Infant categorization as a dynamic process linked to memory

Recency effects are well documented in the adult and infant literature: recognition and recall memory are better for recently occurring events. We explore recency effects in infant categorization, which does not merely involve memory for individual items, but the formation of abstract category representations. We present a computational model of infant categorization that simulates category learning in 10-month-olds. The model predicts that recency effects outweigh previously reported order effects for the same stimuli. According to the model, infant behaviour at test should depend mainly on the identity of the most recent training item. We evaluate these predictions in a series of experiments with 10-month-old infants. Our results show that infant behaviour confirms the model's prediction. In particular, at test infants exhibited a preference for a category outlier over the category average only if the final training item had been close to the average, rather than distant from it. Our results are consistent with a view of categorization as a highly dynamic process where the end result of category learning is not the overall average of all stimuli encountered, but rather a fluid representation that moves depending on moment-to-moment novelty. We argue that this is a desirable property of a flexible cognitive system that adapts rapidly to different contexts.

Underspecification in toddlers' and adults' lexical representations

Recent research has shown that toddlers' lexical representations are phonologically detailed, quantitatively much like those of adults. Studies in this article explore whether toddlers' and adults' lexical representations are qualitatively similar. Psycholinguistic claims (Lahiri & Marslen-Wilson, 1991; Lahiri & Reetz, 2002, 2010) based on underspecification (Kiparsky, 1982 et seq.) predict asymmetrical judgments in lexical processing tasks; these have been supported in some psycholinguistic research showing that participants are more sensitive to noncoronal-to-coronal (pop → top) than to coronal-to-noncoronal (top → pop) changes or mispronunciations. Three experiments using on-line visual world procedures showed that 19-month-olds and adults displayed sensitivities to both noncoronal-to-coronal and coronal-to-noncoronal mispronunciations of familiar words. No hints of any asymmetries were observed for either age group. There thus appears to be considerable developmental continuity in the nature of early and mature lexical representations. Discrepancies between the current findings and those of previous studies appear to be due to methodological differences that cast doubt on the validity of claims of psycholinguistic support for lexical underspecification.

Do infants discriminate non-linguistic vocal expressions of positive emotions?

Adults are highly proficient in understanding emotional signals from both facial and vocal cues, including when communicating across cultural boundaries. However, the developmental origin of this ability is poorly understood, and in particular, little is known about the ontogeny of differentiation of signals with the same valence. The studies reported here employed a habituation paradigm to test whether preverbal infants discriminate between non-linguistic vocal expressions of relief and triumph. Infants as young as 6 months who had habituated to relief or triumph showed significant discrimination of relief and triumph tokens at test (i.e. greater recovery to the unhabituated stimulus type), when exposed to tokens from a single individual (Study 1). Infants habituated to expressions from multiple individuals showed less consistent discrimination in that consistent discrimination was only found when infants were habituated to relief tokens (Study 2). Further, infants tested with tokens from individuals from different cultures showed dishabituation only when habituated to relief tokens and only at 10-12 months (Study 3). These findings suggest that discrimination between positive emotional expressions develops early and is modulated by learning. Further, infants' categorical representations of emotional expressions, like those of speech sounds, are influenced by speaker-specific information.

TRACX2: a connectionist autoencoder using graded chunks to model infant visual statistical learning

Even newborn infants are able to extract structure from a stream of sensory inputs; yet how this is achieved remains largely a mystery. We present a connectionist autoencoder model, TRACX2, that learns to extract sequence structure by gradually constructing chunks, storing these chunks in a distributed manner across its synaptic weights and recognizing these chunks when they re-occur in the input stream. Chunks are graded rather than all-or-nothing in nature. As chunks are learnt their component parts become more and more tightly bound together. TRACX2 successfully models the data from five experiments from the infant visual statistical learning literature, including tasks involving forward and backward transitional probabilities, low-salience embedded chunk items, part-sequences and illusory items. The model also captures performance differences across ages through the tuning of a single-learning rate parameter. These results suggest that infant statistical learning is underpinned by the same domain-general learning mechanism that operates in auditory statistical learning and, potentially, in adult artificial grammar learning.This article is part of the themed issue 'New frontiers for statistical learning in the cognitive sciences'.

