BEHAVIOURAL

BRAIN
RESEARCH
ELSEVIER Behavioural Brain Research 67 (1995) 99-101

Short communication

Development of chromatic discrimination in early infancy

Russell J. Adams a7b**,Mary L. Courage a
“Department of Psychology, Memorial University of Newfoundland, St. John’s, NF, Canada AIB 3X9
bDepartment of Neonatology. St. Glare’s Mercy Hospital, St. John’s, NF, Canada Al C 5B8

Received 6 January 1994; revised 25 August 1994; accepted 25 August 1994

Abstract

Combining a habituation-recovery paradigm with procedures to minimize brightness cues, we evaluated 1-3-month-old? ability to
discriminate among broadband mid- and long-wavelength stimuli. One-month olds showed evidence of discriminating a red (domi-
nant 1= 650 nm) from green (dominant L = 545 nm) but failed to discriminate either of these from a yellow (dominant 1= 585 nm).
At 2 months, infants showed evidence of discriminating 545-nm green from 585~nm yellow, but only at 3 months do they appear to
discriminate 650-nm red from the yellow. Consistent with previous work, these data imply that chromatic discrimination emerges slowly
over the first three postnatal months. Moreover, the developmental pattern is consistent with the notion that early human vision is
limited by general rather than selective immaturities within the requisite retinal and neural mechanisms.

Key words: Color vision; Infant; Chromatic discrimination; Visual development

-

Data from experiments on human infant color vision [ l-41 only when the stimulus size of each is at least 16”. In the
have played a critical role in formulating models of the present experiments we expand on this previous work by
early anatomical and functional development of important using the same methods to examine the development of
visual system mechanisms, particularly those in the retina chromatic discrimination in l-, 2- and 3-month-old human
[ 5,6]. Yet there is still debate over whether early vision is infants.
limited by immaturities in specific visual system mecha- In the first set of experiments, three groups of 20 l-month
nisms (e.g., a particular receptor type or a post-retinal olds (mean age = 4.8 weeks) were tested with pairs of
mechanism) or by more general quantitative immaturities chromatic stimuli that newborns fail to discriminate,
across all visual mechanisms. namely 16” red vs. yellow, 16” yellow vs. green and 8” red
To date, empirical studies indicate that at birth, human vs. green. A fourth group of 20 l-month olds (mean
color vision is very poor. Habituation-recovery tests of age = 5.1 weeks) was tested with a pair of stimuli that
chromatic-achromatic discrimination [ 7- 1 l] have re- newborns discriminate from each other (16” red vs. green).
vealed that even with large stimulus sizes (16-32”), new- The stimuli, apparatus and procedure were described in
borns fail to discriminate from white, several short- detail in a previous paper [ 121 and thus only the most
wavelength (450-nm, 470-nm and 480-nm blue; 493-nm pertinant details will be given here. The stimuli were cre-
blue-green), mid-wavelength (565nm and 572-nm yellow- ated by passing broadband chromatic light produced by a
green) and extraspectral (purple) chromatic stimuli. More- Beseler Dichro MX II color computer through a simple
over, even with those stimuli that newborns have shown optical system and onto a vertical rear-projection screen
evidence of discriminating from white (540-nm green, located 38 cm from the infant’s eyes. A variable aperture
585-nm yellow, 595-nm orange and 650-nm red), the size within the optical system controlled the luminance of the
of the chromatic stimulus must be at least 8-16”. Fur- stimuli. Measurements of the chromatic characteristics of
thermore, tests of chromatic-chromatic discrimination the stimuli taken at the screen revealed that the C.I.E. x-
[ 121 reveal that newborns discriminate only those stimuli and y-coordinates were 0.28 and 0.69 for the green square,
with the greatest spectral separation (red vs. green) and 0.50 and 0.46 for the yellow square and 0.69 and 0.29 for
the red square. From these values, dominant wavelengths
* Corresponding author at address a. Fax: (1) (709) 737-2430. were estimated to be 545 nm, 585 nm and 650 nm for the

