Psychophysiology, 39 ~2002!, 482–490. Cambridge University Press. Printed in the USA.
Copyright © 2002 Society for Psychophysiological Research
DOI: 10.1017.S0048577202010764
Visual categorization during childhood: An ERP study
M. BATTY and M. J. TAYLOR
CerCo, CNRS-UMR 5549, Toulouse, France
Abstract
Categorization is a basic means of organizing the world around us and offers a simple way to process the mass of stimuli
one perceives every day. The ability to categorize appears early in infancy, and has important ramifications for the
acquisition of other cognitive capacities, but little is known of its development during childhood. We studied 48 children
~7–15 years of age! and 14 adults using an animal0nonanimal visual categorization task while event-related potentials
~ERPs! were recorded. Three components were measured: P1, N2, and P3. Behaviorally, the children performed the task
very similarly to adults, although the children took longer to make categorization responses. Decreases in latencies ~N2,
P3! and amplitudes ~P1, N2, P3! with age indicated that categorization processes continue to develop through childhood.
P1 latency did not differ between the age groups. N2 latency, which is associated with stimulus categorization, reached
adult levels at 9 years and P3 latency at 11 years of age. Age-related amplitude decreases started after the maturational
changes in latencies were finished.
Descriptors: Event-related brain potentials, Visual categorization, Development
~Hickling & Gelman, 1995; Odom & Cook, 1996; Rosch, 1976!.
At 4–5 years, children understand that living creatures have particular properties that distinguish them from nonliving ~Gutheil,
Vera, & Keil, 1998; Keil, 1989!, and by 5– 6 years of age, children
correctly categorize animals versus nonanimals ~Thibault, 1999!.
Only a very few studies have examined categorization beyond the
preschool age range. Rosch ~1976! found that whereas children of
3 years of age are able to categorize known stimuli at the basic
level, superordinate classifications mature during school-age years.
Thus, the capacity to categorize is seen early in infancy ~Eimas &
Quinn, 1994!, yet the ability for increasingly abstract or complex
classifications appears to continue to evolve over childhood
~Desprels-Fraysse & Lecacheur, 1996; Gutheil & Gelman, 1997!.
The ability for categorization has important ramifications for the
acquisition of other cognitive capacities, playing a critical role
in memory, reasoning, problem solving, and learning to learn
~O’Sullivan, 1996; Thibault, 1999!.
ERPs have become an important means to investigate cognitive
development and its underlying neural correlates, being the most
readily applied of the neuroimaging techniques in children. Previous studies have demonstrated that ERPs provide sensitive measures of the development of visual attention to stimulus categories
and can evaluate the first stages of visual processing in children
and adults ~Jonkman et al., 1999; Taylor & Khan, 2000; Taylor &
Pang, 1999!. There are a few reports in the literature where categorization processes have been studied with ERPs in adults using
large series of different stimuli ~Antal, Kéri, Kovacs, Janka, &
Benedek, 2000; Thorpe, Fize, & Marlot, 1996!. Thorpe et al.
reported difference waves between ERPs to animals and nonanimals, measured frontally ~onset of 150–170 ms!, reflecting early
differentiation of complex stimuli. An f MRI study showed only
posterior activation for this categorization task ~Fize et al., 2000!,
The process of categorization is a way of organizing the world
around us and reducing the enormous amount of information that
one encounters on a daily basis. Categorizations are often based on
shared common features that are perceived as being typical of the
category, or on an item’s similarity to exemplars or prototypes of
the category ~Smith & Jonides, 2000!. Categories can be groupings
of similar objects ~e.g., chairs! or quite dissimilar objects, which
share an abstract feature, or physical features that give a “family
resemblance” ~Krascum & Andrews, 1998; Rosch & Mervis, 1975!.
Thus, objects within a group can differ widely depending on the
breadth of the category ~e.g., living vs. nonliving, compared to
crows vs. ravens!. There are three levels of categories described in
the literature, basic ~bird!, subordinate ~crow!, and superordinate
~animals!, with the similarities among members of a level decreasing with the higher levels ~i.e., superordinate!.
Despite the fundamental nature of this cognitive process, categorization appears to undergo considerable maturation. Studies of
categorization in infants with preferential looking paradigms have
shown that at 3 months of age, infants could form perceptual
category representations ~Mareschal & Quinn, 2001! and distinguish dogs ~or cats! from birds ~Quinn, Eimas, & Rosenkrantz,
1993!. At 4 months, infants are able to learn new categories, with
either real or artificial stimuli ~Quinn & Eimas, 1996, for a review!. The age at which a child will group items in the same
manner as an adult depends on the type of categorization required
We thank Hélène Chevalier and Denis Fize for their help in setting up
this study, and the Fondation pour la Recherche Médicale for financial
support.
Address reprint requests to: Magali Batty, CerCo, CNRS-UMR 5549,
133 Rte de Narbonne, Faculté de Médecine de Rangueil, 31062 Toulouse,
France. E-mail: [email protected].