When Looks Are Everything

The goal of this research was to examine mechanisms underlying early induction—specifically, the relation between induction and categorization. Some researchers argue that even early in development, induction is based on category-membership information, whereas others argue that early induction is based primarily on similarity. Children 4 and 5 years of age participated in two types of tasks: categorization and induction. Both tasks were performed with artificial animal-like categories in which appearance was pitted against category membership. Although the children readily acquired category-membership information and subsequently used this information in categorization tasks, they ignored category membership during the induction task, relying instead on the appearance of items. These results support the idea that early in development, induction is similarity based.

Conceptual influences on induction: A case for a late onset

This research examines the mechanism of early induction, the development of induction, and the ways attentional and conceptual factors contribute to induction across development. Different theoretical views offer different answers to these questions. Six experiments with 4- and 5-year-olds, 7-year-olds and adults (N=208) test these competing theories by teaching categories for which category membership and perceptual similarity are in conflict, and varying orthogonally conceptual and attentional factors that may potentially affect inductive inference. The results suggest that early induction is similarity-based; conceptual information plays a negligible role in early induction, but its role increases gradually, with the 7-year-olds being a transitional group. And finally, there is substantial contribution of attention to the development of induction: only adults, but not children, could perform category-based induction without attentional support. Therefore, category-based induction exhibits protracted development, with attentional factors contributing early in development and conceptual factors contributing later in development. These results are discussed in relation to existing theories of development of inductive inference and broader theoretical views on cognitive development.

Infant categorization

In this article, we review the principal findings on infant categorization from the last 30 years. The review focuses on behaviorally based experiments with visual preference, habituation, object examining, sequential touching, and inductive generalization procedures. We propose that although this research has helped to elucidate the 'what' and 'when' of infant categorization, it has failed to clarify the mechanisms that underpin this behavior and the development of concepts. We outline a number of reasons for why the field has failed in this regard, most notably because of the context-specific nature of infant categorization and a lack of ground rules in interpreting data. We conclude by suggesting that one remedy for this issue is for infant categorization researchers to adopt more of an interdisciplinary approach by incorporating imaging and computational methods into their current methodological arsenal. WIREs Cogn Sci 2010 1 894-905 For further resources related to this article, please visit the WIREs website.

Overcoming catastrophic interference in connectionist networks using Gram-Schmidt orthogonalization

Connectionist models of memory storage have been studied for many years, and aim to provide insight into potential mechanisms of memory storage by the brain. A problem faced by these systems is that as the number of items to be stored increases across a finite set of neurons/synapses, the cumulative changes in synaptic weight eventually lead to a sudden and dramatic loss of the stored information (catastrophic interference, CI) as the previous changes in synaptic weight are effectively lost. This effect does not occur in the brain, where information loss is gradual. Various attempts have been made to overcome the effects of CI, but these generally use schemes that impose restrictions on the system or its inputs rather than allowing the system to intrinsically cope with increasing storage demands. We show here that catastrophic interference occurs as a result of interference among patterns that lead to catastrophic effects when the number of patterns stored exceeds a critical limit. However, when Gram-Schmidt orthogonalization is combined with the Hebb-Hopfield model, the model attains the ability to eliminate CI. This approach differs from previous orthogonalisation schemes used in connectionist networks which essentially reflect sparse coding of the input. Here CI is avoided in a network of a fixed size without setting limits on the rate or number of patterns encoded, and without separating encoding and retrieval, thus offering the advantage of allowing associations between incoming and stored patterns. PACS Nos.: 87.10.+e, 87.18.Bb, 87.18.Sn, 87.19.La.