0166-4328/95/$9.50 0 1995 Elsevier Science B.V. All rights reserved

SSDI 0166-4328(94)00143-X
100 R.J. Adams, M.L. Courage / Behavioural Brain Research 67 (1995) 99-101

green, yellow and red squares, respectively. Excitation pu- among any of the three measures. These data imply that
rity was approximately 0.80 for all stimuli. except for the ability to discriminate red from green with
To minimize the possibility that infants make chromatic 8 ’ stimuli, there is little improvement in infants’ chromatic
discriminations on the basis of brightness cues, our design discrimination from birth to 1 month, at least within this
attempts to make brightness cues irrelevant by first famil- spectral region (the Rayleigh region). [ 121
iarizing or habituating infants to a wide range of continu- In a subsequent experiment, we used the same methods
ally varying or ‘jittering’ luminances. As we have argued and analyses to test two groups of 20 2-month olds (mean
at length on many previous occasions [8-121 and have age = 9.2 weeks) with the 16” stimuli that l-month olds
verified with control experiments [ 91, this paradigm effec- failed to discriminate, (i.e., 16” green vs. yellow and 16”
tively reduces the possibility that infants use achromatic red vs. yellow). To be consistent with our previous study,
cues to discriminate among chromatic stimuli. we also tested a third group of 2-month olds with the
Testing proceeds in two phases. In the habituation (fa- stimuli (16 ’ red vs. green) that both newborns and
miliarization) phase, the baby is shown repeatedly one of l-month olds discriminate. As expected, the analyses re-
the chromatic stimuli (e.g., the 16” green) that, on each vealed that 2-month olds discriminated the red from the
trial, varies in luminance from 0.34 to 1.16 log cd/m2. The green (PC 0.005) and, unlike l-month olds, they also dis-
order of the luminances is random but is counterbalanced criminated the green from the yellow (Pe: 0.05). However,
across infants. Using an infant control procedure [ 121, like younger infants, 2-month olds showed no evidence of
each infant’s looking time is recorded by two experienced discriminating the yellow from the red.
observers. one of whom is prevented from determining the In a final experiment, we retested the 16” yellow/red
chromatic charateristics of the stimuli. The habituation combination with a group of 20 3-month olds (mean
phase continues until the infant’s fixation time decreases age= 13.3 weeks). Unlike %-month olds, 3-month olds
to a preset criterion. At this point, we assume that the showed strong evidence of discriminating this pair of chro-
infant has become ‘bored’ with the color of the habituat- matic stimuli (P<O.OOl). Table 1 presents a summary of
ing stimulus and is also ignoring any trial-to-trial bright- the data from the present experiments as well as those
ness changes. In the subsequent test phase, the baby is from the previous study with neonates [ 121. Table 1 shows
shown two trials with a second chromatic stimulus (e.g., for each group of infants, the mean fixation times to the
the 16” red) of mid-range luminance (0.64 log cd/m2) and familiar and novel chromatic stimuli during the test as well
two more trials with the familiar habituating stimulus (e.g.,
the 16” green) in an ABBA or BAAB order. The lumi- Table 1
nance of the familiar stimulus shown during the test is a Mean fixation times to the novel and familiar chromatic stimuli shown
value that is within the range of habituating luminances during the test phase and to the familiar stimulus during the last two
habituation trials. The asterisks denote those conditions in which infants
but one that was not shown during the habituation phase.
show statistical evidence of looking longer at the novel chromatic test
This value is also counterbalanced across babies. stimulus than at the familiar chromatic stimulus shown either during the
We expect that if the infant can discriminate the novel test trials or during the last two habituation trials. Note that the newborn
data are from a previous report [ 121
chromatic stimulus from the familiar chromatic stimulus
on the basis qfchromatic cues, he/she should look longer at
the novel stimulus during the test trials. Conversely, a baby Group Fixation time
(s)
who cannot discriminate the chromatic stimuli on the basis
of chromatic cues should not look longer at the novel Novel stimulus Familiar stimulus Last 2 habituation
stimulus, even if he/she can discriminate a (perhaps sub-
NB 16” R-G* 16.7* 7.9 8.8
stantial) brightness difference between the novel and fa-
NB8” R-G 17.9 13.3 14.4
miliar chromatic squares. For each baby, we calculate NB 16” Y-G 15.1 10.6 11.9
three response measures: mean fixation time to the novel NB 16” R-Y 15.7 18.4 14.5