482
Visual categorization during childhood
although the difference potential was evident widely spread over
the scalp ~VanRullen & Thorpe, 2001!. Antal et al. also measured
a number of ERP components in an animal0nonanimal task and
found N1, P2, and N2 components that were more negative for
nonanimal stimuli, suggesting that the categorization process started
early at N1 ~150 ms!. In a single study of visual categorization with
adults and children, ERPs were used to examine the classification
of prototypical and nonprototypical cats and dogs; unlike the
preceding two studies, however, there were many repetitions of the
same stimuli per category ~Ellis & Nelson, 1999!. For the children
~6-year-olds!, P3 amplitude was larger to prototypes than to nonprototypes, whereas for adults P3 latency was shorter to prototypes
than to nonprototypes. Thus, the prototypicality of stimuli influenced the P3 in a different manner in children than adults, suggesting that the use of prototypes in categorization changes with
age ~e.g., Meints, Plunkett, & Harris, 1999!, although developmental change per se was not investigated, nor the early stages of
information processing.
Given the importance of categorization for a wide range of
cognitive functions, but the limited knowledge about how this
ability develops, we were interested in determining if, and if so
which, stage~s! of this processing change over childhood. Some
categorization skills appear during infancy and behavioral studies
have shown considerable development during the preschool years,
but there is a paucity of studies during childhood or into adolescence. There are many examples in development where a capability, although appearing early in life, continues to develop until
nearly adulthood, such as face expertise ~Chung & Thompson,
1995; Taylor, Edmonds, McCarthy, & Allison, 2001! or processing
related to language ~e.g., Holcomb, Coffey, & Neville, 1992;
Taylor, 1988!. Thus, we chose a superordinate category that was
behaviorally and semantically possible in young children, but the
development of which may well extend with experience into the
school-age years ~Rosch, 1976!. Children from 7 to 15 years of age
and adults were included and the task was a broad-based categorization, animal versus nonanimal. We used natural images, rather
than line drawings ~Ellis & Nelson, 1999! or words ~Taylor &
Eals, 1996! to have stimuli as realistic as possible, and to allow
processing independent of reading skills. To determine if processing stages developed differentially over childhood, we assessed
several levels of processing by measuring the following ERP
components: P1, which reflects early visual processing or encoding, N2 which is implicated in categorization processes, and P3,
reflecting a final stage of evaluation or memory updating.
Method
Participants
Sixty-two participants were included; 14 young adults ~mean 23.9 6
1.0 years, 7 women! and 48 children, divided into four age groups:
7–8 years ~mean 8.0 6 0.7 years, n 5 12, 8 girls!, 9–10 years
~mean 10.0 6 0.6 years, n 5 14, 9 girls!, 11–12 years ~mean 11.8 6
0.6 years, n 5 12, 3 girls!, 13–15 years ~mean 14.1 6 0.7 years, n 5
10, 5 girls!. The children had been recruited via friends and
colleagues within the institution and local schools. All children
were given the abbreviated WISC-III ~block design and vocabulary! to ensure that all were in normal cognitive range for their age.
All participants had normal or corrected to normal eyesight. There
were 5 left-handed children, spread across age groups and 2 lefthanded adults. The experimental protocol was approved by the
“Comité Opérationnel pour l’Ethique dans les Sciences de la Vie”
of the Centre National de la Recherche Scientifique, and all par-
483
ticipants ~as well as the parents of the children! gave informed,
written consent.
Experimental Task
The present study used an animal categorization task adapted from
the one used by Thorpe et al. ~1996! with adults. Stimuli were 540
color photographs culled from CDs of natural scenes. Twenty-five
percent were targets representing animals in their natural environments ~mammals, birds, reptiles, fish!; the remainder were nontargets that included landscapes, trees, flowers, and man-made objects
without animals ~Figure 1!. Pictures ~18 3 12 cm! were presented
in the center of a computer screen at a distance of 50 cm, for 80 ms
in random order using E-Prime, which synchronized the presentation of the stimuli and triggers. The interstimulus interval ~ISI!
varied between 1,100 and 1,500 ms. Each of the 540 pictures was
presented once, none repeated. Participants responded to target
trials ~animals! by pressing the space bar as quickly and as accurately as possible with their dominant hand; they did not respond
to nontargets ~go0no-go task!. The trials were divided into three
equal blocks, and each of these was also divided into three subblocks, such that a pause was given every 60 trials, and a longer
pause after 180 trials. This allowed even the youngest children to
complete the task without undue fatigue. The experimental trials
were preceded by 20 practice trials, to accustom the participants to
the rapid presentation rate and range of stimuli; none of these
images were included in the subsequent trial blocks.
ERP Recording and Analysis
ERPs were recorded from 30 active electrodes ~Ag0Agcl! applied
with an EasyCap ~FMS Falk Minow; www.easycap.de! electrode
cap ~FP1, FP2, F3, F4, F7, F8, Fz, FT9, FT10, FP5, FP6, T7, T8,
C3, C4, CP5, CP6, TP9, TP10, P3, P4, P7, P8, Pz, POz, PO9,
PO10, O1, O2, Oz!. Cz served as reference during the recording
and was reinstated when the average reference was calculated.