From perceptual to language-mediated categorization

From at least two months onwards, infants can form perceptual categories. During the first year of life, object knowledge develops from the ability to represent individual object features to representing correlations between attributes and to integrate information from different sources. At the end of the first year, these representations are shaped by labels, opening the way to conceptual knowledge. Here, we review the development of object knowledge and object categorization over the first year of life. We then present an artificial neural network model that models the transition from early perceptual categorization to categories mediated by labels. The model informs a current debate on the role of labels in object categorization by suggesting that although labels do not act as object features they nevertheless affect perceived similarity of perceptually distinct objects sharing the same label. The model presents the first step of an integrated account from early perceptual categorization to language-based concept learning.

Visual recognition of age class and preference for infantile features: implications for species-specific vs universal cognitive traits in primates

Despite not knowing the exact age of individuals, humans can estimate their rough age using age-related physical features. Nonhuman primates show some age-related physical features; however, the cognitive traits underlying their recognition of age class have not been revealed. Here, we tested the ability of two species of Old World monkey, Japanese macaques (JM) and Campbell's monkeys (CM), to spontaneously discriminate age classes using visual paired comparison (VPC) tasks based on the two distinct categories of infant and adult images. First, VPCs were conducted in JM subjects using conspecific JM stimuli. When analyzing the side of the first look, JM subjects significantly looked more often at novel images. Based on analyses of total looking durations, JM subjects looked at a novel infant image longer than they looked at a familiar adult image, suggesting the ability to spontaneously discriminate between the two age classes and a preference for infant over adult images. Next, VPCs were tested in CM subjects using heterospecific JM stimuli. CM subjects showed no difference in the side of their first look, but looked at infant JM images longer than they looked at adult images; the fact that CMs were totally naïve to JMs suggested that the attractiveness of infant images transcends species differences. This is the first report of visual age class recognition and a preference for infant over adult images in nonhuman primates. Our results suggest not only species-specific processing for age class recognition but also the evolutionary origins of the instinctive human perception of baby cuteness schema, proposed by the ethologist Konrad Lorenz.

Using saliency maps to separate competing processes in infant visual cognition

This article presents an eye-tracking study using a novel combination of visual saliency maps and "area-of-interest" analyses to explore online feature extraction during category learning in infants. Category learning in 12-month-olds (N = 22) involved a transition from looking at high-saliency image regions to looking at more informative, highly variable object parts. In contrast, 4-month-olds (N = 27) exhibited a different pattern displaying a similar decreasing impact of saliency accompanied by a steady focus on the object's center, indicating that targeted feature extraction during category learning develops across the 1st year of life. These results illustrate how the effects of lower and higher level processes may be disentangled using a combined saliency map and area-of-interest analysis.

The influence of social information on children's statistical and causal inferences

Constructivist accounts of learning posit that causal inference is a child-driven process. Recent interpretations of such accounts also suggest that the process children use for causal learning is rational: Children interpret and learn from new evidence in light of their existing beliefs. We argue that such mechanisms are also driven by informative social cues and suggest ways in which such information influences both preschoolers' and infants' inferences. In doing so, we argue that a rational constructivist account should not only focus on describing the child's internal cognitive mechanisms for learning but also on how social information affects the process of learning.

Contributions of attentional style and previous experience to 4-month-old infants' categorization

We examined how infants' categorization is jointly influenced by previous experience and how much they shift their gaze back-and-forth between stimuli. Extending previous findings reported by Kovack-Lesh, Horst, and Oakes (2008), we found that 4-month-old infants' (N = 122) learning of the exclusive category of cats was related to whether they had cats at home and how much they shifted attention between two available stimuli during familiarization. Individual differences in attention assessed in an unrelated task were not related to their categorization. Thus, infants' learning is multiply influenced by past experience and on-line attentional style.

Correspondences between what infants see and know about causal and self-propelled motion

The associative learning account of how infants identify human motion rests on the assumption that this knowledge is derived from statistical regularities seen in the world. Yet, no catalog exists of what visual input infants receive of human motion, and of causal and self-propelled motion in particular. In this manuscript, we demonstrate that the frequency with which causal agency and self-propelled motion appear in the visual environment predicts infants' understanding of these motions. In an observational study, an infant wearing a head-mounted camera saw people act as agents in causal events three times more often than he saw people engaged in self-propelled motion. Subsequent experiments with the habituation paradigm revealed that infants begin to generalize self-propulsion to agents in causal events between 10 and 14 months of age. However, infants cannot generalize causal agency to a self-propelled object at 14 or 18 months unless the object exhibits additional cues to animacy. The results are discussed within a domain-general framework of learning about human action.