test square, mean time to the familiar test square and as 1M 16” R-G 47.8** 23.7 24.4
a reliability check, mean time during the last two trials of 1M 8” R-G 48.6** 24.3 25.4
1M 16’ Y-G 24.7 24.1 22.8
the habituation phase. One-way ANOVA and post-hoc
1M 16” R-Y 30.7 25.3 21.7
Neuman-Keuls analyses revealed that l-month olds ha-
2M 16” R-G 35.9** 21.3 22.2
bituated to 16 O or 8 ’ green squares, looked longer at novel 2M 16” Y-G 28.9* 20.1 21.0
16” or 8” red squares than at the familiar green squares 2M 16” R-Y 28.9 26.4 26.1
shown either during the test or during the last two habitu- 3M 16” R-Y 39.9*** 17.8 16.2
ation trials (PC 0.005). Conversely, those groups habitu-
ated to 16’ yellow squares and tested with a novel 16” red * PiO.05 **p <0.005; *** P<O.OOl.
or with a novel 16” green square, showed no differences * Dominant wavelenth: G = 545 nm; Y= 585 nm; R = 650 nm.
 R.J. Adams, M.L. Courage / Be.havioural Brain Research 67 (1995) 99-101
as the mean fixation time to the familiar chromatic stimu-
lus during the last two trials of habituation. It is apparent
from Table 1 that for all groups there was little difference
in fixation time to the familiar stimulus during the test
compared to that during the last two trials of the habitu-
ation phase (P>O.O5 for all groups). Note also the rela-
tively short fixation times of newborns. The mean fixation
time for newborns (averaged across all conditions) was
only 13.8 s, as compared to 33.3 s, 25.6 s and 24.6 s, for
l-, 2- and 3-month olds, respectively. The well-docu-
mented limitations in neonates’ wakefulness and atten-
tiveness [8-111 likely account for this difference.
Our results imply that young infants’ ability to discrimi-
nate chromatic stimuli is relatively poor. Even at 2-months
of age, infants still fail to discriminate two large chromatic
stimuli (16” red and yellow) which are widely separated
within the Rayleigh region of the spectrum (i.e., the region
comprising wavelengths above 540 nm). Do these dis-
crimination failures within the Rayleigh region imply that
young infants are like traditional adult protanopes or deu-
teranopes who lack specific chromatic mechanisms? Sev-
eral aspects of our data and those of others argue against
this interpretation. First, the few successful Rayleigh dis-
criminations made by l- and 2-month olds in the present
study as well as by neonates in a previous study [ 121,
suggests that unlike protanopes or deuteranopes, young
infants possess at least two retinal receptor mechanisms
(likely the LWS and MWS cones) and a post-receptoral
mechanism (likely the R/G channel) that is capable of
comparing the output of the two cone types. This sugges-
tion is corroborated by recent electrophysiological data
[ 131 showing that at least up to the level of the primary
visual cortex, young infants’ contrast sensitivity thresholds
to bichromatic [ R/(R + G)] sinusoidal gratings appear best
explained by the existence of functional MWS and LWS
cones and an accompanying post-receptoral mechanism.
However, despite the apparent presence of these
mechanisms, it is obvious that they are far from mature
during early infancy. As an alternative to a ‘missing
mechanism’ explanation, Banks and his colleagues [5]
offer a simpler yet perhaps more elegant solution. Based
on examination of the anatomical differences between sev-
eral parameters of young infants’ and adults’ optical and
retinal mechanisms, they propose that much of the exist-