Electrode impedances were below 5 kV. EOG was monitored with
a supraorbital and two outer canthi electrodes. ERPs were collected using the NeuroScan 4.0 amplified with SynAmps ~Neuroscan Labs, www.neuro.com!, with a gain of 500 and a sampling
rate of 500 Hz. Off-line reaveraging was conducted; continuous
files were epoched into 1,100-ms sweeps, including a 100-ms
prestimulus baseline and digitally filtered with a bandpass of
0.1–30 Hz. Trials with ocular or muscular artefact that exceeded
6150 mV for children and 690 mV for adults were rejected.
Data Analyses
EEG data trials were averaged by category ~animal, nonanimal!;
those trials with an incorrect behavioral response were excluded.
The components P1, N2, and P3 ~see Figure 2! were measured at
the electrodes where their amplitudes were maximal ~Picton et al.,
2000!. The P1 peak was measured at occipital electrodes ~O1, O2,
and POz!, N2 at central electrodes ~C3, C4, and Cz!, and P3 at Pz
only, as P3 had a very constrained distribution around Pz with this
task ~see Figure 2!. The peak amplitudes were measured from the
prestimulus baseline and the latencies from stimulus onset. The
data were submitted to repeated measures ANOVAs ~Age ~5! 3
Stimulus ~2! 3 Electrode ~3!!. Type I errors associated with inhomogeneity of variance were controlled by decreasing the degrees
of freedom using the Greenhouse–Geisser epsilon ~E!. The probability estimates were based on these reduced degrees of freedom;
the uncorrected degrees of freedom are given in the text, along
with the epsilon values. The data were analyzed as a function of
age, stimulus category, and electrodes, and for P1 and N2 as
484
M. Batty and M.J. Taylor
Figure 1. Examples of the 540 stimuli used, although all were presented in color; three targets ~animals! on the left, and three
nontargets ~nonanimals! on the right, are shown.
function of hemisphere, with post hoc tests using Bonferroni
statistics.
Results
Behavioral Data
Performances were very good. The average percentage of correct
responses ~target and nontargets! was 98.78% ~60.14! for the
adults. There was a significant age effect for accuracy, F~4,57! 5
3.739, p , .009, with an increase in accuracy with increasing age.
Nevertheless, children performed very well, with 96.3% ~60.83!
accuracy for the youngest. Few errors were made, but it was
interesting to note that errors made by adults and by children were
very often to the same pictures.
No reaction time ~RT! difference was found between correct responses to targets ~animals! and error responses to nontargets ~nonanimals!; only correct response RTs were analyzed further. In adults,
the mean RT was 414 ms. Our data showed the usual decrease in
reaction times with age ~Figure 3!. Adults’ RTs differed significantly from the RTs of children 11–12 years old ~475 ms, p , .02!,
9–10 years old ~520 ms, p , .0001!, and 7–8 years old ~524 ms, p ,
.0001!, but not from the 13–15-year-olds ~430 ms, p 5 .820!.
ERPs
P1. P1 latency showed a significant age effect, F~4,57! 5 20.878,
p , .0001, driven by the adults’ latencies being shorter ~105.36 6
1.32 ms! than children’s ~118.73 6 1.43 ms!, but there were no
differences among the groups of children ~Figure 4a!. The latency
showed a significant electrode effect, F~2,114! 5 77.2, p , .0001;
the P1 was shorter at POz than at O1 and O2, with no difference
between O1 and O2.
For P1 amplitude there were main effects of age, F~4,57! 5
18.78, p , .0001 ~Figure 4b!, electrode, F~2,114! 5 34.36, p ,
.0001, stimulus, F~1,57! 5 60.37, p , .0001 ~Figure 5a!, and
an Age 3 Electrode interaction, F~8,114! 5 3.99, p , .0001,
E 5 .933. The adults’ P1 amplitude was smaller than that of the
groups of children 12 and younger, p , .0001 ~Figure 4b!. This
decrease in amplitude with age was marked; the youngest children had a P1 about six times the amplitude of the adults. The
electrode effect was due to P1 at POz being smaller than at O1
and O2, p , .0001, but there was no significant difference
between the lateral electrodes. The stimulus effect was due to
P1 being slightly but consistently larger across age groups when
evoked by targets ~mean 18.24 mV! than by nontargets ~17.18 mV;
Figure 4b!.
Visual categorization during childhood
485
Figure 2. Averaged ERPs waveforms for targets for adults, 7–8-year-olds, and 11–12-year-olds displayed at 13 of the electrode sites.
Components measured are indicated by arrows.
N2. In adults, N2 had a mean latency of 260 ms. A significant
age main effect was seen, F~4,57! 5 32.85, p , .0001, due to
decreasing latencies with increasing age, although only the latency
for the youngest children ~7–8-year-olds, 293 ms!, showed a significant difference with other groups, p , .0001 ~Figure 4c!.
