The Stability of Visual Habituation during the First Year of Life

Author(s): John Colombo, D. Wayne Mitchell, Marion O'Brien and Frances Degen Horowitz
Source: Child Development, Vol. 58, No. 2 (Apr., 1987), pp. 474-487
Published by: Wiley on behalf of the Society for Research in Child Development
Stable URL: http://www.jstor.org/stable/1130524
Accessed: 27-06-2016 05:35 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact [email protected].

Wiley, Society for Research in Child Development are collaborating with JSTOR to digitize,
preserve and extend access to Child Development

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
The Stability of Visual Habituation during the
First Year of Life

John Colombo, D. Wayne Mitchell, Marion O'Brien, and

Frances Degen Horowitz
University of Kansas

COLOMBO, JOHN; MITCHELL, D. WAYNE; O'BRIEN, MARIoN; and HoRowrTz, FRANCES DEGEN. The
Stability of Visual Habituation during the First Year of Life. CHILD DEVELOPMENT, 1986, 57, 474-
487. The short-term reliability and long-term stability of visual habituation and dishabituation in
infancy were assessed in a sample of 186 infants from 4 age groups (3-, 4-, 7-, and 9-month-olds) seen
for 2 within-age sessions and in a sample of 69 infants seen longitudinally at 3, 4, 7, and 9 months of
age. Moderate week-to-week reliability (r's = .30-.50) was observed for duration-based and mag-
nitude of habituation variables at all ages, although better reliability was evident at 4 and 9 months
than at 3 and 7 months. In most cases, the reliability of habituation magnitude measures was
attributable to the reliability of the peak fixation alone. Data from the longitudinal sample suggested
that only the duration of peak fixation was consistently stable across the ages tested, although
stability for several measures emerged across the 7-9-month testing. No consistent reliability or
stability emerged for the presence or magnitude of dishabituation in either sample.

Early reports on the visual attention of sis makes individual differences in habitua-
the human infant (Berlyne, 1958; Fantz, 1958) tion particularly interesting to investigators
presented new paradigms for research in de- searching for early developmental continuity
velopmental psychology. Following Fantz's or for measures in infancy that might prove
(1964) initial report of a decline in fixation predictive of intellectual or cognitive out-
time to a redundantly presented visual stimu- come in later childhood.
lus, a sizable theoretical literature grew
around the habituation of infants' visual atten- A major step toward the assessment of
tion. Much of the current literature on infant individual differences in habituation was the
habituation is based on a model of habituation development of the infant-control proce-
proposed by Sokolov (1963) that posits a wan- dure (Horowitz, Paden, Bhana, & Self, 1972),
ing of the orienting response as a function of which replaced experimenter-determined
the formation of an internal representation of presentation durations with habituation crite-
the stimulus shown. Although alternative ac- ria based on the infant's own individual pat-
counts of visual habituation have also been tern of fixation. Within this procedure, infants
set forth (e.g., Bashinski, Werner, & Rudy, may vary on the total amount of attention di-
1985; Dannemiller & Banks, 1983; Rovee- rected toward the stimulus (Mitchell &
Collier & Gekoski, 1979), the assumption Steiner, 1984), on the speed or rate of their
common to these theoretical approaches has habituation (DeLoache, 1976; McCall, Ken-
been that visual habituation reflects some nedy, & Dodds, 1977; Miller et al., 1979; Mil-
type of learning in the infant. This hypothe- ler, Sinnott, Short, & Hains, 1976), and on

This work was supported by a National Institute of Child Health and Human Development
(NICHHD) postdoctoral fellowship (5T32HD07173-04), Public Health Service Biomedical Re-
search Resources (University of Kansas Biomedical grant 4309), and by an Alcohol, Drug, and
Mental Health Administration grant (1 R01 MH41395-01) awarded to John Colombo. Preparation of
the manuscript was supported in part by a NICHHD grant (1 R01 HD18290-lA1) to Frances
Degen Horowitz. These results were presented at the International Conference on Infant Studies in
Los Angeles, CA, April 1986. We gratefully acknowledge the assistance of Carla Dykeman and
Laura Fine in subject recruitment, Diana Anderson-Goetz, Yvonne Caldera, Madelyn Moss, and
Keith Nagle as observers, the staff of the University of Kansas Regents Center for their cooperation,
Dr. Diane Bythell and Jeff Coldren for their comments on an earlier version of the manuscript, and
Dr. Susan Embretson for statistical consultation. We are especially grateful to the parents of the 255
infants who participated in this set of studies. Address correspondence and requests for reprints to
John Colombo, Infant Study Center, Department of Human Development, University of Kansas,
Lawrence, KS 66045.

[Child Development, 1987, 58, 474-487. ? 1987 by the Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/87/5802-0016801.00]

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
 Colombo et al. 475

their pattern of fixations before habituation pose (i.e., discrimination of the affective cate-
(Bornstein & Benasich, 1986; McCall, 1979). gory).

