Review TRENDS in Cognitive Sciences Vol.8 No.

11 November 2004

Language and Conceptual Development series

Thought before language

Jean M. Mandler
Department of Cognitive Science, University of California San Diego, La Jolla, CA 92093, USA

To learn language infants must develop a conceptual Perceptual categories versus concept formation
base onto which language can be mapped. Recent Although concept formation might begin at birth, it is
research in infant cognitive development shows that at difficult to differentiate perceptual learning from concept
least by 9 months of age infants have developed a formation in the earliest months of life. One of the main
conceptual system sufficiently rich to allow language to techniques for studying perceptual learning in the first
begin. Evidence for this system is shown by categoriz- 6 months is the familiarization/preferential-looking tech-
ation of objects above and beyond their perceptual nique in which infants are shown a series of pictures of
appearance, problem-solving, long-term recall of events, instances of one category (e.g. cats) and then shown a
and inductive inferences. During the next year, early picture from another category (e.g. a dog) [5]. Longer
concepts gradually become refined. However, at the looking at the dog indicates that a perceptual category
time when language takes off they are often still less of cats (or pictures of cats?) has been formed. The
specific than many words in daily use, accounting for the findings with this method have been that by 3 months
phenomenon of overextension of word meaning. infants can form ‘basic-level’ categories of animals, such as
dogs and cats, as well as furniture, such as tables and
This review outlines the extensive evidence now available chairs, on the basis of faces or common shapes. There is
for conceptual thought before language, and discusses the some evidence that 3-month-olds can form a more global
varied uses to which preverbal concepts are put: problem- category of mammals as well [6], perhaps also by general-
solving, recall, inferential thought, and, of course, izing across faces.
language acquisition itself. Until recently, it was con- This use of the preferential-looking technique tells us
sidered established that babies are purely sensorimotor the sorts of perceptual categories infants can learn, but
creatures who have not yet created a conceptual system does not test for concept formation. This difference is
and so cannot think. According to Piaget [1], infants learn important, because much perceptual learning is an
many motor and perceptual skills in the first year, but not automatic procedure (as Piaget understood) that does
until midway in the second year do they begin to acquire not require conceptual thought [7–9]. Perceptual cate-
the conceptual system that will enable them to recall the gories or schemas are necessary for recognizing instances
past, learn language, and engage in mental problem- of a category, but do not in themselves specify what
concept the category represents, for example, animate or
solving. Motor and perceptual skills, including perceptual
inanimate things. Some possible ways to get from
categorization, are classic examples of procedural knowl-
perceptual schemas to concepts are discussed in Box 1.
edge (sometimes termed implicit knowledge), whereas
Nevertheless, some infancy researchers believe that
recall and mental problem-solving are classic examples of
separate perceptual and conceptual processes are not
declarative (sometimes termed explicit knowledge) [2,3].
required [10]. In this ‘one-process’ view, until language
The latter kind of knowledge requires an accessible
teaches unseen characteristics, concepts consist simply of
conceptual system.
perceptual categories, such as dogs, plus associated
The aim of this article is to summarize some of the
perceptions.
literature indicating that infants are building an acces-
In addition to asking how infants learn to recognize
sible conceptual system from an earlier age than tradi-
dogs or cups, we want to know what they think a dog or a
tionally thought. To bring the topic into manageable cup is. Some of the methods used to study this kind of
bounds, it omits the extensive work on infants’ learning conceptual knowledge and its use in preverbal thinking
about basic physical principles, such as that objects are are the object-examination test, which can uncover non-
solid and continue to exist when they disappear from sight perceptually-based categorization, deferred imitation,
[4]. Instead it concentrates on concepts more closely which measures recall of the past, generalized imitation,
related to early language learning: object kinds, simple which measures inductive inference (see Box 2), and
actions, and spatial relations. For the same reason the mental problem-solving (problem-solving without overt
brief conclusions about mapping language onto the trial and error).
conceptual base concentrate on early word learning rather
than the acquisition of syntax, which follows after the first
words. Preverbal problem-solving
Corresponding author: Jean M. Mandler ([email protected]). Little work has been done on mental problem-solving, but
Available online 29 September 2004 there is some research on means–end analyses in which
www.sciencedirect.com 1364-6613/$ - see front matter Q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.tics.2004.09.004
 Review TRENDS in Cognitive Sciences Vol.8 No.11 November 2004 509

Box 1. What kind of mechanism can create a preverbal concept?