Although there was an overall effect of age on N2 amplitude
due to a decrease with age, F~4,57! 5 5.20, p , .001, there were
no differences among the groups of children, nor between adults
and adolescents; adults’ N2 amplitude ~24.99 mV! differed from
the amplitude of 11–12-year-olds ~29.71 mV, p , .02!, from the
9–10-year-olds ~210.12 mV, p , .005!, and from 7–8-year-olds
~210.64 mV, p , .002; Figure 4d!. A hemisphere effect, F~2,114! 5
25.71, p , .0001, due to consistently larger amplitudes over the
right hemisphere site ~29.43 mV vs. 27.36 mV!, was seen across
the five age groups.
P3. P3 latency was longer for targets ~animals: 442 ms! than
nontargets ~nonanimals: 415 ms!, F~1,57! 5 11.97, p , .001;
Figures 4e and 5c. There was also an age effect on P3 latency,
F~4,57! 5 71.71, p , .0001. In adults, the mean latency was about
370 ms, which did not differ significantly from the 13–15- or
11–12-year-olds, but the latency was much longer for 9–10- and
7–8-year-olds, 495 ms and 570 ms, respectively ~Figure 4e!.
P3 amplitude was much larger to targets, F~1,57! 5 239.35,
p , .0001 ~Figures 4f and 5c!. P3 to stimuli including animals was
13.86 mV and only 5.85 mV to stimuli that were nonanimals.
Although there was a small age effect, F~4,57! 5 2.86, p , .03, it
was due to amplitudes only for the targets decreasing with age
~Figure 4f !.
Discussion
Figure 3. Mean reaction times ~in milliseconds! with error bars for adults
and the four groups of children.
The present study demonstrates that the ability to categorize images depending upon whether they contain an animal or not appears to be essentially mature at a behavioral accuracy level in our
youngest age group. However, significant changes were still occurring in the ERPs that varied with the component measured.
These data suggest that some aspects of this superordinate catego-
486
M. Batty and M.J. Taylor
Figure 4. Mean latencies ~in milliseconds! and amplitudes ~in microvolts! with error bars of the three components, P1 ~a,b!, N2 ~c,d!,
and P3 ~e,f !, across the five age groups, averaged over the electrode sites where the peaks were measured ~i.e., for P1: O1, O2, and
POz; for N2: C3, C4, and Cz; P3 was measured only at Pz!. Targets ~animals! are in solid lines and nontargets ~nonanimals! are in dotted
lines.
rization mature early, whereas other aspects evolve more slowly
over childhood.
Behavioral Data
The hit rate for adults was very high, at 98.8%, consistent with
reports in the literature with a similar task ~Antal et al., 2000;
Fabre-Thorpe, Richard & Thorpe, 1998; Thorpe et al., 1996!. It is
important to note that despite a significant difference in hit rate
with age, the children’s performance was also very high, being at
96% for the 7–8-year-olds. This is an impressive performance,
when the stimuli were presented for only 80 ms and subjects had
no prior knowledge of the size, location, number, or species of
animal that could appear in the pictures. Some animals were also
only partially visible and the scenes were so varied that they
offered no reliable cues. The errors made by the children, across
age groups, were also in response to the same images that produced errors in adults. For example, a picture of a mushroom in the
forest was classified as “animal” by 47% of subjects, perhaps due
to its shape like a jellyfish, as other pictures of vegetables or fruits
did not produce errors. In contrast, a scene with a man on a horse
was classified in 32.3% of cases as “nonanimal,” we suppose, as
the horse was seen as a means of transport, or the image of the man
served to inhibit the “animal” categorization. These behavioral
data demonstrate that the categorization processes that children
used for this task were apparently the same as those used by adults.
The ability to discriminate animals on the basis of very little
information is developed by an early age. The RTs however,
decreased steadily with age as is seen ubiquitously ~Kail, 1993!,
with the mean RTs for the youngest group of children being 110 ms
longer than for the adults.
ERPs
Age effects. Amplitudes of the three components decreased with
age but with differing patterns. Early ERP components are often
larger in children than adults, and this effect is particularly marked
with rapid presentation rates of stimuli as used in this task ~e.g.,
Theunissen, Alain, Chevalier, & Taylor, 2001!. The effects of
maturation on P1 seemed to start very early, before 7 years of age,
and continued during childhood. This could reflect the automation
of visual encoding processes expressed as a decrease in the number
of neurones involved. Although the number of synapses decreases
after 8 years of age ~Huttenlocher, 1990!, and this underlying
neuroanatomic maturation may be the cause of the P1 amplitude
decrease with age, there may also be cognitive or attentional
factors, as this is certainly the case for the other peaks measured.
For example, N2 amplitude decreased with age, but this started
only at adolescence. P1 showed no decrease in latency during
childhood but an abrupt change between 13–15-year-old children
Visual categorization during childhood
Figure 5. Averaged ERP waveforms for targets in solid lines and nontargets in dotted lines for all age group, at O1 ~a!, Cz ~b! and
Pz ~c!. Note that the X and Y axes are different for each set of plots, to maximize the component shown.