Despite the acceptance of visual habitua- Taken together, these studies suggest
tion as an indicator of important cognitive pro- that at least some aspects of infant visual
cesses and its widespread use in infant re- habituation constitute reliable individual dif-
search (for a recent comprehensive review, ference measures. Direct comparison among
see Bornstein, 1985), there are remarkably these studies is difficult, however, as they
few data on its basic parametric or psycho- vary in the age of the infants tested, in inter-
metric features. For example, throughout the test intervals, and in the specific habituation
period of research on individual differences procedures employed. More systematic esti-
in visual habituation, estimates of such reli- mates of the consistency and stability of early
ability were essentially lacking in the litera- visual habituation measured under consistent
ture, and the few reports available on this procedural and stimulus conditions are still
topic were not encouraging (McCall, 1979). needed, especially in light of recent indica-
More recently, some scattered estimates of tions that early visual behavior may be pre-
the internal consistency of early habituation dictive of later intellectual outcome (Fagan &
and dishabituation measures have begun to Singer, 1983; Miller et al., 1979; Miller et al.,
appear. For example, Ritz, Woodruff, and Fa- 1977; Miller, Spiridigliozzi, Ryan, Callan, &
gen (1984) tested 4-month-olds twice within a McLaughlin, 1980; Rose & Wallace, 1985;
session and again 1 week later and examined Sigman, Cohen, Beckwith, & Parmelee, 1985;
the consistency of total looking time and num- Slater, 1985; see also Bornstein & Sigman,
ber of looks to habituation. Total looking time 1986). We report here the results of study of
was significantly correlated within the im- the short- and long-term stability of infant vi-
mediate sessions, but habituation rate (num- sual behavior at four points during the first 9
ber of looks to habituation) was not, and months of life. One cross-sectional sample of
neither measure was reliable 1 week later. infants was tested twice at each of four ages
More recently, Bornstein and Benasich (1986) (3, 4, 7, and 9 months), while another group
examined the reliability of habituation in 5- was tested longitudinally at each of these four
month-old infants across a 1-week interval ages. All the sessions were conducted using
with both social (faces) and nonsocial (geo- standard infant-control habituation proce-
metric figures) visual stimuli. Using a latent dures run under similar stimulus conditions,
variable approach derived from factor analytic therefore providing data that are directly com-
methods, they report some internal consis- parable across ages and samples.
tency for a habituation "construct" (consisting
of habituation slope, length of initial looking, Method
and percent decrement from initial looks to
criterion). Bornstein and Benasich also re- SUBJECTS
ported moderate test-retest stability for three The cross-sectional sample consisted of
habituation patterns initially described by 186 infants from four age groups: 3, 4, 7, and 9
McCall (1979). months of age. A longitudinal sample con-
sisted of 69 infants, with each infant tested
A separate and largely unexamined issue
once within a week of the third-, fourth-,
concerns the stability of habituation measures
seventh-, and ninth-month birthdays for a
across early development. McCall's (1979)
maximum of four testings. All infants were
longitudinal study of infants' visual and car-
recruited by mail and telephone from the
diac responses in fixed-trial procedures at 5
largely Caucasian, middle-class suburbs of
and 10 months of age yielded little individual
Kansas City, Missouri. Attrition data due to
stability and little cross-measure reliability.
fussiness/sleepiness or failure to return for ap-
Miller et al. (1979) used a fixed-trial paradigm
pointments (two consecutive broken appoint-
to examine the longitudinal stability of habit- ments or notification of withdrawal from the
uation rate (i.e., magnitude of decrement)
study) across the tests for the cross-sectional
from early infancy (at 2, 3, or 4 months of
sample are shown in Table 1 and for the lon-
age) to 27 and 39 months of age and found
gitudinal sample in Table 2.
moderate stability. More recently, Byrne,
Clark-Tousenard, Hondas, and Smith (1985) STIMULI
habituated infants to a series of different smil-
ing individuals and found significant consis- The habituation and dishabituation stim-
tency across ages of 4-7 months in total look- uli were eight color slides of individuals'
ing time, duration of first look, average look, faces. Four of the pictured individuals were
and dishabituation to a different emotional female, and four were male. Males did not

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
476 Child Development
TABLE 1

ATTRITION DATA FOR THE CROSS-SECTIONAL SAMPLE

Age Group Total N Lost Test 1 Lost Test 2 Final Na

3 months ..... 47 9 state change 4 no returns 34

4 months ..... 46 4 state change 8 no returns 34
7 months ..... 56 ... 6 no returns 50
9 months ..... 37 ... 3 no returns 34
Total 186 13 state change 21 no returns 152
a Infants on whom data for two complete habituation sessions are available and who
are included in all analyses involving habituation variables. An additional 12 infants (five
3-month-olds, four 4-month-olds, one 7-month-old, and two 9-month-olds) fussed or fell
asleep during posthabituation sequences and were therefore eliminated from dishabitua-
tion analyses.

have facial hair, and none of the individuals APPARATUS

wore glasses. All wore a dark blue jacket to
Testing took place in a 2 x 3-m booth
conceal differences in clothing, and all were
covered on three walls and on top with black
photographed in front of the same dull yellow fabric. The fourth wall was constructed of
background.
black plywood, and a .3 x .3-m translucent
screen was set in its center. Stimuli were rear
We chose faces as stimuli for two reasons.
First, we wanted to strike a balance between projected onto the screen during the session,
and infants viewed stimuli while sitting on a
assessing consistency of visual behavior with
parent's lap. Infants sat approximately .5 m
identical stimuli (in which case, results might
from the testing wall. The screen subtended a
be severely limited in generality) and assess-
visual angle of 280.
ing consistency across stimuli that varied
completely (where such differences might ob- Observers viewed infants through 5 mm
scure or override any consistency that ex- peepholes situated on either side of the
isted). The facial stimuli we chose were not screen. Fixations were coded via corneal
identical in individual features and yet were reflections, with fixation timing and habitua-
not drastically different from one another in tion criterion calculations under the control of
their overall configuration. The second factor a microprocessor. An additional observer con-
relevant to our choice of stimuli was that fa- trolled presentation of the slides and viewed
cial stimuli provide a class of stimulus that the infant via video monitor. In approximately
has been used successfully in many studies 25% of the sessions, two observers coded
across the entire span of the first year (Byrne fixations for reliability purposes. Observers'
et al., 1985; Fagan, 1974; Rose, 1981). Addi- records of individual fixation durations had a
tionally, Bornstein and Benasich (1986) found mean correlation of .97 (SD = .10) across all
reliability to be similar in studies using both ages, and observer reliability did not vary
faces and geometric patterns with 5-month- across infants' ages (.94 at 3 months, .99 at 4
olds. months, and .97 at 7 and 9 months).

TABLE 2

ATTRITION DATA FROM LONGITUDINAL STUDY

Age at Infants Infants Missed

Test Tested (N) Completed (N)a Testing
3 months ......... 69 58
4 months ......... 60 57 9
7 months ......... 53 50 7
9 months ......... 42 42 11
Total 207 27

a Completion refers to habituation sequence only. This table shows only those in-
fants who completed testing at any particular age; these infants may or may not have
completed any previous testing. A total of 35 infants completed habituation testings at all
four ages, 27 of whom also completed posthabituation testings at all four ages.