Unless concepts are innate, there must be some way to convert example, Barsalou’s perceptual symbols [52]. Image-schemas are the
perceptual experience into a form that makes thinking possible. One abstract spatial representations used by cognitive linguists to
proposal is Karmiloff-Smith’s representational redescription, in which represent the primitive meanings underlying language [53]. Serendi-
procedural information is redescribed into a declarative format pitously, they also provide a good description of a number of infants’
independent of perception [50]. A related proposal is perceptual early concepts, in that they emphasize, as do infants, what objects do –
meaning analysis, a mechanism well suited for preverbal redescrip- how they move and interact – rather than their physical appearance.
tion [51]. I proposed this as a mechanism that extracts and Some common image-schemas are PATH, UP–DOWN, and LINK, and
summarizes incoming perceptual information, from which it creates spatial relations such as CONTAINMENT. For example, PATH is the
a store of primitive meanings that can be combined to form accessible simplest representation of any object moving through space, omitting
concepts such as animal. These meanings are abstract descriptions of all details of the object or the trajectory itself. Further analysis will
what is happening in the scenes infants observe; for example, an highlight different kind of paths, such as up and down, and animate or
object might be described as ‘starts to move by itself’ and ‘interacts inanimate. Herein lie the early conceptions that enable verb learning.
contingently with another object.’ Such meanings arise from attentive Similarly, LINK represents contingencies between events, another
analysis of spatial information and differ in format from perceptual or aspect of the world that infants learn from an early age. Differences in
motor schemas, in that they allow conscious thought. paths and their links define animals as different from inanimate
I suggested that the meanings created by perceptual meaning objects. From such simple beginnings the concepts arise that under-
analysis are represented in the form of image-schemas [51], although pin the learning of the linguistic distinction between agents and
there are other formats that could serve the same functions – for patients.