487
488
and adults. This late modification could be due to the large amplitude difference of this component between children and adults,
producing a spurious shift in the latency, with the disappearance of
the large positive slow wave underlying the P1 in children.
N2 latency reached adult levels very early ~9–10 years!. Other
studies have also reported an asymptote in N2 latency across a
variety of tasks by about 11 years, suggesting that the process
associated with this component, evaluation and categorization of
the stimulus, is mature at this age ~Johnson, 1989; Taylor, 1995!.
However, the onset of amplitude changes in N2 in adolescence
suggests that improvements in categorization processing could
develop late and continue until adulthood. It has been shown that
children learn basic categorizations early and superordinate levels
during later childhood ~Mervis & Crisafi, 1982; Rosch, 1976;
Thibault, 1999!, which could account for the continued development of this processing.
P3 amplitude to nontargets did not change with age for the ages
studied. In contrast, P3 to targets showed amplitude decreases with
age and, like N2, this decrease started only after 12 years of age.
We found that the P3 latency continued to decrease until 11–12
years, when adult latency level was reached. In contrast to N2,
there is greater variability in the age at which the P3 reaches adult
levels, ranging up until adulthood for more difficult tasks ~Taylor,
1995!.
Thus, P1, N2, and P3 amplitudes decreased with age, P1 decreased very early, from 7 years, and continued during childhood,
whereas the later components, N2 and P3, showed amplitude
decreases only later, after 12 years of age. Latencies also changed
with age, again, with each component having a different time
course. For the three peaks measured in this study, the latency
maturation was faster than the maturation in amplitude.
Task effects (target0nontarget). Across age groups, the amplitude of the P1 was larger to targets than to nontargets. Thus,
discrimination of targets from nontargets occurred very early, in
the first stage of processing. P1 is described as the earliest endogenous potential ~e.g., Mangun, 1995!, but most reports have linked
it with processing of spatial information. Some recent studies have
also shown that P1 is sensitive to nonspatial features ~Han, Fan,
Chen, & Zhuo, 1997; Taylor & Khan, 2000!, but in simple tasks
where subjects could easily apply top-down processing ~e.g., attend only to red rectangles!, the amplitude and0or latency effects
on P1 were more readily explained. Recent reports have found
ERP differences at 50–75 ms for more complex discrimination
tasks ~Debruille, Guillem, & Renault, 1998; Mouchetant-Rostaing,
Giard, Bentin, Aguera, & Pernier, 2000!, although others have
reported such early effects to be due to low-level processing
~VanRullen & Thorpe, 2001!. The present P1 effect could reflect
a very rapid, although likely not accurate, global processing of
stimuli significant to humans ~Mouchetant-Rostaing, Giard,
Delpuech, Echallier, & Pernier, 2000!. This is consistent with other
suggestions that the occipital cortex is directly involved in recognizing animals, and that this is not due to bottom-up feature
processing ~Martin, Ungerleider, & Haxby, 2000!.
N2 is classically associated with categorization processes ~Ritter et al., 1984!, but explanations of effects vary. Some research
has suggested that N2 reflects target detection, as its amplitude is
larger when evoked by a target, and its latency changes with
reaction time ~Paz-Caballero & Garcia-Austt, 1992!. Kopp, Mattler, Goertz, and Rist ~1996!, using a hybrid choice reaction go0
no-go procedure, suggested that N2 reflects either the detection of
targets or the inhibition of inappropriate response tendencies. In
M. Batty and M.J. Taylor
the present study, the paradigm required both target detection and
inhibition of inappropriate responses due to the brief presentation
time. Also targets, like nontargets, were so varied and without
easy visual cues to distinguish them ~both contained landscapes,
water scenes, etc.!, that the same processing was required for both,
for their correct categorization. These factors could explain the
lack of significant N2 differences as a function of target0nontarget
category.
P3 is generally thought to be involved in processes requiring
attention to targets and final stimulus evaluation or memory updating ~Picton, 1992!. As is seen ubiquitously, P3 amplitude was
much larger to targets than nontargets. P3 is not usually sensitive
to physical characteristics of the stimulus, but is modulated by the
difficulty of the task. Thus, the target P3 latency could estimate the
difficulty of discriminating targets from nontargets. These data
show that this was not a difficult task even for the children, as the
P3 latencies were shorter than often seen in the developmental
literature ~such as with semantic classification or memory tasks;
Berman & Friedman, 1993; Taylor & Eals, 1996; Taylor & Smith,
1995!. The marked differences in amplitude and the developmental
pattern of P3s measured, when evoked by targets or by nontargets,
suggested that the P3 evoked by targets was a P3b, and the P3
evoked by nontargets was a P3a. P3b is often not measured to
nontarget stimuli, as it is linked with target detection. Thus, the
component we measured within the same time window but on the
nontarget trials was most likely P3a. A positive component in the
300– 600-ms time range, with a smaller amplitude and shorter
latency to nontarget stimuli, has been described as P3a in a visual
attention task ~Eals & Taylor, 1996!. The current findings are also
consistent with the literature that describes P3a as an automatic
component, little or unaffected by attention ~Picton, 1992!. This
automatic aspect reflects the low level processes of P3a, and could
explain our finding of no developmental change in this peak after
7 years of age. The differences between target and nontarget P3s
reinforces the earlier conclusion that there is rapid categorization
of the stimuli as “animal” or “nonanimal,” reflected in the P1 and
N2 and completed before the P3.