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
 Colombo et al. 477
PROCEDURE habituation or posthabituation stimulus) on
Infants were seated in the booth on a par- completing the four longitudinal visits.
ent's lap and habituated to one of the face
stimuli using a standard infant-control visual VARIABLES OF INTEREST: HABITUATION
habituation procedure (Horowitz et al., 1972). INDICES
Parents were given explicit instructions not to
interfere with the infant's looking, and in all
Eight continuous habituation measures,
cases followed these instructions. plus a categorical variable of habituation pat-
tern, constituted the core variables for stabil-
The criterion for habituation was two ity analyses in this study. Each variable has
consecutive fixations at a decrement of 50% or served as a focal point in various past research
more from the mean of the two longest previ- reports on the topic of visual habituation.
ous fixations within the session. The habitua-
Duration Variables
tion phase lasted for a minimum of five looks,
as the initial criterion was calculated from the a) Duration of peak fixation.-Length of
two longest fixations out of the first three the longest fixation observed during the
habituation sequence.
looks. A fixation was defined as any look at
the stimulus 1 sec or longer in duration, and b) Duration of first fixation.
fixations were terminated by a look away of
1 sec or longer. These parameters have been
c) Average fixation duration.-Calcu-
lated by dividing total fixation time by num-
shown to maximize interobserver reliability
ber of fixations to habituation.
and to minimize subject loss (Colombo &
Horowitz, 1985). d) Total fixation time.-The sum of the
The habituation procedure was the same duration of all fixations to the habituation
for all infants, but infants were randomly as- stimulus, again including the two criterion
fixations.
signed to one of two posthabituation condi-
tions. Approximately two-thirds of each age e) Total interlook interval.-After the
group was assigned to the Immediate Change termination of a fixation within the standard
condition, in which they were shown four infant-control procedure, the stimulus is
presentations of a novel face immediately fol- turned off for 2 sec and then reilluminated.
lowing habituation. Novel faces were se- Since the latency to return to the stimulus af-
lected randomly from the pool of faces oppo- ter it has been reilluminated can vary, Mitch-
site in gender from the habituation face. The ell and Steiner (1984) have suggested that the
remaining third of each age group was as- time infants spend "off" the stimulus be-
signed to the Lagged Change condition, in tween fixations (which, when totaled for the
which they were shown the habituation face session, yields the "total interlook interval"
for four more fixations before the presentation variable) may be important to infants' dis-
of a novel face for four fixations (Bertenthal, crimination performance, either in terms of
Haith, & Campos, 1983). In each sample, the memory interference (Rose, 1981) or perhaps
individual infant's assigned condition (Im- as a measure of motivation.
mediate Change or Lagged Change) was held
constant across repeated testings. Magnitude of Habituation
The amount of habituation shown by an
For the cross-sectional sample, infants infant may vary (McCall & Kagan, 1970).
were recruited for a first testing within 1 week Ruddy and Bornstein (1982) have also re-
of their age-group birthday and were seen for ported that a measure of the magnitude of
a second session 1-2 weeks thereafter. The
habituation was predictive of later cognitive
intertest interval for all age groups averaged outcome. Similarly, Miller et al. (1977) and
8.1 days (SD = 3.0 days) and did not vary Miller et al. (1980) have reported relations be-
among age groups. The second testing was tween a decrement measure and later per-
identical in procedure to the first, except that formance on a number of cognitive and lan-
the gender of faces shown in the second test guage measures. We calculated two such
habituation and posthabituation sequences decrement measures through the 50% habitu-
was opposite of that shown in the first test ation criterion.
sequences.

For the longitudinal sample, stimuli for a) Magnitude of habituation from looks
each testing were chosen so that the gender of 1-2.-Calculated by the formula
the habituation/posthabituation faces was al-
ternated for each infant across adjacent tests (M12 - GRIT)
and so that infants saw each face once (as a NFIX

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
478 Child Development
where M12 is the mean duration of the first other paradigms that purport to study prever-
two looks, CRIT is the mean duration of the bal perception and cognition and has been
two criterion looks, and NFIX is the number found to correlate with later measures of intel-
of fixations through habituation. lectual outcome (e.g., Fagan & Singer, 1983;
O'Connor, Parmelee, & Cohen, 1984; Rose &
b) Magnitude of habituation from the
Wallace, 1985). Because dishabituation may
peak look.-Calculated by the formula
be considered as either a categorical or a con-
tinuous measure, we examined the reliability
(PEAK - CRIT) and stability of both the presence and the
(NFIX - LOCPK)' magnitude of dishabituation.

where PEAK is the duration of the peak We categorized the presence or absence
of dishabituation on the basis of the differ-
fixation, CRIT is the mean duration of the two
ence between the durations of infants' last cri-
criterion fixations, NFIX is the number of
terion fixation and their first look at the novel
fixations through habituation, and LOCPK is
the location of the peak look in the sequence. face. If this difference was larger than the dif-
ference between the last criterion fixation and
Nonduration Variables first posthabituation fixation for infants in the
a) Number offixations through habitua- Lagged Change condition (i.e., where no
tion.-The number of looks to the stimulus to change of stimulus occurred) from the same
habituation has been used as one index of the age group, infants were categorized as "dis-
rate or speed of habituation (McCall et al., habituators."
1977; Ritz et al., 1984). We have included the
two criterion fixations in this variable. For magnitude of dishabituation, we cal-
culated dishabituation using the proportion
b) Location of peak fixation in the se-
quence.-McCall (1979) has suggested that NOV1
different habituation patterns may be charac-
(NOV1 + LAST)'
terized by the placement of the peak fixation.
This variable is the ordinal value (e.g., 1 = where NOV1 is the duration of fixation to the
first, 2 = second, etc.) of the fixation on which
first presentation of the novel face, and LAST
the peak fixation occurred.
is the duration of fixation to the last presenta-
c) Habituation pattern.-Bornstein and tion of the habituated-to face. A proportion
Benasich (1986) and McCall (1979) have was chosen to avoid the statistical problems
described three global types of habituation associated with the use of difference scores
patterns. The three types of patterns may be (Appelbaum & McCall, 1983).
generally characterized as (1) "exponential
decrease," typified by a steep decrement from
a peak fixation occurring early in the se- Data Reduction and Preliminary
quence; (2) "increase-decrease," in which the Analyses
infant actually increases fixation time before HABITUATION INDICES
the decrement to criterion; and (3) "fluctuat-
ing," which is characterized by at least two Because infant fixation data are often ex-
peak looks, which may come late in the tremely variable, we tested the various distri-
habituation session, or by at least two changes butions for normality and tested individual
in the direction of fixation time (i.e., increase scatterplots for outliers. Most of the distribu-
or decrease) before reaching habituation crite- tions were in fact severely skewed, and there-
rion. fore we applied square-root transformations to
all raw habituation variables to normalize the
Bornstein and Benasich (1986) have pro-
distributions. (A constant was added to the
vided explicit definitions by which individual
magnitude of habituation variables to avoid
patterns may be categorized, and we followed
taking the square root of negative values.) Re-
these definitions in categorizing infants' pat-
siduals analyses were employed for each
terns for this study. Once familiar with the
bivariate distribution to identify outliers
categorization system, two judges were able
whose data would disproportionately affect
to classify the looking patterns with 100% ac- individual correlations. Infants were excluded
curacy.
from correlation analyses if their standard-
ized residuals within the bivariate distribu-
VARIABLES OF INTEREST: DISHABITUATION
INDICES tion were 4.0 or greater (Cohen & Cohen,
1983). This is a conservative practice; two in-
The infant's response to novelty has been fants were excluded from one of the longitu-
taken as an index of cognition within this and dinal correlations, but no more than one in-