infants create a sub-goal in the service of a main goal. Preverbal recall of the past
Willatts [11] used a novel two-step problem in which The first major body of evidence for an accessible conceptual
infants had to retrieve a hidden object from under a system in the first year of life was obtained from deferred
cover that was beyond their reach at the far end of a imitation studies [12]. In this technique an event is modeled
pullable cloth. He found that the ability to construct a while infants watch, and after a delay they are given the
plan to solve this means–end problem becomes possible objects used in the modeling and encouraged to imitate what
between 8 and 9 months of age. Before that time, he they observed. Before modeling, as a baseline condition
suggested, infants learn various sensorimotor routines against which to compare post-modeling performance, the
about how to uncover objects and how to obtain out-of- objects are given to the infants to see what they spon-
reach objects by pulling, but have not yet developed the taneously do with them. The rationale behind the technique
conceptual processes required to work out a complex is that to reenact an event requires the same kind of
goal path by thought alone. conceptualization and retrieval as to retell it [13].
The technique has uncovered the rudiments of declara-
tive memory in infants as young as 6 months, showing
that they can reproduce actions demonstrated for them
Box 2. Generalized imitation: a method to test preverbal after a delay of 24 h [14]. By 9 months recall memory is
inductive inference robust, and infants can reproduce not just single actions
Inductive inference sounds like a sophisticated accomplishment, but but event sequences after delays of a month [15]. An
it is basically just a process of generalization. In the generalized example from Bauer et al. [16] of a 9-month-old reproduc-
imitation technique an event is modeled but instead of providing the ing a novel two-step sequence after a delay of a month is
same props used for modeling, different objects are provided and shown in Figure 1. By 10 or 11 months, infants can recall
infants must choose between them for their imitations. (As in other
sequences after a delay of several months [15,17] and
imitation methods, a baseline period is given first to see what infants
spontaneously do with the various test items). A typical example single actions even up to a year later [18]. Recall after
consists of modeling a dog being given a drink from a cup. Then the weeks or months indicates that memory consolidation
infant is given the cup, but instead of the same dog, a different dog, processes are at work [19], which is essential to the
or a cat or bird, is provided, along with an exemplar from another development of an accessible conceptual system.
category, such as an airplane. Which of these items infants choose (if
any) for their imitations gives an indication of the breadth of the
That deferred imitation requires declarative knowledge
concept under study – that is, how far infants have generalized their and cannot be accomplished on procedural grounds alone
observations. Imitating a complex event (as opposed to echoing a (as happens, for example, in motor learning or in
sound or perhaps mimicking clapping hands) requires conceptual repetition priming) is shown by the fact that it can occur
understanding, as amply shown by Piaget [54]. after a delay following single-trial observational learning
It was recently suggested that imitation does not require
conceptual understanding because toddlers do not understand
without any opportunity to practise the response [12].
models as symbols for real objects [55]. However, understanding Older children can also verbalize the events they acted out
symbols is not necessary for imitation; just as for pretend play, which before they learned to talk [20]. A final piece of evidence is
occurs throughout the second year, all that is required is for the that whereas infants can act out events they have
modeled event to bring to mind the relevant concepts. That this
observed in the past, adult amnesic patients, who are
happens is shown by toddlers’ scale errors, such as trying to sit in a