Infants and toddlers show enlarged visual sensitivity to nonaccidental compared with metric shape changes

Some shape changes are more important for object perception than others. We used a habituation paradigm to measure visual sensitivity to a nonaccidental shape change—that is, the transformation of a trapezium into a triangle and vice versa—and a metric shape change—that is, changing the aspect ratio of the shapes. Our data show that an enhanced perceptual sensitivity to nonaccidental changes is already present in infancy and remains stable into toddlerhood. We have thus established an example of how early visual perception deviates from the null hypothesis of representing similarity as a function of physical overlap between shapes, and does so in agreement with more cognitive, categorical demands.

From Perceptual Categories to Concepts: What Develops?

People are remarkably smart: they use language, possess complex motor skills, make non-trivial inferences, develop and use scientific theories, make laws, and adapt to complex dynamic environments. Much of this knowledge requires concepts and this paper focuses on how people acquire concepts. It is argued that conceptual development progresses from simple perceptual grouping to highly abstract scientific concepts. This proposal of conceptual development has four parts. First, it is argued that categories in the world have different structure. Second, there might be different learning systems (sub-served by different brain mechanisms) that evolved to learn categories of differing structures. Third, these systems exhibit differential maturational course, which affects how categories of different structures are learned in the course of development. And finally, an interaction of these components may result in the developmental transition from perceptual groupings to more abstract concepts. This paper reviews a large body of empirical evidence supporting this proposal.

Categorical structure among shared features in networks of early-learned nouns

The shared features that characterize the noun categories that young children learn first are a formative basis of the human category system. To investigate the potential categorical information contained in the features of early-learned nouns, we examine the graph-theoretic properties of noun-feature networks. The networks are built from the overlap of words normatively acquired by children prior to 2(1/2) years of age and perceptual and conceptual (functional) features acquired from adult feature generation norms. The resulting networks have small-world structure, indicative of a high degree of feature overlap in local clusters. However, perceptual features--due to their abundance and redundancy--generate networks more robust to feature omissions, while conceptual features are more discriminating and, per feature, offer more categorical information than perceptual features. Using a network specific cluster identification algorithm (the clique percolation method) we also show that shared features among these early-learned nouns create higher-order groupings common to adult taxonomic designations. Again, perceptual and conceptual features play distinct roles among different categories, typically with perceptual features being more inclusive and conceptual features being more exclusive of category memberships. The results offer new and testable hypotheses about the role of shared features in human category knowledge.

Developmental changes in visual object recognition between 18 and 24 months of age

Two experiments examined developmental changes in children's visual recognition of common objects during the period of 18 to 24 months. Experiment 1 examined children's ability to recognize common category instances that presented three different kinds of information: (1) richly detailed and prototypical instances that presented both local and global shape information, color, textural and featural information, (2) the same rich and prototypical shapes but no color, texture or surface featural information, or (3) that presented only abstract and global representations of object shape in terms of geometric volumes. Significant developmental differences were observed only for the abstract shape representations in terms of geometric volumes, the kind of shape representation that has been hypothesized to underlie mature object recognition. Further, these differences were strongly linked in individual children to the number of object names in their productive vocabulary. Experiment 2 replicated these results and showed further that the less advanced children's object recognition was based on the piecemeal use of individual features and parts, rather than overall shape. The results provide further evidence for significant and rapid developmental changes in object recognition during the same period children first learn object names. The implications of the results for theories of visual object recognition, the relation of object recognition to category learning, and underlying developmental processes are discussed.