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
 Colombo et al. 479

fant was excluded from any of the other standard vs. first fixation to novel) ANOVAS
correlations reported. further indicated that dishabituation to the
Pearson r's were calculated for the con-
novel face in the Lagged Change condition
did not significantly differ from that in Im-
tinuous habituation variables across testing
mediate condition (all Familiar/Novel Stimu-
sessions to gauge the continuum-based short-
lus effects reached p < .01, with no significant
and long-term reliabilities of the assessments.
two-way interactions), thus allowing us to
The test-retest reliability of habituation pat-
pool the conditions for assessment of stability.
tern was evaluated with kappa, a reliability
coefficient for categorical data that adjusts for We tested the reliability of infants' cat-
chance agreement (Cohen, 1960). egorization as "dishabituators" or "nondis-
habituators" across repeated testings with
DISHABITUATION kappa. For magnitude of habituation, Pearson
r's were calculated after we performed resid-
The inclusion of a Lagged Change group
uals analyses, as done for habituation indices,
at each age in the experimental design al- and arcsin transformed the dishabituation
lowed group comparisons to evaluate in
proportion to avoid rectangular distributions
which of the four age groups dishabituation
often generated by proportion scores.
could actually be attributed to discrimination
of the novel from the familiar stimulus as op-
posed to spontaneous dishabituation due to Results and Discussion
regression. For the cross-sectional sample, SHORT-TERM RELIABILITY:
these analyses were mixed design Condition CROSS-SECTIONAL SAMPLE RESULTS
(2) x Testing (2) x Pre/Post (2: last criterion
fixation vs. first posthabituation fixation) Habituation Indices
ANOVAs performed separately for each age Zero-order correlations for the continu-
group. (For the longitudinal sample, the Test- ous variables and kappas for the pattern of
ing factor was excluded.) In these analyses, habituation variable are presented in Table 3.
the Condition x Pre/Post interaction indi- Overall, duration-based variables (peak fixa-
cated discrimination, as we would expect that tion, duration of first fixation, average fix-
only infants in the Immediate Change condi- ation duration) and magnitude of habitua-
tion would show dishabituation since only tion variables proved to be the most reliable
they saw a novel stimulus on the first post- variables. Duration of first fixation was the
habituation look. only variable to show statistically significant
reliability at all ages; the test-retest reliability
Across both samples, only the 3-month-
of peak and mean fixation was significant at
olds failed to discriminate stimulus changes,1
all ages except 3 months. Reliabilities for total
although this was not because the 3-month- fixation and total interlook interval were sta-
olds failed to dishabituate. Rather, both 3-
tistically significant at 4 and 9 months and
month-old dishabituation analyses indicated
when all ages were considered together.
that significant posthabituation increases
occurred for both Immediate Change and The reliability of the infants' magnitude
Lagged Change conditions. This rules out the of habituation was to some degree dependent
possibility that 3-month-olds' habituation was both on age and on which fixation was used as
attributable to state variables or adaptation. the initial curve point; decrement from the
We could not infer discrimination in these first two fixations was reliable at 3, 4, and 7
young infants, however, because sufficient months and across all ages. Decrement from
"spontaneous" recovery occurred in the the peak fixation was reliable in all except
Lagged Change condition to prevent sig- the 7-month-olds and very reliable when
nificant Pre/Post x Condition interactions in test-retest reliability was calculated across
the analyses. Subsequent Condition (2) x all ages. None of the nonduration variables
Familiar/Novel Stimulus (2: last fixation to (number of fixations to habituation, location of

SThe Condition x Pre/Post interaction terms for these analyses in the cross-sectional sample
are F(1,29) = 0.93, p = N.S., for the 3-month-olds; F(1,26) = 12.52, p = .001, for the 4-month-olds;
F(1,47) = 8.12, p = .006, for the 7-month-olds; and F(1,30) = 6.38, p = .017, for the 9-month-olds.
For the longitudinal sample, the interaction terms are, for 3-month-olds, F(1,48) = 0.94, p = N.S.;
for 4-month-olds, F(1,53) = 3.43, p = .05; for 7-month-olds, F(1,48) = 4.69, p < .05; and for 9-month-
olds, F(1,40) = 21.55, p < .001. In all cases in which the interaction was significant, the pattern of
dishabituation was as expected, with larger Pre/Post fixation increases in the Immediate Change
condition than in the Lagged Change condition.

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
480 Child Development
TABLE 3

RELIABILITY OF HABITUATION VARIABLES FROM THE CROSS-SECTIONAL SAMPLE

3 Months 4 Months 7 Months 9 Months All Ages

(N = 34) (N = 34) (N = 50) (N = 34) (N = 152)a
Duration variables:
Duration of peak fixation .... .22 .49** .41** .35* .47***
Duration of first fixation .... .37* .58*** .36** .39* .46***
Average fixation fixation .... .29 .46** .42** .47** .47***
Total fixation time
to habituation ............ .26 .41* .09 .54*** .36***
Total interlook interval ..... .17 .36* .23 .32* .23*
Habituation magnitude:
From looks 1-2 ............ .33* .32* .33* .08 .38***
From peak look ............ .32* .65*** .19 .41* .48***
Nonduration variables:
Number of fixations
to habituation ............ -.07 .08 .25 -.05 .17
Location of peak fixation .... -.03 -.04 .09 .18 .09
Habituation pattern ......... -.14 -.36 -.37 .17 -.13
Dishabituation variables:
Presence of
dishabituation ............ .05 .17 .00 - .36 - .03
Magnitude of
dishabituation ............ .16 .03 -.18 - .22 -.03

a This column contains r's run on all infants, regardless of age.