little model chair [56]. Being reminded of what a model represents is capable of being primed by previous experience but
only the first part of symbolic understanding; it also requires incapable of recall, cannot [21].
representational insight (i.e. understanding reference), a process
that is still incomplete in the second year [57]. Imitation occurs
somewhat earlier than pretend play, presumably because the Some preverbal concepts
modeling provides reminder cues and relieves the infant of the
Given that by the middle of the first year infants have
need to plan an event.
begun to form accessible conceptual representations that
www.sciencedirect.com
510 Review TRENDS in Cognitive Sciences Vol.8 No.11 November 2004

which are perceptually highly similar, but at the same age

fail to differentiate dogs from fish, which differ consider-
ably in appearance. It has been suggested that the models
of birds and planes might have been correctly categorized
because all the birds had eyes and all the planes had
wheels [10], but this does not explain why infants also
found it easy to categorize different kinds of wheeled
vehicles whilst failing to differentiate dogs from fish [23].
Furthermore, although 3-month-olds can perceptually
differentiate tables and chairs [6], 7–11-month-old infants
fail to categorize them on the object-examination test, and
Figure 1. Recall without words. A 9-month-old recalling a novel two-step event
sequence seen several times (but not practised) a month before. To turn on the
only at 11 months are kitchen utensils categorized as
light, the infant had to drop the car down the L-shaped chute, then push it with the different from furniture [24]. No perceptual explanation
rod so it rolled against the far wall, which turned on the light. Reproduced from [16] has been forthcoming for this pattern of successes and
with permission of Blackwell Publishing.
failures. (One can, however, use the object-examination
they use for recall and problem-solving, the question task to form perceptual categories in real time if enough
becomes what these early concepts are like. How do infants trials are given [25]).
characterize the world surrounding them? One test that Pauen [26] systematically varied the between-category
similarity of small-scale models of animals and furniture.
has been used to investigate early concepts is object-
In high-similarity conditions, each item had legs, curved
examination. In this test a few small-scale models of a
as well as rectilinear parts, and black-and-white dots that
category, such as animals, are given to the infant to examine
could be interpreted as eyes in the animals and knobs or
one at a time. Then a model from a new category, such as
decorations in the furniture. Even when there was high
vehicles, is given to the infant. Longer examination of the
between-category similarity among the items, 10- and
new instance indicates that the infant differentiates the two 11-month-olds categorized the items appropriately on the
categories. Although in many ways similar to categorization object-examination test and did so as much as when
studies using preferential-looking, the object-examination between-category similarity was low. As before, such data
test frequently gives different results, perhaps because patterns are difficult to explain on perceptual grounds
examination requires focused attention, which constitutes alone, again indicating that non-perceptual knowledge
only a portion of looking time [22]. was controlling performance.
The object-examination test has shown that by 7
months, infants differentiate models of animals, vehicles
and furniture from each other [23,24]. Because of the Preverbal inductive inference
within-class similarities in appearance of exemplars from The generalized imitation test (Box 2) has been used to
each of these three domains, this finding might be due study the inferences infants make, by asking how far
solely to perceptual categorization. However, 9-month- infants generalize properties associated with a concept.
olds also differentiate birds from airplanes (see Figure 2), Infants (like adults) must rely on inductive inference to