Hemispheric asymmetries. N2 amplitude was larger over the
right than the left hemisphere. This asymmetry was present across
age groups and for both targets and nontargets. These results are in
disagreement with those reported by Parrot, Doyon, Démonet,
and Cardebat ~1999! and Marsolek ~1995!, who found a left
hemispheric superiority in categorization processing. However,
Heinze, Hinrichs, Scholz, Burchert, and Mangun ~1998! observed an effect of the type of treatment ~global vs. local! on
N2 lateralization, with an N2 left lateralization when processing
was local, and a right lateralization when global processing was
required. The stimuli used in Parrot et al.’s and Marsolek’s
categorization tasks were simple geometric forms and would
have required only local processing for task performance. Other
studies that have investigated global versus local processing of
visual stimuli have also found asymmetries, larger over the right
hemisphere, for global processing starting at the N2 latency
~Evans, Shedden, Hevenor, & Hahn, 2000; Proverbio, Minniti,
& Zani, 1998!. In the present task, stimuli were rapidly presented complex visual scenes in which the target detection would
have required a global treatment of the picture. Thus, we suggest that the differences of lateralization were due to the type of
categorization required by the task, and our results argue for
global processing for this broad-based categorization task being
present and lateralized, from 7 years of age.
Visual categorization during childhood
489
Conclusions
At first glance, the similarities observed in the accuracy and in the
errors made in categorizing the stimuli across age groups suggested that the categorization processes are not only functional
in young children but also qualitatively very similar to the same
processes in adults. Yet the time to respond to the stimuli was more
than 100 ms longer for the children. There are a number of possible
explanations for this effect. The developmental literature is replete
with reports of the steadily increasing speed of responses over
childhood ~e.g., Kail, 1993!, which suggests this difference is due,
at least in part, to development in the motor system. However, the
ERP analyses demonstrated that there is also cognitive development. N2 and P3 showed latency decreases in the earlier age
group~s! reflecting increased speed for the cognitive processes
associated with these components. All three components showed
decreasing amplitudes with age, which suggests a progressive
decrease in the number of neurons, or cortical area, recruited or
required for task performance. Thus, despite the task being easy
for the children, underlying neural processing was still developing
over the age range tested.
It is important to note that the amplitude and latency did not
change in a parallel manner for the components. The maturation of
the latencies preceded that of the amplitudes; the latter continued
to change until adulthood. Thus, a process appeared to reach its
maximal or adult speed, and then the cortical activation implicated
decreased. The earlier the latency of the component, the sooner
the maturation changes in latency were completed, and the sooner
the amplitude changes started. N2, for example, reflects the process of categorization, and the latency reached adult levels after 9
to 10 years, but the amplitude changes did not start until 12 years
of age.
The present investigation confirms adult studies, and shows
that children, like adults, are capable of correctly categorizing
complex visual stimuli, even when presented very briefly ~80 ms!.
The finding of the N2 lateralization across age groups, in combination with the behavioral data, suggest that the ability to
make this superordinate categorization of “animal” is largely
mature at 7 years of age. Furthermore, the fact that there were
no Age 3 Stimulus interactions suggests that the processes linked
with the P1 and P3 changed only in a qualitative manner across
age groups. The continued developmental changes may, speculatively, be linked with more general cognitive factors that continue to mature until adulthood and that may well underlie more
abstract categorization tasks.
REFERENCES
Antal, A., Kéri, S., Kovacs, G., Janka, Z., & Benedek, G. ~2000!. Early and
late components of visual categorization: An event-related potential
study. Cognitive Brain Research, 9, 117–119.
Berman, S., & Friedman, D. ~1993!. A developmental study of ERPs during
recognition memory: Effects of picture familiarity, word frequency, and
readability. Journal of Psychophysiology, 7, 97–114.
Chung, M. S., & Thomson, D. M. ~1995!. Development of face recognition. British Journal of Psychology, 86, 55–87.
Debruille, J. B., Guillem, F., & Renault, B. ~1998!. ERPs and chronometry
of face recognition: Following-up Seeck et al. and George et al. NeuroReport, 9, 3349–3353.
Desprels-Fraysse, A., & Lecacheur, M. ~1996!. Children’s conception of
object, as revealed by their categorizations. Journal of Genetic Psychology, 157, 49– 64.
Eals, M., & Taylor, M. J. ~1996!. ERPs reflecting parallel and serial
processing of color arrays. Functional Neuroscience, EEG Suppl, 46,
301–313.
Eimas, P. D., & Quinn, P. C. ~1994!. Studies on the formation of perceptually based basic-level categories in young infants. Child Development, 65, 903–917.