* p < .05.
** p < .01.
*** p < .001.

peak look, and pattern of habituation) showed bauer, & Harmon, 1976; Emde & Harmon,
any reasonable reliability at any age studied. 1983). It is also possible that the lower re-
liabilities may be more specifically related to
Over all ages, then, most habituation
changes in infants' response to social stimuli
measures possessed reasonable short-term re-
(i.e., faces) at these ages.
liability. As Table 3 suggests, however, there
was some age-related variability in the test- Dishabituation
retest reliability of habituation variables. In- Kappas for the categorical measure of dis-
fants' habituation indices appeared to be habituation and Pearson r's calculated for the
more reliable at 4 and 9 months than at 3 and
dishabituation proportion are also presented
7 months. There are, for example, seven sta- in Table 3. None of these correlations was
tistically significant test-retest correlations at statistically significant, nor were these results
4 months, and six significant correlations at affected by calculating correlations separately
9 months, but only three at 3 months and four for the Immediate Change and Lagged
at 7 months. Furthermore, the mean of the Change conditions. We were not able to doc-
test-retest correlations for the nine continu-
ument any reliability for either the presence
ous variables is .36 at 4 months and .30
or the magnitude of dishabituation (or "recov-
at 9 months but .20 at 3 months and .26
ery") of attention at any age across the short
at 7 months. These lower reliabilities at 3 and term.
7 months are not attributable to a truncation
of variability of these infants' visual behavior; Controlling for the Influence of the Peak
the variability of 7-month-olds is equal to that Fixation
of the 9-month-olds, and 3-month-olds were The zero-order correlations suggested
the most variable of all age groups. These age that the most reliable variables from infants'
differences in reliability may be interpreted visual habituation sessions across the first
as reflecting periods of rapid developmental 9 months were duration and magnitude of
change or reorganization, as the attenuated re- habituation variables. These variables are not
liability at 3 and 7 months may be related to completely independent of one another, how-
the developmental transitions that have been ever, as the peak fixation is common to the
documented at approximately these ages by calculation of all of them. Since the peak
Emde and his colleagues (Emde, Gaens- fixation per se was comparatively reliable, it

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
 Colombo et al. 481

seemed possible that the reliability of some or correlations for both magnitude of habituation
all of the other variables to which it contrib- variables. This was first done by partialing out
utes might be attributed to the peak fixation the influence of the peak fixation from the re-
alone. In order to test this possibility, we per- liability of each of the components of the dec-
formed a number of secondary analyses. rement equations not directly involving the
We first examined the contribution of the
peak fixation in the computation (i.e., the
mean of the two criterion looks and the mean
peak fixation to total fixation time and average
fixation duration variables. For total fixation
of first two looks). None of the partial correla-
tions on these variables for individual age
time, we subtracted the peak fixation from the
groups was statistically significant, and the
total and calculated the reliability of this non-
partial correlations across all ages combined
peak fixation total, while for average fixation were .15 for habituation criterion and .12 for
duration we calculated the infants' mean fix-
the mean duration of the first two fixations.
ation excluding the peak look. The reliabili-
ties of these two variables across the cross- These results strongly suggest that the reli-
ability of those components was determined
sectional sample are presented in Table 4.
primarily by the reliability of the peak fixation
The reliabilities of these nonpeak duration
alone. This finding was further confirmed by
variables show little or no reliability at 3
calculating the reliabilities of each of the mag-
months but moderate reliability from 4
nitude of habituation variables per se while
months on. Although the reliability of non-
directly partialing out the influence of the
peak total looking time at 7 months of age did
peak fixations. When this is done, the test-
not approach significance, mean nonpeak fixa-
retest reliability of these variables no longer
tions at 7 months were moderately reliable.
reaches statistical significance. None of the
We next examined the contribution of the partial correlations for separate age groups ex-
peak fixation to the reliability of the duration ceeds .13, and the respective test-retest corre-
of infants' first fixations. Since the first fixation lations for decrement from the peak fixation
is in fact the peak fixation in 40%-60% of the and decrement from the mean of the first two
cases with young infants (e.g., Bornstein & looks are .11 and .16 when the peak fixations
Benasich, 1986; McCall, 1979), it was possi- are partialed out.
ble that the reliability of this variable was due
In summary, then, while duration mea-
directly to the peak fixation. We therefore ex-
sures appear to possess reasonable short-term
amined the reliability of first fixations in cases
test-retest reliability regardless of whether
in which neither fixation was the peak fixa-
they include the peak fixation, the significant
tion. Since the location of the peak fixation in
reliability observed for magnitude of habitua-
the habituation sequence was not reliable, the tion variables does seem attributable to the
number of infants showing nonpeak first
contribution of the peak fixation.
fixations in each of the two testings is consid-
erably lower than the overall sample size.
However, these nonpeak first fixations were LONG-TERM STABILITY: LONGITUDINAL
relatively reliable: Pearson r's were .63 at 3 SAMPLE RESULTS
months (N = 10), .52 at 4 months (N = 13), Habituation Indices
.59 at 7 months (N = 22), .43 at 9 months (N =
The results of the longitudinal study indi-
18), and .57 (N = 63) across all ages. All corre- cated that infants' visual behavior is not as
lations reached at least the p < .05 level of reliable over several months as it is from
significance.
week to week. Pearson product-moment cor-
Finally, we looked at the contribution of relation coefficients for continuous habitua-
the peak fixations to the observed test-retest tion variables and kappas for habituation pat-

TABLE 4

TEST-RETEST RELIABILITY OF NONPEAK FIXATION VARIABLES:

CROSS-SECTIONAL SAMPLE

Variable 3 Months 4 Months 7 Months 9 Months All Ages

Total nonpeak looking

time to habituation ........... .16 .31** .04 .43** .22**
Average nonpeak
fixation duration ............. .23 .38** .33** .49** .36**