ing infant psychophysical data can be accounted for by
positing that young infants possess all of the requisite vi-
sual mechanisms, but that these are reduced quantita-
tively. In other words, relative to adults, infants’ visual
mechanisms operate inefficiently. With regard to color vi-
sion, the inefficiency model predicts that chromatic dis-
101
crimination is possible, but is limited to relatively large
chromatic stimuli with relatively wide spectral separation.
These predictions are based on estimates of discrimina-
tion thresholds or ‘Weber fractions’ for different combi-
nations of chromatic stimuli. Given the spectral charac-
teristics of the stimuli employed in the present studies, the
model would predict that the red-green pair would have
the lowest Weber fraction and thus be the ‘easiest’ to
discriminate. The Weber fractions for the yellow-green
pair would be higher than those for red-green and even
higher still for the yellow-red pair. The developmental pat-
tern reported here appears consistent with the predictions
of the inefficiency model and helps further support the
emerging conclusion that young infants’ vision is a quan-
titatively rather than qualitatively reduced form of adults’.
This work was supported by Natural Sciences and En-
gineering Research Council of Canada Grant # 3-32-674.
Many thanks to Michele Mercer for critical comments on
an earlier version of the manuscript and to all of the in-
fants and their parents for enthusiastic cooperation.
111Hamer, R.D., Alexander, K. and Teller, D.Y., Rayleigh discrimi-
nations in young human infants, Vision Res., 22 (1982) 575-587.
121Clavadetscher, J.E., Brown, A.M., Ankrum, C. and Teller, D.Y.,
Spectral sensitivity and chromatic discriminations in 3-and ‘I-week-
old infants, J. Opt. Sot. Am. A, 5 (1988) 2093-2105.
131 Packer, O., Hartmann, E.E. and Teller, D.Y., Infant color vision:
The effect of test field size on Rayleigh discriminations, Vision Res.,
24 (1984) 1247-1260.
[41 Vamer, D., Cook, J.E., Schneck, M.E., McDonald, M. and Teller,
D.Y., Tritan discriminations by l- and 2-month-old human infants,
Vision Res., 25 (1985) 821-831.
151 Banks, M.S. and Bennett, P.J., Optical and photoreceptoral im-
maturities limit the spatial and chromatic vision of human neo-
nates, J. Opt. Sot. Am. A, 5 (1988) 2059-2079.
[61 Brown, A.M., Development of visual sensitivity to light and color
vision in human infants: a critical review, Vision Res., 30 (1990)
1159-1188.
[71 Morrone, M.C., Burt, D.C. and Fiorentini, A., Development of
infant contrast sensitivity to chromatic stimuli, Vision Res., 37
(1993) 2535-2552.
PI Adams, R.J., Further exploration of human neonatal chromatic-
achromatic discrimination, J. Exp. Child Psychol., in press.
[91 Adams, R.J. Maurer, D. and Cashin, H.C., The influence of stimu-
lus size on newborns’ discrimination of chromatic from achromatic
stimuli, Vision Res., 30 (1990) 2023-2030.
Adams, R.J., Maurer, D. and Davis, M., Newborns’ discrimination
of chromatic from achromatic stimuli, J. Exp. Child Psychol., 41
(1986) 267-281.
Adams, R.J., Courage, M.L., and Mercer, M.E., Deficiencies in
human neonates’ color vision: Photoreceptoral and neural expla-
nations, Behav. Brain Res., 43 (1991), 109-114.
Adams, R.J., Newborns’ discrimination among mid- and long-
wavelength stimuli, J. Exp. Child Psycho/., 47 (1989) 130- 141.
Allen, D., Banks, M.S. and Norcia, A.M., Does chromatic sensi-
tivity develop more slowly than luminance sensitivity?, Vision Res.,
33 (1993) 2553-2562.