Figure 2. Categorizing animate and inanimate things. 9-month-olds categorize these model birds and airplanes as different kinds of things, despite their perceptual similarity.

www.sciencedirect.com
 Review TRENDS in Cognitive Sciences Vol.8 No.11 November 2004 511

extend their knowledge beyond their particular experi- and self-motion [31,32], between contingent and non-
ences. So, an infant might only have observed people and contingent interaction among objects [33], and between
perhaps a dog or a cat eat, but will infer from these biological and non-biological motion [34]. These global
observations that other animals eat too. A series of studies conceptualizations of actions are sufficient to form robust
showed that after seeing modeled behavior with one concepts of animate and inanimate things (Box 1), but we
animal, 9- to 14-month-olds would use any other animal still have relatively little information about when under-
to imitate animal-specific behaviors, such as drinking or standing of action becomes functional understanding, as
sleeping, but not a non-animal [27–29]. Similarly, they opposed to mere motion understanding.
would use any vehicle to imitate keying or giving a ride. Between 6 and 12 months infants discriminate goal-
By contrast, they typically refused to imitate modeled directed behavior from non-purposeful behavior whether
actions that were inappropriate to a kind, such as putting carried out by computer-generated shapes or by real
a vehicle to bed or giving it a drink [27]. (Giving the people [35–38]. They also parse continuous actions into
infants a choice might be crucial for this result; if an units organized around goals [39]. This understanding of
inappropriate action only is modeled, they might be actions, however, can still fairly be said to be more
induced to copy it). These results confirm not only that primitive than functional understanding. The latter
infants have developed global concepts, such as animal requires coordinating knowledge of specific object kinds
and vehicle, but also that they use these broad concepts to with specific actions. However, at least some detailed
limit their generalizations (some of which might later understanding begins in this period, as shown by the
need correction, such as that fish drink or that airplanes preverbal problem-solving abilities discussed earlier [11].
are opened with keys). At the same time, infants are
sensitive to the difference between domain-specific proper- Differentiation of preverbal global concepts
ties, such as eating and sleeping, and domain-general The results described above show that infants have global
properties, such as going indoors or being washed. In the concepts of animals, vehicles, and furniture (and also
domain-general case they are willing to generalize across plants [28]). The question arises of whether they have any
domain boundaries, and thus exhibit quite different concepts that will more-or-less match the first nouns they
behavior than in the domain-specific case [28]. learn. We know that 1-year-old infants see the difference
By narrowing the available choices to a single domain, between dogs and cats and between tables and chairs, but
the technique can also be used to specify more exactly how do they have any knowledge of the behavioral or func-
infants conceive of various animals and vehicles. When tional differences that make these items conceptually
14-month-olds were shown events such as giving a dog a different? This issue has had little exploration, but one
drink from a cup and then given the cup along with a series of studies using generalized imitation [28,40] found
different dog and either a cat, an unfamiliar mammal, or a little evidence for ‘basic-level’ concepts (i.e. those that
bird, they were indiscriminate in their generalizations match commonly used nouns) before the middle of the
when mammals were substituted – they chose a cat or second year. For example, at 14 months infants chose a
unfamiliar mammal for their imitations as often as the pan as often as a cup to imitate giving a doll a drink, a bird
other dog [28]. They were less likely to choose a bird. It is as often as a dog to imitate a dog chewing on a bone, and a
as if they were saying, ‘I saw you give a land animal a wrench as often as a hammer to imitate hammering
drink’. However, they often used the bird as a second (Figure 3). By 19–20 months imitation was largely correct
choice, as if they were saying, ‘I saw you give a land animal
a drink, but birds drink too’. By contrast, the infants were
100
more selective in their imitations with vehicles. They were
more likely to match the ‘basic-level’ category of the Appropriate
modeled item (car or motorcycle) and less likely to Inappropriate
80
generalize from land vehicles to airplanes. It appears
% of actions performed