** p < .01.

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
482 Child Development

tern are reported in Table 5 for all infants short-term assessments in the cross-sectional
on whom complete habituation sessions are study.
available at the two comparison ages. (As Dishabituation
noted in Table 2, not all infants completed all
Correlations for presence and magnitude
four testings, so sample sizes vary within the
of dishabituation across ages in the longitudi-
matrices.)
nal sample are also presented in Table 5. For
Of all the habituation variables tested, magnitude of dishabituation, only the 4-9-
only duration of peak fixation and magnitude month assessment was observed to be stable;
of habituation calculated from peak yielded this is also reflected in the categorical dis-
significant and generally systematic relations habituation measure, although the relation for
across ages. Peak fixation was the most stable this latter measure does not reach statistical
across most of the first year (only the 4-9- significance. This isolated relation is difficult
month correlation does not reach signifi- to interpret because it is the only significant
cance). The only other variables to yield any longitudinal correlation for this variable, be-
cross-age stability across the first year were cause it occurs over a long (5-month) interval
average fixation duration and duration of first in the absence of significant correlations over
look, although the pattern of those relations shorter intervals, and because dishabituation
does not suggest a simple developmental in- did not exhibit within-age reliability at these
terpretation. two ages. It may be that this is not a spurious
relation, however, given that the 4- and 9-
Our 4-7-month correlations differ some- month habituation assessments were the most
what from those reported by Byrne et al. internally reliable in the cross-sectional sam-
(1985), who report 4-7-month correlations ple and that Byrne et al. (1985) found a 4-7-
ranging from .37 to .53 for various duration month dishabituation relation similar in mag-
measures (stability for peak look was not re- nitude.
ported). This magnitude of relation is ap-
Controlling for the Influence of the Peak
proximated in our data only by the peak look Fixation
variable. This discrepancy may be due to
As for the cross-sectional study, we tried
stimulus or procedural differences; Byrne
to determine the contribution of the peak
et al. (1985) employed a constant multiple-
fixation to the longitudinal stability of the du-
stimulus sequence during habituation, while
ration and magnitude of habituation variables
we used one stimulus per habituation se-
in a set of secondary analyses. We first sub-
quence.
tracted the peak fixation from total and mean
As with the cross-sectional study, none of looking time equations. The longitudinal sta-
the nonduration variables (number of looks to bility of total and mean nonpeak looking time
habituation, location of peak look, habituation variables are presented in Table 6. While the
pattern) showed any stability from one age to duration of infants' peak fixation is relatively
another. We did examine the reliability of in- stable across the first 9 months, stability in
fants' habituation pattern across more than infants' nonpeak looking was not observed
two testings, however, and some reliability until 7 months, as only the 7-9-month correla-
did emerge. Of the 35 infants who completed tion is significant (r = .39, p < .05). The corre-
each of the four longitudinal habituation ses- lation for mean nonpeak fixations is statisti-
sions, 23 (67%) were classified as having the cally significant between the 3- and 9-month
same habituation pattern on at least three of testings as well as between the 7- and 9-
the four occasions. Twenty of the 23 infants month testings.
showed this consistency on the fluctuating What little stability was observed for du-
pattern, while the other three showed consis- ration of first fixation across the first year
tency on the exponential decrease pattern. (Table 5) appears attributable to the peak look
Even after we adjusted for the frequency of alone. The longitudinal stability of first fixa-
the occurrence of these two patterns in the tions in cases in which neither of the first
distribution to correct for possible chance looks was a peak fixation was found to be very
findings, this finding was statistically signifi- low. None of the age-to-age correlations ex-
cant (p = .02, binomial test). It may be that ceeds .12, and none approaches statistical
two assessments are not adequate to demon- significance.
strate sufficient reliability for this variable,
but over the course of a number of repeated We also computed a series of longitudi-
assessments, consistency can be observed. nal correlations in which the influence of the
This would also explain the lack of reliability peak fixations was partialed from the criterion
in habituation pattern observed across two components of the habituation magnitude

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
 TABLE 5

STABILITY OF VISUAL ATTENTION INDICES: LONGITUDINAL SAMPLE

4 Months 7 Months 9 Months

Duration measures:
Duration of peak fixation:
3 months ..................................... .35* .30* .37*
4 months ........................................... .32* .20
7 months ........................................... .49*
Duration of first fixation:
3 months ...................................... .36* -.07 .04
4 months ....................... ...... ............ .28* .36*
7 m onth s ........................................... - .17
Mean fixation duration:
3 months ..................................... .34* .18 .48*
4 months ....................................... ..13 -.08
7 m onths ........................................... .38*
Total fixation through habituation:
3 months ...................................... .21 .09 .37*
4 months ........................................ .08 .07
7 m onths ................................. ..........26
Total interlook interval:
3 months .................................... .21 .03 .28
4 months .............. ......................... .11 -.13
7 months ........................................... .32*
Magnitude of habituation variables:
Looks 1-2 to criterion:
3 months ...................................... .25 .15 .27
4 months ...................... .................. -.14 .02
7 months ........................................... -.03
Peak look to criterion:
3 months ...................................... .50** .43** .53**
4 months ........................................ .36* .23
7 months ........................................ .36*
Nonduration variables:
Number of fixations to habituation:
3 months ......................................... .05 .12 .16
4 months ........................................... -.19 .05
7 months ........................................... .24
Location of peak fixation:
3 months ...................................... .11 .08 .28
4 months ........................................ .20 .00
7 months .......................................... .27
Habituation pattern:
3 months ...................................... .18 .01 -.10
4 months ........................... ............. -.11 .17
7 months ..................... 6 .......................19
Dishabituation variables:
Categorical dishabituation:
3 months ..................................... -.23 -.06 -.18
4 months ........................................... -.07 .23
7 months ........................................... -.06
Magnitude of dishabituation:
3 months ..................................... -.19 -.01 -.19
4 months ........................................ -.15 .37**
7 months ........................... ............. -.15

NOTE.-Table includes all infants who completed habituation sessions at each age reported. N's for correlations
range from 50 (3-4 months) to 36 (3-9 months). The statistics reported for pattern of habituation are kappas.
*p< .05.
** p < .01.

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
484 Child Development
TABLE 6

STABILITY OF NONPEAK DURATION VARIABLES:

LONGITUDINAL SAMPLE

4 Months 7 Months 9 Months

Total nonpeak fixation time:

3 months ....... ................................... ..08 .11 .27
4 m onths ............................................ .11 .10
7 m onths ........................................... .39*
Nonpeak mean fixation duration:
3 months ....... ................................... . .23 .17 .38*
4 m onths .............. ........................... .08 .00
7 m onths .............................................38*
* < .05.

equations, as in the cross-sectional sample. sures of infant visual behavior, as assessed

Across the earlier months these correlations within a standard infant-control habituation
were not statistically significant, suggesting paradigm, are attributable to some character-
that the significant stability seen in Table 5 istic of the individual infant, as opposed to
for magnitude of habituation from peak fix- situational influences. Overall, infants' visual
ation from 3 and 4 months of age to 7 and 9 habituation was fairly reliable over the short
months of age was attributable to the stability term and less reliable over the longer term.
of the peak fixation alone. This finding was While the comparison across the two samples
further confirmed when we partialed the peak fits the expected pattern of an inverse relation
fixation directly from the habituation mag- between test-retest reliability and test-retest
nitude variables; none of the resulting partial interval, the pattern of results within the lon-
correlations from 3 to 7 months remains sig- gitudinal sample itself did not. There were
nificant when this is done. However, the du- nearly as many significant correlations be-
ration of infants' criterion looks was stable tween the most distant testings (i.e., between
from 7 to 9 months of age even when the peak 3 and 9 months we observed three significant
fixations were partialed out (r = .37, p < .05), longitudinal relations) as between the most
suggesting some stability for magnitude of proximate testings (four significant correla-
habituation in and of itself across those ages. tions between 3 and 4 months and four be-
This finding was further supported by correla- tween 7 and 9 months). Such a pattern of rela-
tions that directly partialed out the influence tions has also been reported elsewhere (e.g.,
of the peak fixations from the longitudinal sta- Miller et al., 1979) and seems to support
bility of the peak look to criterion decrement claims for nonlinear development elsewhere
variable. Even when the peak fixation is par- in the infancy literature (e.g., Emde et al.,
tialed out, a significant relation remained for 1976).
this variable between 7 and 9 months (r =
.45, p < .01). Because this habituation mag- Furthermore, some measures of infants'
nitude variable was not reliable within the 7- visual habituation proved to be more reliable
and 9-month cross-sectional groups when the than others. A consistent finding across both
effect of the peak fixation was removed, this is the samples tested was that fixation duration
a curious finding that warrants replication. and magnitude of habituation variables were
more reliable than number of fixations, loca-
In sum, when the influence of peak look
tion of peak look, or habituation pattern vari-
was removed from most duration and mag- ables. The most reliable and stable of these
nitude of habituation measures, stability still
remained between the 7- and the 9-month
measures was the infants' longest, or "peak,"
look during a testing session, which showed
testings for total nonpeak fixation time, mean
good week-to-week reliability and consistent
nonpeak fixation, and magnitude of habitua-
tion from the first two fixations. These find-
stability across longitudinal tests spanning the
period from 3 to 9 months. Other nonpeak
ings suggest some stabilization in more as- duration measures were reliable from week to
pects of infants' visual habituation in the
week starting at 4 months, but month-to-
second half of the first year of life.
month stability for total nonpeak and mean
General Discussion
nonpeak fixation duration measures together
did not appear until between 7 and 9 months
The aim of this project was to determine of age. Secondary analyses in each sample
the degree to which commonly used mea- also consistently suggested that much of the