Toddlers can adaptively change how they categorize: same objects, same session, two different categorical distinctions

Two experiments demonstrate that 14- to 18-month-old toddlers can adaptively change how they categorize a set of objects within a single session, and that this ability is related to vocabulary size. In both experiments, toddlers were presented with a sequential touching task with objects that could be categorized either according to some perceptually salient dimension corresponding to a taxonomic distinction (e.g. animals vs. vehicles) or to some less obvious dimension (e.g. rigid vs. deformable). In each experiment, children with larger productive vocabularies responded to both dimensions, showing evidence of sensitivity to each way of categorizing the items. Children with smaller productive vocabularies attended only to the taxonomically related categorical grouping. These experiments confirm that toddlers can adaptively shift the basis of their categorization and highlight the dynamic interaction between the child and the current task in early categorization.

Ten-month-olds' selective use of visual dimensions in category learning

There is now general consensus that infants can use several different visual properties as the basis for categorization. Nonetheless, little is known about when and whether infants can be guided by contextual information to select the relevant properties from amongst those available to them. We show here that by 10 months of age infants can be biased, through observational learning, to use one or the other of two object properties for classification. Two groups of infants watched an actress classifying objects by either shape (the Shape group) or surface pattern (the Pattern group). When subsequently presented with two test trials which contradicted either one or the other of the classification rules, infants in the two groups looked longer to the classification event that was incompatible with the rule that group had been familiarized to. These results are discussed with reference to the development of selective feature processing in infancy.

The interplay between perceptual organization and categorization in the representation of complex visual patterns by young infants

The relation between perceptual organization and categorization processes in 3- and 4-month-olds was explored. The question was whether an invariant part abstracted during category learning could interfere with Gestalt organizational processes. A 2003 study by Quinn and Schyns had reported that an initial category familiarization experience in which infants were presented with visual patterns consisting of a pacman shape and a complex polygon could interfere with infants' subsequent good continuation-based parsing of a circle from visual patterns consisting of a circle and a complex polygon. However, an alternative noninterference explanation for the results was possible because the pacman had been presented with greater frequency and duration than had the circle. The current study repeated Quinn and Schyns's procedure but provided an equivalent number of familiarization trials and duration of study time for the infants to process the pacman during initial familiarization and the circle during subsequent familiarization. The results replicated the previous findings of Quinn and Schyns. The data are consistent with the interference account and suggest that a cognitive system of adaptable feature creation can take precedence over organizational principles with which a perceptual system comes pre-equipped.

How Do Simple Connectionist Networks Achieve a Shift From "Featural" to "Correlational" Processing in Categorization?

Three developmental connectionist models simulate a purported shift from "featural" to "correlational" processing in infant categorization (models: Gureckis & Love, 2004/this issue; Shultz & Cohen, 2004/this issue; Westermann & Mareschal, 2004/this issue; empirical data: Cohen & Arthur, 2003; Younger, 1985; Younger & Cohen, 1986). In this article, the way in which the models are able to simulate the behavioral data is revealed, and their respective theoretical commitments are evaluated. Together the models argue that the shift from featural to correlational processing in infant categorization might be illusory, as these models are able to replicate the key behavioral features while processing correlations right from the start. As such, they claim the behavioral description of a shift is not reflected at the level of mechanism.

Basic-level category discriminations by 7- and 9-month-olds in an object examination task

This study examines 7- and 9-month-olds' ability to categorize cats as separate from dogs, and dogs as separate from cats in an object examination task. In Experiment 1, 7- and 9-month-olds (N = 30) familiarized with toy cat replicas were found to form a category of cat that included novel cats but excluded a dog and an eagle. In Experiment 2, 7- and 9-month-olds (N = 30) familiarized with toy dog replicas were found to form a category of dog that included a novel dogs and a novel cat but excluded an eagle. These results mirror those of 3- to 4-month-olds tested with visual preference methods and stand in contrast to previously reported object examination results. Analyses of the distribution of features in the exemplars used to familiarize infants suggest that, like the 3- to 4-month-olds, the 7- and 9-month-olds in these studies form categories within the task, and on the basis of feature distributions.