that infants (in the urban culture under study) categorize

animals mainly on the basis of their behavior but are 60
somewhat more advanced in their categorization of
vehicles, using both function (or location?) and
40
appearance.
Even for vehicles, however, the data suggest that
conceptualizing objects initially depends on conceptualiz- 20
ing what they do. This phenomenon might depend on
infants’ attention to motion. One study found that infants
at five-and-a-half months were more likely to process and 0
Natural Artifact Natural Artifact
remember information about what people around them properties properties properties properties
were doing than what they or the objects being used
looked like [30]. In these experiments infants watched 14 months 20 months
videos that showed actions such as brushing teeth or hair. TRENDS in Cognitive Sciences
Neither the objects nor the faces of the actors were
Figure 3. Generalizing ‘basic-level’ properties by 14- and 20-month olds. Household
encoded as well as the activities. Young infants are artifacts are differentiated before animals and plants. Figure adapted from [28], with
also sensitive to the difference between caused motion permission of Elsevier.

www.sciencedirect.com
512 Review TRENDS in Cognitive Sciences Vol.8 No.11 November 2004

for a variety of household artifacts and for vehicles, but Infants thus demonstrated concepts of tight and loose fit
not until 24 months for animals and plants. Thus, we see a per se (i.e. abstracted away from other information),
narrowing of global concepts to something more refined at showing their readiness to learn a language such as
about the time that the noun explosion in language takes Korean, which includes degree of fit as a component of its
place. At 20 months, concepts of natural kinds lag behind terms for containment. Interestingly, Korean adults
artifacts (Figure 3). A likely reason for the different rate of showed the same sensitivity but English-speaking adults
differentiation between the animal and artifact domains is did not, indicating one of the ways that language can affect
the difference in daily experience and opportunities to habits of thought [47].
observe crucial differences between what objects do, or is
done to them.
The emphasis on actions in differentiating objects does Conclusions: mapping language onto preverbal
not necessarily mean that the actions themselves are concepts
precisely understood. Functional understanding can be This summary of preverbal thought has shown that the
quite general for long periods. However, Booth and conceptual basis necessary to understand simple language
Waxman [41] showed that 14-month-olds were more likely develops over the course of the first year. During this time
to categorize novel objects if shown a function for them at infants develop extensive concepts of objects such as
the same time. A study presented at the International animals, actions such as drinking, and spatial relations
Conference on Infant Studies (Träuble and Pauen, May such as containment. They can put these together in a
2004) found that 11-month-old infants categorized novel form sufficient to recall the past, solve simple problems,
objects on the basis of their overall perceptual similarity, and make inductive inferences. At the same time, their
without paying much attention to what their parts look concepts are often more general than the words used in
like. But if some parts were shown to produce interesting daily speech. This mismatch between concept and word
results, the infants then switched to categorizing the leads to the familiar phenomenon of overextension, in
objects on the basis of their parts instead of overall which early words have a broader extension for children
similarity. Even infants as young as 5 months can be
than they do for adults. Even at age 2, as many as 30% of
primed to attend to surface details if paired with
common nouns are still understood too broadly, for
functional information [42].
example, understanding (and using) the word dog to
include foxes, and the word cake to include pies [48].
Relational concepts
A similar phenomenon occurs with verbs. For example,
Infants also develop a number of spatial relational
young Korean learners use a single verb meaning un-fit
concepts over the course of the first year, such as
for any act of separation, whereas Korean adults make
containment. Baillargeon and her colleagues have shown
several finer distinctions, and young English learners use
that containment starts out as a global, relatively
the prepositions off and out to convey a similarly wide
undifferentiated concept that gradually becomes refined
variety of acts of separation [49].
[43]. For example, the concept of containment does not
An interesting question is whether the refinement in
initially take into account variables of the height or width
of a container with respect to the object being contained, conceptualization that is taking place in the first half of
but does so by 7–8 months [44]. In a similar fashion, the second year leads to the noun explosion or, conversely,
infants gradually acquire an abstract notion of above and whether learning nouns helps the refining process (see
below [45]. Even 3-month-olds who are habituated to a also Box 3). Hearing different words consistently used for,
picture of a figure above a line will dishabituate if the say, two animals that the child has interpreted as minor
figure is moved to below the line. But they will not perceptual variants, should lead the child to pay attention
dishabituate if a different figure is used in the test until to the differences. On the other hand, naming alone does
6–7 months. At first these spatial relations appear to be not facilitate categorization of novel objects in 14-month-
perceptually bound to the objects instantiating them. To olds unless accompanied by functional information,
go beyond this and abstract aboveness itself from the whereas by 18 months, labels alone do so [41]. Findings
rest of the display requires further analysis beyond what of this kind suggest that before language becomes well-
the perceptual system normally provides, suggesting established, conceptual differentiation plays a larger role
that it is an achievement of perceptual meaning analysis
(see Box 1). Box 3. Questions for future research
In related work on containment, it has been shown that † What other kinds of mental problem-solving do infants engage in
5-month-olds are sensitive to change in an object from towards the end of the first year?
tight to loose containment (or vice versa) [46]. By 9 months † What is the exact role of functional understanding in infants’
infants categorize containment as tight- or loose-fitting in differentiation of global concepts?
† Is the lag in conceptual differentiation of animals and plants
an abstract fashion. For example, after seeing several very
compared with artifacts due only to the difference in daily
different looking objects put into tight containment (such experience, or is there some other factor that makes natural kinds
as a book put into a slipcase and a cork into a bottle), more difficult to learn (such as lack of functional information)?
infants from both English-speaking and Korean-speaking † Is it the usual case that before language becomes well-established
homes dishabituated when shown still another object put concept development plays a larger role in word learning than word
learning plays in concept development, and if so, does this change
into loose containment, such as a pencil put into a pencil
towards the end of the second year?
cup (and vice versa from loose to tight containment) [47].
www.sciencedirect.com
 Review TRENDS in Cognitive Sciences Vol.8 No.11 November 2004 513