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
 Colombo et al. 485

short-term reliability and much of the stability liabilities between two single items on a
observed for magnitude of habituation vari- psychometric instrument administered under
ables at early ages could be confidently attrib- different protocols (i.e., different visual stim-
uted to the correlated reliability of the peak uli). The findings of moderate reliability in
fixation. Only between 7 and 9 months of age this light strongly suggest some internally
did a magnitude of habituation variable dem- consistent process. However, the amount of
onstrate any significant stability independent variance accounted for by these estimates of
of the stability of the peak look. reliability never exceeded 28%. It would
therefore appear that, if visual attention as-
Our findings suggest that fixation dura- sessments are to be used in studies of devel-
tion variables, which are not central to tradi-
opmental prediction, multiple assessments
tional visual habituation models, might
are clearly preferable.
provide better access to the processes that de-
termine infant visual behavior, whether those Under the procedural and stimulus con-
processes are cognitive or motivational. Fur- ditions of these studies, we were unable to
thermore, our finding that fixation duration observe any consistent reliability or stability
variables carry much of the weight of the reli- of the infant's visual dishabituation to a novel
ability of infant visual behavior is consistent presentation.2 Dishabituation was not found
with other recent findings that link fixation- to be reliable at any age in the cross-sectional
duration measures with later intellectual or sample, and a positive relation was observed
cognitive function. For example, Sigman et al. on only one of six comparisons in the longitu-
(1985) reported that prolonged visual fixation dinal sample. As with any null result, these
at 4 months of age predicted poorer IQ scores findings must be interpreted with caution;
at 8 years in a large preterm sample. Simi- this is especially true of data from the 3-
larly, Slater (1985) reported high negative cor- month-olds, whose group analyses did not
relations (r's from -.45 to -.73) between show significant dishabituation beyond that
fixation duration measures taken during in- seen in infants not immediately receiving a
fancy and later performance on preschool in- novel stimulus following dishabituation. It is
telligence tests. If we accept duration of important to note that the lack of reliability
fixation as an index of cognitive processing, it observed here does not discount the validity
appears reasonable to hypothesize that infants of using dishabituation as a group measure of
with shorter peak fixations may process visual discrimination or of discrimination in a
stimuli faster or more efficiently than infants given individual at a given point in time, as
with longer peak fixations. suggested previously by McCall (1979). It
Nonduration-based habituation variables does suggest, however, that dishabituation
across the number of stimuli employed here
typically did not yield statistically significant
may not be a reliable measure of individual
short-term or long-term reliability in these
infants' visual discrimination abilities. Our
studies. Other authors have argued that mea-
sures such as number of fixations to habitua- findings of low stability for recovery within
infancy contrast with recent reports of a posi-
tion may not be a valid individual difference
measure (Dannemiller, 1984). While Born- tive relation between the response to novelty
in the paired-comparison paradigm and child-
stein and Benasich (1986) reported stable
hood IQ (Fagan, 1985; Fagan & Singer, 1983;
habituation patterns across two testings at 5
Rose & Wallace, 1985). Perhaps the test phase
months, we found reliability in habituation
of the paired-comparison paradigm is more
patterns only when more than two testings
were considered. sensitive than the single-presentation test
trial commonly used in the standard habitua-
The moderate stability of certain mea. tion-dishabituation procedure. In forcing the
sures of visual attention across various points infant to choose between the familiar and the
in the first year is encouraging, especially if novel stimuli, the former procedure may give
one considers the reported correlations as re- the infant a cue for retrieval of the previously

2 We have presented only two of many ways to calculate dishabituation. The most common
continuous measure is a pre/post difference score, which is not appropriate for correlational methods
(see Appelbaum & McCall, 1983). We have, however, calculated and examined reliabilities for this
difference score, and they are no better than those reported here for the dishabituation proportion.
Similarly, we have chosen one of a number of arbitrary cutoffs for categorizing dishabituation: an
increase in fixation to the novel stimulus that exceeds that of the control condition. Other possible
cutoffs may be more liberal (e.g., any fixation increase above the criterion fixation) or more stringent
(e.g., a doubling of fixation above the criterion fixation, which can be seen as "reversing" the 50%
decrement criterion). We have also performed these analyses, with again no improvement in the
reliability of dishabituation.

