container in particular,

suggesting that their attention was summoned by more than the perceptual qualities of the

container. Therefore, it is reasonable to suspect that the infants were focusing on an

additional, unseen, aspect of the approaching container (perhaps the mental representation

of the location of the toy hidden inside), and that this process may have been reflected in

the HR as well. Thus, whereas orienting may account for the HR pattern in the Delay

Interval, it is probably an insufficient explanation for the sustained decelerations

characteristic of HR in the Approach Interval. However, although these facts strongly

suggest that HR during this interval reflect anticipation, they leave unanswered the

question of why decelerations failed to increase over blocks.

One possible means of resolving whether decelerations in the Delay and Approach

Interval were Dis or D2s is by comparing the infants' HR to their overall accuracy on the

task. It is reasonable to imagine that, for example, if the decelerations are truly a *

continuation ofa DI elicited at the onset of the Delay Interval (as the infant oriented to

the clown on the distracter screen) then attention to the memory task would be usurped, in

essence, by the distracter, leading to worse performance on location recall for larger D s.

On the other hand, if the decelerations reflect an anticipatory process, a D2, as the infant

prepares to search in the correct container, then those trials with accurate searches should

be associated with larger decelerations.

The results from the HR X Accuracy ANOVA, in which the magnitude of HR

decelerations were compared on trials on which the infant was correct versus incorrect, on

the first and second half of the testing session (see Figure 11) helps to resolve this issue.

Upon first glance at this data, it is clear that, in general HR is more deceleratory on the

first than the second block, an effect which seems to be consistent with the orienting

hypothesis. Likewise, the difference in HR on the correct and incorrect during the Delay

Interval was not significant, an outcome which appears to reinforce the idea that infants

were simply orienting to the distracter during this interval. However, the orienting

hypothesis does not explain the significant difference in HR, during the Approach Interval,

on correct versus incorrect trials. The clown face is the same on all trials, regardless of

the infants' accuracy yet infants are decelerating more in this interval on the trials on

which they are ultimately correct, at least in the first block. During the Approach Interval,

habituation of an orienting response may account for the overall differences between

block 1 and 2 but it can not account for the difference in HR, on correct and incorrect

trials, within the first block. Furthermore, given that the stimuli presented on trials in

block 1 are perceptually identical, the only reasonable factor left which can account for the

differential HR on these two types of trials (correct versus incorrect), is a different

cognitive process, at work early in the session, which precedes correct versus incorrect


One potential candidate for such a process is "sustained attention", as

conceptualized by Porges (1976, 1980), Richards (1988) and Richards & Casey (1992).

In essence, sustained attention is defined as a specialized form of attention, in which

"orienting is maintained or amplified in order to process information in the stimulus" (Berg

and Richards, 1997, in press, pg 7). Furthermore, sustained attention is postulated to

contain two important subcomponents: a "selective" quality (the tendency to "focus

information processing on specific aspects of a task or stimulus") and an "intensive"

quality (referring to the "functional" effect that sustained attention has on the task at


In Richard's paradigm, 14, 20 and 26-week-old infants were exposed to one of

two trial types: 1) trials in which a single, centrally located, stimulus was presented, and 2)

those with the same central stimulus plus a competing stimulus in the periphery. HR was

monitored on both types of trials and suggested that the older infants, especially, oriented

to the central stimulus on both trial types, for 4-5 sees, producing HR decelerations which

remained below baseline throughout those seconds. Thus, even on trials in which there

was a distracter, infants were able to select the relevant (central) stimulus and sustain their

attention to it (as indicated by the below baseline HR), ignoring the distracting event in the

periphery (Richards, 1985).

Although the HR patterns exhibited in study 2 do not conform entirely to the

theoretical definition of sustained attention it is possible to use this conceptual framework

to describe much of what is occurring in study 2. For example, in Richard's paradigm

infants focused and sustained their attention to a centrally located stimulus, enabling them

to encode this stimulus as most relevant and, therefore, resist looking to a distracter

presented in the periphery. Perhaps infants in the present study are focusing attention,

utilizing the same basic process (attend to the relevant information), but in the opposite

direction: they are focusing on the periphery, especially toward the Target zone, in order

to resist distraction from the central stimulus (the clown face). This preference certainly

fits with the "selective" aspect of sustained attention as infants focus attention on the

Target container in particular during the Approach Interval. It also supports the presence

of the "intensive" component of sustained attention as trials with the larger decelerations

are, indeed, those on which the subjects are more accurate, suggesting that the focus of

attention upon approach, facilitates correct searching behavior.

One slight discrepancy between the sustained attention scenario for HR in the

Approach Interval and Richard's conceptualization of sustained attention is that

decelerations accompanying sustained attention are supposed to reflect the initial

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EMHPK1I7J_S2GCOA INGEST_TIME 2013-09-28T02:26:48Z PACKAGE AA00014283_00001