in word learning than word learning plays in conceptual 27 Mandler, J.M. and McDonough, L. (1996) Drinking and driving don’t
differentiation. mix: inductive generalization in infancy. Cognition 59, 307–335
28 Mandler, J.M. and McDonough, L. (1998) Studies in inductive
inference in infancy. Cogn. Psychol. 37, 60–96
References 29 McDonough, L. and Mandler, J.M. (1998) Inductive generalization in
1 Piaget, J. (1952) The Origins of Intelligence in the Child, International 9- and 11-month olds. Dev. Sci. 1, 227–232
Universities Press 30 Bahrick, L.E. et al. (2002) Attention and memory for faces and actions
2 Schacter, D.L. et al. (2000) Memory systems of 1999. In The Oxford
in infancy: the salience of actions over faces in dynamic events. Child
Handbook of Memory (Tulving, E. and Craik, F.I.M., eds), Oxford
Dev. 73, 1629–1643
University Press
31 Leslie, A.M. (1988) The necessity of illusion: perception and thought in
3 Squire, L.R. and Knowlton, B.J. (1995) Learning about categories in
infancy. In Thought without Language (Weiskrantz, L. ed.), Clarendon
the absence of memory. Proc. Natl. Acad. Sci. U. S. A. 92, 12470–12474
Press
4 Baillargeon, R. (2001) Infants’ physical knowledge: of acquired
expectations and core principles. In Language, brain, and cognitive 32 Poulin-Dubois, D. et al. (1996) Infants’ concept of animacy. Cogn. Dev.
development: Essays in honor of Jacques Mehler (Dupoux, E. ed.), MIT 11, 19–36
Press 33 Frye, D. et al. (1983) Object-person discrimination and communication
5 Quinn, P.C. and Eimas, P.D. (1996) Perceptual cues that permit at 3 and 10 months. Dev. Psychol. 19, 303–309
categorical differentiation of animal species by infants. J. Exp. Child 34 Arterberry, M.E. and Bornstein, M.H. (2001) Three-month-old infants’
Psychol. 63, 189–211 categorization of animals and vehicles based on static and dynamic
6 Behl-Chadha, G. (1996) Basic-level and superordinate-like categorical attributes. J. Exp. Child Psychol. 80, 333–346
representations in early infancy. Cognition 60, 105–141 35 Csibra, G. et al. (1999) Goal attribution without agency cues: the
7 Moscovitch, M. et al. (1994) Memory without conscious recollection: a perception of ‘pure reason’ in infancy. Cognition 72, 237–267
tutorial review from a neuropsychological perspective. In Attention 36 Csibra, G. (2003) One-year-old infants use teleological representation
and Performance XV: Conscious and Nonconscious Information of actions productively. Cogn. Sci. 27, 111–133
Processing (Umiltá, C. and Moscovitch, M., eds), MIT Press 37 Woodward, A.L. and Sommerville, J.A. (2000) Twelve-month-old
8 Reed, J.M. et al. (1999) Learning about categories that are defined by infants interpret action in context. Psychol. Sci. 11, 73–77
object-like stimuli despite impaired declarative memory. Behav. 38 Sodian, B. et al. (2004) Do infants apply the principle of rational action
Neurosci. 113, 411–419 to human agents? Infant Behav. Dev 27, 31–41
9 Tillmann, B. et al. (2000) Implicit learning of tonality: a self- 39 Baldwin, D.A. et al. (2001) Infants parse dynamic action. Child Dev.
organizing approach. Psychol. Rev. 107, 885–913 72, 708–717
10 Mareschal, D. and Quinn, P.C. (2001) Categorization in infancy. 40 Mandler, J.M. and McDonough, L. (2000) Advancing downward to the
Trends Cogn. Sci. 5, 443–450 basic level. J. Cogn. Dev. 1, 379–403
11 Willatts, P. (1997) Beyond the ‘couch potato’ infant: how infants use 41 Booth, A.E. and Waxman, S. (2002) Object names and object functions
their knowledge to regulate action, solve problems, and achieve goals. serves as cues to categories for infants. Dev. Psychol. 38, 948–957
In Infant Development: Recent Advances (Bremner, G. et al., eds), 42 Wilcox, T. and Chapa, C. (2004) Priming infants to attend to color and
Psychology Press
pattern information in an individuation task. Cognition 90, 265–302
12 Meltzoff, A.N. (1988) Infant imitation and memory: nine-month-olds
43 Baillargeon, R. and Wang, S. (2002) Event categorization in infancy.