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
486 Child Development

shown stimulus. The paired-comparison pro- Dannemiller, J. (1984). Infant habituation criteria:
cedure also likely allows better methodologi- 1. A Monte Carlo study of the 50% decrement
cal control for spontaneous dishabituation that criterion. Infant Behavior and Development, 7,
can inflate individual estimates of discrimina- 147-166.
tion. Detailed task-demand analyses of the Dannemiller, J., & Banks, M. (1983). Can selective
differences between the two test-phase tech- adaptation account for early visual habituation?
niques should help to describe more pre- Merrill-Palmer Quarterly, 29, 151-158.
cisely the processes involved in different as- DeLoache, J. (1976). Rate of habituation and visual
pects of infant visual memory and should also memory in infants. Child Development, 47,
help to isolate what may be early precursors 145-154.
to childhood intellectual performance. Emde, R., Gaensbauer, T., & Harmon, J. (1976).
Emotional expression in infancy: A biobe-
havioral study. Psychological Issues: Mono-
References
graph Series, 10.
Appelbaum, M., & McCall, R. (1983). Design and Emde, R., & Harmon, J. (1983). Continuities and
analysis in developmental psychology. In M. M. discontinuities in development. New York:
Haith & J. J. Campos (Eds.), P. H. Mussen Plenum.
(Series Ed.), Handbook of child psychology: Fagan, J. (1974). Infant recognition memory: The
Vol. 2. Infancy and developmental psychobiol- effects of length of familiarization and type of
ogy (pp. 415-476). New York: Wiley. discrimination task. Child Development, 45,
Bashinski, H., Werner, J., & Rudy, J. (1985). Deter- 351-356.
minants of infant visual fixation: Evidence for a Fagan, J. (1985, May). Early novelty preferences
two-process theory. Journal of Experimental and later intelligence. Paper presented at the
Psychology, 39, 580-598. meeting of the Society for Research in Child
Berlyne, D. (1958). The influence of albedo and Development, Toronto.
complexity on visual fixation in the human in- Fagan, J., & Singer, L. (1983). Infant recognition
fant. British Journal of Psychology, 56, 315- memory as a measure of intelligence. In
318. C. Rovee-Collier (Ed.), Advances in infancy
Bertenthal, B., Haith, M., & Campos, J. (1983). The research (Vol. 2, pp. 31-78). Norwood, NJ:
partial-lag design: A method for controlling Ablex.
spontaneous regression in the infant-control Fantz, R. (1958). Pattern vision in young infants.
paradigm. Infant Behavior and Development, Psychological Review, 8, 43-47.
6, 331-338. Fantz, R. (1964). Visual experience in infants: De-
Bornstein, M. (1985). Habituation as a measure of creased attention to familiar patterns relative to
visual information processing in human in- novel ones. Science, 146, 668-670.
fants: Summary, systematization, and synthe- Horowitz, F. D., Paden, L., Bhana, K., & Self, P.
sis. In G. Gottleib & N. Krasnegor (Eds.), De- (1972). An infant control procedure for study-
velopment of audition and vision during the ing infant visual fixations. Developmental Psy-
first year of postnatal life: A methodological chology, 7, 90.
overview (pp. 253-295). Norwood, NJ: Ablex. McCall, R. (1979). Individual differences in the pat-
Bornstein, M., & Benasich, A. (1986). Infant habitu- tern of habituation at 5 and 10 months of age.
ation: Assessments of individual differences Developmental Psychology, 15, 559-569.
and short-term reliability at 5 months. Child McCall, R., & Kagan, J. (1970). Individual differ-
Development, 57, 87-99. ences in the infant's distribution of attention to
Bornstein, M., & Sigman, M. (1986). Continuity in stimulus discrepancy. Developmental Psychol-
mental development from infancy. Child De- ogy, 2, 90-98.
velopment, 57, 251-274. McCall, R., Kennedy, C., & Dodds, C. (1977). The
Byme, J., Clark-Tousenard, M., Hondas, B., & interfering effect of distracting stimuli on the
Smith, I. (1985, May). Stability of individual infant's memory. Child Development, 48, 79-
differences in infant visual attention. Paper 87.
presented at the meeting of the Society for Miller, D., Ryan, E., Aberger, E., McGuire, M.,
Research in Child Development, Toronto. Short, E., & Kenny, D. (1979). Relationships
Cohen, J. (1960). A coefficient for agreement for between assessments of habituation and cogni-
nominal scales. Educational and Psychological tive performance in the early years of life. In-
Measurement, 20, 37-46. ternational Journal of Behavioral Develop-
Cohen, J., & Cohen, P. (1983). Applied multiple ment, 2, 159-170.
regression/correlation analysis for the behav- Miller, D., Ryan, E., Short, E., Ries, P., McGuire,
ioral sciences. Hillsdale, NJ: Erlbaum. M., & Culler, P. (1977). Relationships between
Colombo, J., & Horowitz, F. D. (1985). A para- early habituation and later cognitive perform-
metric study of the infant control procedure. ance in infancy. Child Development, 48, 658-
Infant Behavior and Development, 8, 117-121. 661.

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms
 Colombo et al. 487

Miller, D., Sinnott, J., Short, E., & Hains, A. (1976). memory: A predictor of later cognitive func-
Individual differences in habituation rates and tioning in preterm infants. Child Development,
object concept performance. Child Develop- 56, 843-852.
ment, 47, 528-531. Rovee-Collier, C., & Gekoski, M. (1979). The eco-
Miller, D., Spiridigliozzi, G., Ryan, E., Callan, M., nomics of infancy: A review of conjugate rein-
& McLaughlin, J. (1980). Habituation and forcement. In H. Reese & L. Lipsitt (Eds.), Ad-
cognitive performance: Relationships between vances in child development and behavior
measures at four years of age and earlier assess- (Vol. 13, pp. 195-256). New York: Academic
ments. International Journal of Behavioral De- Press.

velopment, 3, 131-146. Ruddy, M., & Bornstein, M. (1982). Cognitive cor-

Mitchell, D. W., & Steiner, L. (1984, April). In- relates of infant attention and maternal stimula-
dividual differences in habituation perfor- tion over the first year of life. Child Develop-
mance: Implications for recovery behavior. ment, 53, 183-188.
Paper presented at the International Confer- Sigman, M., Cohen, S., Beckwith, L., & Parmelee,
ence on Infant Studies, New York. A. (1985, July). Infant attention in relation to
O'Connor, M., Parmelee, A., & Cohen, S. (1984). intellectual abilities in childhood. Paper pre-
Infant auditory discrimination in preterm and sented at the meeting of the International Soci-
full-term infants as a predictor of 5-year intelli- ety for the Study of Behavioral Development,
gence. Developmental Psychology, 20, 159-165. Tours.

Ritz, E., Woodruff, A. B., & Fagen, J. (1984). Short- Slater, A. (1985, July). The relationship between
and long-term stability of habituation rate and infant attention and learning, and linguistic
looking time in four-month-old infants. Journal and cognitive abilities at 18 months and 4.5
of Genetic Psychology, 144, 285-286. years. Paper presented at the meeting of the
Rose, S. (1981). Developmental changes in infants' International Society for the Study of Behav-
retention of visual stimuli. Child Development, ioral Development, Tours.
52, 227-233. Sokolov, E. (1963). Perception and the conditioned
Rose, S., & Wallace, I. (1985). Visual recognition reflex. New York: Macmillan.

This content downloaded from 128.163.2.206 on Mon, 27 Jun 2016 05:35:41 UTC
All use subject to http://about.jstor.org/terms