Ellis, A. E., & Nelson, C. A. ~1999!. Category prototypicality judgements
in adults and children: Behavioral and electrophysiological correlates.
Developmental Neuropsychology, 15, 193–211.
Evans, M. A., Shedden, J. M., Hevenor, S. J., & Hahn, M. C. ~2000!. The
effect of variability of unattended information on global and local
processing: Evidence for lateralization at early stages of processing.
Neuropsychologia, 38, 225–239.
Fabre-Thorpe, M., Richard, G., & Thorpe, S. J. ~1998!. Rapid categorization of natural images by rhesus monkeys. NeuroReport, 9, 303–308.
Fize, D., Boulanouar, K., Chatel, Y., Ranjeva, J. P., Fabre-Thorpe, M., &
Thorpe, S. ~2000!. Brain areas involved in rapid categorization of
natural images: An event-related fMRI study. NeuroImage, 11, 634– 643.
Gutheil, G., & Gelman, S. A. ~1997!. Children’s use of sample size and
diversity information within basic-level categories. Journal of Experimental Child Psychology, 64, 159–174.
Gutheil, G., Vera, A., & Keil, F. C. ~1998!. Do houseflies think? Patterns
of induction and biological beliefs in development. Cognition, 66,
33– 49.
Han, S., Fan, S., Chen, L., & Zhuo, Y. ~1997!. On the different processing
of wholes and parts: A psychological analysis. Journal of Cognitive
Psychology, 9, 687– 698.
Heinze, H. J., Hinrichs, H., Scholz, M., Burchert, W., & Mangun, G. R.
~1998!. Neural mechanisms of global and local processing: A com-
bined PET and ERP study. Journal of Cognitive Neuroscience, 10,
485– 498.
Hickling, A. K., & Gelman, S. A. ~1995!. How does your garden grow?
Early conceptualization of seeds and their place in the plant growth
cycle. Child Development, 66, 856–876.
Holcomb, P. J., Coffey, S. A., & Neville, H. J. ~1992!. Visual and auditory
sentence processing: A developmental analysis using event-related potentials. Developmental Neuropsychology, 8, 203–241.
Huttenlocher, P. R. ~1990!. Morphometric study of human cerebral cortex
development. Neuropsychologia, 28, 517–527.
Johnson, R., Jr. ~1989!. Developmental evidence for modality-dependent
P300 generators: A normative study. Psychophysiology, 26, 651– 667.
Jonkman, L. M., Kemner, C., Verbaten, M. N., Van Engeland, H., Kenemans,
J. L., Camfferman, G., Buitelaar, J. K., & Koelega, H. S. ~1999!.
Perceptual and response interference in children with attention-deficit
hyperactivity disorder, and the effects of methylphenidate. Psychophysiology, 36, 419– 429.
Kail, R. ~1993!. Processing time decreases globally at an exponential rate
during childhood and adolescence. Journal of Experimental Child
Psychology, 56, 254–265.
Keil, F. C. ~1989!. Concepts, kinds, and cognitive development. Cambridge,
MA: MIT Press.
Kopp, B., Mattler, U., Goertz, R., & Rist, F. ~1996!. N2, P3 and the
lateralized readiness potential in a no-go task involving selective response priming. Electroencephalography and Clinical Neurophysiology, 99, 19–27.
Krascum, R. M., & Andrews, S. ~1998!. The effects of theories on children’s acquisition of family-resemblance categories. Child Development, 69, 333–346.
Mangun, G. R. ~1995!. Neural mechanisms of visual selective attention.
Psychophysiology, 32, 4–18.
Mareschal, D., & Quinn, P. C. ~2001!. Categorization in infancy. Trends in
Cognitive Sciences, 10, 443– 450.
Marsolek, C. J. ~1995!. Abstract visual-form representations in the left
cerebral hemispheres. Journal of Experimental Psychology: Human
Perception and Performance, 21, 375–386.
Martin, A., Ungerleider, L. G., & Haxby, J. V. ~2000!. Category specificity
and the brain: The sensory0motor model of semantic representations of
objects. In M. S. Gazzaniga ~Ed.!, The new cognitive neurosciences,
~2nd ed., pp. 1023–1036!. Cambridge, MA: MIT Press.
Meints, K., Plunkett, K., & Harris, P. L. ~1999!. When does an ostrich
become a bird? The role of typicality in early word comprehension.
Developmental Psychology, 35, 1072–1078.
490
Mervis, C. B., & Crisafi, M. A. ~1982!. Order of acquisition of subordinate-,
basic-, and superordinate-level categories. Child Development, 53,
258–266.
Mouchetant-Rostaing, Y., Giard, M. H., Bentin, S., Aguera, P.-E., & Pernier,
J. ~2000!. Neurophysiological correlates of face gender processing in
humans. European Journal of Neuroscience, 12, 303–310.
Mouchetant-Rostaing, Y., Giard, M. H., Delpuech, C., Echallier, J. F., &
Pernier, J. ~2000!. Early signs of visual categorization for biological and
non-biological stimuli in humans. NeuroReport, 11, 2521–2525.