in immediate and deferred tests. Child Dev. 59, 217–225
Trends Cogn. Sci. 6, 85–93
13 Mandler, J.M. (1990) Recall of events by preverbal children. In The
44 Aguiar, A. and Baillargeon, R. (1998) Eight-and-a-half-month-old
Development and Neural Bases of Higher Cognitive Functions
infants’ reasoning about containment events. Child Dev. 69, 636–653
(Diamond, A. ed.), New York Academy of Sciences
14 Collie, R. and Hayne, H. (1999) Deferred imitation by 6- and 9-month- 45 Quinn, P.C. (2003) Concepts are not just for objects: categorization of
old infants: more evidence for declarative memory. Dev. Psychobiol. spatial relation information by infants. In Categories and Concepts in
35, 83–90 Early Development (Rakison, D.H. et al., eds), Oxford University Press
15 Carver, L.J. and Bauer, P.J. (2001) The dawning of a past: the 46 Hespos, S.J. and Spelke, E.S. (2004) Conceptual precursors to
emergence of long-term explicit memory in infancy. J. Exp. Psychol. language. Nature 430, 453–456
Gen. 130, 726–745 47 McDonough, L. et al. (2003) Understanding spatial relations: flexible
16 Bauer, P.J. et al. (2003) Development in long–term explicit memory infants, lexical adults. Cogn. Psychol. 46, 229–259
late in the first year of life: behavioral and electrophysiological indices. 48 McDonough, L. (2002) Early concepts and early language acquisition:
Psychol. Sci. 14, 629–635 what does similarity have to do with either? In Representation,
17 Mandler, J.M. and McDonough, L. (1995) Long-term recall in infancy. Memory, and Development: Essays in Honor of Jean Mandler (Stein,
J. Exp. Child Psychol. 59, 457–474 N.L. et al., eds), Erlbaum
18 McDonough, L. and Mandler, J.M. (1994) Very long-term recall in 49 Bowerman, M. and Choi, S. (2001) Shaping meanings for language:
infants: infantile amnesia reconsidered. Memory 2, 339–352 universal and language-specific in the acquisition of spatial semantic
19 Zola, S.M. and Squire, L.R. (2000) The medial temporal lobe and the categories. In Language Acquisition and Conceptual development
hippocampus. In The Oxford Handbook of Memory (Tulving, E. and (Bowerman, M. and Levinson, S.C., eds), Cambridge University Press
Craik, F.I.M., eds), Oxford University Press 50 Karmiloff-Smith, A. (1992) Beyond modularity: A Developmental
20 Bauer, P.J. et al. (2002) Making the past present: later verbal Perspective on Cognitive Science, MIT Press
accessibility of early memories. J. Cogn. Dev 3, 21–47 51 Mandler, J.M. (2004) The Foundations of Mind: Origins of Conceptual
21 McDonough et al. (1995) The deferred imitation task as a nonverbal Thought, Oxford University Press
measure of declarative memory. Proc. National Acad. Sci. U. S. A. 92, 52 Barsalou, L.W. et al. (2003) Grounding conceptual knowledge in
7580–7584 modality-specific systems. Trends Cogn. Sci. 7, 84–91
22 Lansink, J.M. et al. (2000) The distribution of infant attention during 53 Lakoff, G. (1987) Women, Fire, and Dangerous Things: What
object examination. Dev. Sci. 3, 163–170
Categories Reveal about the Mind, University of Chicago Press
23 Mandler, J.M. and McDonough, L. (1993) Concept formation in
54 Piaget, J. (1951) Play, Dreams and Imitation in Childhood, Routledge
infancy. Cogn. Dev. 8, 291–318
& Kegan Paul
24 Mandler, J.M. and McDonough, L. (1998) On developing a knowledge
55 Younger, B.A. and Johnson, K.E. (2004) Infants’ comprehension of toy
base in infancy. Dev. Psychol. 34, 1274–1288
replicas as symbols for real objects. Cogn. Psychol. 42, 207–242
25 Younger, B.A. and Furrer, S.D. (2003) A comparison of visual
56 DeLoache, J.S. et al. (2004) Scale-errors offer evidence for a
familiarization and object-examining measures of categorization in
9-month-old infants. Infancy 4, 327–348 perception–action dissociation early in life. Science 304, 1027–1029
26 Pauen, S. (2002) Evidence for knowledge-based category discrimin- 57 DeLoache, J.S. (2004) Becoming symbol-minded. Trends Cogn. Sci. 8,
ation in infancy. Child Dev. 73, 1016–1033 66–70

www.sciencedirect.com