Odom, R. D., & Cook, G. L. ~1996!. Valuing of identity, distribution of
attention, and perceptual salience in free and rule-governed classifications. Journal of Experimental Child Psychology, 61, 173–189.
O’Sullivan, J. T. ~1996!. Children’s metamemory about the influence of
conceptual relations on recall. Journal of Experimental Child Psychology, 62, 1–29.
Parrot, M., Doyon, B., Démonet, J. F., & Cardebat, D. ~1999!. Hemispheric
preponderance in categorical and coordinate visual processes. Neuropsychologia, 37, 1215–1225.
Paz-Caballero, M. D., & Garcia-Austt, E. ~1992!. ERP components related
to stimulus selection processes. Electroencephalography and Clinical
Neurophysiology, 82, 369–376.
Picton, T. W. ~1992!. The P300 wave of the human event-related potential.
Journal of Clinical Neurophysiology, 9, 456– 479.
Picton, T. W., Bentin, S., Berg, P., Donchin, E., Hillyard, S. A., Johnson, R.,
Miller, G. A., Ritter, W., Ruchkin, D. S., Rugg, M. D., & Taylor, M. J.
~2000!. Guidelines for using human event-related potentials to study
cognition: Recording standards and publication criteria. Psychophysiology, 37, 127–152.
Proverbio, A. M., Minniti, A., & Zani, A. ~1998!. Electrophysiological
evidence of a perceptual precedence of global vs. local visual information. Cognitive Brain Research, 4, 321–334.
Quinn, P. C., & Eimas, P. D. ~1996!. Perceptual cues that permit categorical
differentiation of animal species by infants. Journal of Experimental
Child Psychology, 63, 189–211.
Quinn, P. C., Eimas, P. D., & Rosenkrantz, S. L. ~1993!. Evidence for
representations of perceptually similar natural categories by 3 monthold and 4 month-old infants. Perception, 22, 463– 475.
Ritter, W., Ford, J. M., Gaillard, A. W. K., Harter, M. R., Kutas, M.,
Näätänen, R., Polich, J., Renault, B., & Rohrbaugh, J. ~1984!. Cognition
and event-related potentials I. The relation of negative potentials and
cognitive processes. Annals of the New York Academy of Sciences, 425,
24–38.
Rosch, E. H. ~1976!. Basic objects in natural categories. Cognitive Psychology, 8, 382– 439.
M. Batty and M.J. Taylor
Rosch, E. H., & Mervis, C. B. ~1975!. Family resemblances: Studies in the
internal structure of categories. Cognitive Psychology, 7, 575– 605.
Smith, E. E., & Jonides J. ~2000!. The cognitive neuroscience of categorization. In M. S. Gazzaniga ~Ed.!, The new cognitive neurosciences
~pp. 1013–1022!. Cambridge, MA: MIT Press.
Taylor, M. J. ~1988!. Developmental changes in ERPs to visual language
stimuli. Biological Psychology, 26, 321–338.
Taylor, M. J. ~1995!. The role of event-related potentials in the study of
normal and abnormal cognitive development. In F. Boller & J. Grafman
~Eds.!, Handbook of neuropsychology ~Vol. 10, pp. 187–211!. Amsterdam: Elsevier.
Taylor, M. J., & Eals, M. ~1996!. An event-related potential study of
development using visual semantic tasks. Journal of Psychophysiology,
10, 125–139.
Taylor, M. J., Edmonds, G. E., McCarthy, G., & Allison, T. ~2001!. Eyes
first! Eye processing develops before face processing in children.
NeuroReport, 12, 1671–1676.
Taylor, M. J., & Khan, S. C. ~2000!. Top-down modulation of early
selective attention processes in children. International Journal of Psychophysiology, 37, 135–147.
Taylor, M. J., & Pang, E. W. ~1999!. Developmental changes in early
cognitive processing. In C. Barber, G. C. Celesia, I. Hashimoto, & R.
Kakigi ~Eds.!, Functional neuroscience: Evoked potentials and magnetic fields ~EEG Supple. 49, pp. 145–153!. Amsterdam: Elsevier.
Taylor, M. J., & Smith, M. L. ~1995!. Age-related ERP changes in verbal
and nonverbal memory tasks. Journal of Psychophysiology, 9, 283–297.
Theunissen, E. L., Alain, C., Chevalier, H., & Taylor, M. J. ~2001!. Binding
occurs at early stages of processing in children and adults. NeuroReport, 12, 1949–1954.
Thibault, J. P. ~1999!. Le développement conceptuel. In J. A., Rondal, E.
Esperet, ~Eds!, Manuel de psychologie de l’enfant ~pp. 343–384!.
Sprimont: P. Mardaga.
Thorpe, S. J., Fize, D., & Marlot, C. ~1996!. Speed of processing in the
human visual system. Nature, 381, 520–522.
VanRullen, R., & Thorpe, S. J. ~2001!. The time course of visual processing: From early perception to decision making. Journal of Cognitive
Neuroscience, 13, 454– 461.
~Received May 31, 2001; Accepted December 10, 2001!