Infants' sensitivity to correlations between static and dynamic features in a category context

The role of labels and motions in infant category learning

We investigated the impact of two highly salient transient features, labels and motions, on novel visual category learning in 10-month-old infants. In three eye-tracking experiments, infants were presented with exemplars from two novel categories either accompanied by category-specific labels, accompanied by category-specific motions, or in silence. Labels (Experiment 1) and motions (Experiment 2) were presented using a gaze-contingent design in which these transient features were triggered by infants' fixations. Gaze-contingent transient features, despite being redundant, had a strong impact on categorization. The results revealed that both labels and motions support infants' category formation. Furthermore, both labels and motions promoted similarity-focused exploration, whereas no such pattern was found when infants learned the categories in silence. Analyses of visual exploration patterns revealed that infants readily form expectations about motion properties of categories and that these expectations drive their looking behavior.

Second-Order Correlation Learning of Dynamic Stimuli: Evidence from Infants and Computational Modeling

We present two habituation experiments that examined 20- and 26-month-olds' ability to engage in second-order correlation learning for static and dynamic features, whereby learned associations between two pairs of features (e.g., P and Q, P and R) are generalized to the features that were not presented together (e.g., Q and R). We also present results from an associative learning mechanism that was implemented as an autoencoder parallel distributed processing (PDP) network in which second-order correlation learning is shown to be an emergent property of the dynamics of the network. The experiments and simulation demonstrate that 20- and 26-month-olds as well as neural networks are capable of second-order correlation learning in a category context for internal features of dynamic objects. However, the model predicts-and Experiment 3 demonstrates-that 20- and 26-month-olds are unable to encode second-order correlations in a noncategory context for dynamic objects with internal features. It is proposed that the ability to learn second-order correlations represents a powerful but as yet unexplored process for generalization in the first years of life.

Curiosity-based learning in infants: a neurocomputational approach

Infants are curious learners who drive their own cognitive development by imposing structure on their learning environment as they explore. Understanding the mechanisms by which infants structure their own learning is therefore critical to our understanding of development. Here we propose an explicit mechanism for intrinsically motivated information selection that maximizes learning. We first present a neurocomputational model of infant visual category learning, capturing existing empirical data on the role of environmental complexity on learning. Next we "set the model free", allowing it to select its own stimuli based on a formalization of curiosity and three alternative selection mechanisms. We demonstrate that maximal learning emerges when the model is able to maximize stimulus novelty relative to its internal states, depending on the interaction across learning between the structure of the environment and the plasticity in the learner itself. We discuss the implications of this new curiosity mechanism for both existing computational models of reinforcement learning and for our understanding of this fundamental mechanism in early development.

Red to green or fast to slow? Infants' visual working memory for "just salient differences"

In this study, 6-month-old infants' visual working memory for a static feature (color) and a dynamic feature (rotational motion) was compared. Comparing infants' use of different features can only be done properly if experimental manipulations to those features are equally salient (Kaldy & Blaser, 2009; Kaldy, Blaser, & Leslie, 2006). The interdimensional salience mapping method was used to find two objects that each were one Just Salient Difference from a common baseline object (N = 16). These calibrated stimuli were then used in a subsequent two-alternative forced-choice preferential looking memory test (N = 28). Results showed that infants noted the color change, but not the equally salient change in rotation speed.

Effects of categorical labels on similarity judgments: a critical analysis of similarity-based approaches

Our goal in the present study was to evaluate the claim that category labels affect children's judgments of visual similarity. We presented preschool children with discriminable and identical sets of animal pictures and asked them to make perceptual judgments in the presence or absence of labels. Our findings indicate that children who are asked to make perceptual judgments about identical items judge discriminable items less accurately when making subsequent similarity judgments. Thus, labels do not generally affect children's perceptual similarity judgments; rather, children's reliance on labels to make similarity judgments appears to be attributable to flaws in the methodological approaches used in prior studies. These results have implications for the role of perceptual and conceptual information in children's categorization and induction.

Preschool-age children and adults flexibly shift their preferences for auditory versus visual modalities but do not exhibit auditory dominance

The goal of this study was to evaluate the claim that young children display preferences for auditory stimuli over visual stimuli. This study was motivated by concerns that the visual stimuli employed in prior studies were considerably more complex and less distinctive than the competing auditory stimuli, resulting in an illusory preference for auditory cues. Across three experiments, preschool-age children and adults were trained to use paired audio-visual cues to predict the location of a target. At test, the cues were switched so that auditory cues indicated one location and visual cues indicated the opposite location. In contrast to prior studies, preschool-age children did not exhibit auditory dominance. Instead, children and adults flexibly shifted their preferences as a function of the degree of contrast within each modality, with high contrast leading to greater use.

Learning how actions function: the role of outcomes in infants' representation of events

Action is a fundamental component of object representations. However, little is known about how infants represent actions performed on objects. Across four experiments, we tested the hypothesis that at 10 months of age (N=80) infants represent the general ability of actions to produce outcomes (sounds). Experiments 1A and 1B showed that infants encode actions and associate actions and object appearances in events in which actions produced no sound outcomes. Experiment 2 showed that infants associate the presence or absence of outcomes with actions. Experiment 3 showed, in contrast, that infants did not associate the presence or absence of outcomes with object appearances. Together, these studies suggest that infants encode the outcome potential of specific actions. We discuss the implications of these findings for our understanding of the development of action representations.

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.

Specifying the role of function in infant categorization

Research demonstrates that object functions facilitate artifact categorization in infancy. To explicate the nature and magnitude of this effect, 16-month-olds participated in three studies. In Experiment 1, categorization was facilitated more by object functions than by distinctive motions, suggesting that the motion properties of function cannot fully explain its influence. In Experiment 2, infants failed to categorize when each category exemplar performed a different function, thus revealing the importance of shared functionality in facilitating categorization. In Experiment 3, infants were tested after each new exemplar was introduced. When object functions were provided during training, infants were more likely to appropriately extend the novel categories on the very first trial. This suggests that function reduces the need for exposure to multiple exemplars in forming categories. Together, these findings confirm the conceptual nature of the facilitative effect of function on early categorization.

Connecting cues: overlapping regularities support cue discovery in infancy

The present work examined the discovery of linguistic cues during a word segmentation task. Whereas previous studies have focused on sensitivity to individual cues, this study addresses how individual cues may be used to discover additional, correlated cues. Twenty-four 9-month-old infants were familiarized with a speech stream in which syllable-level transitional probabilities and an overlapping novel cue served as cues to word boundaries. Infants' behavior at test indicated that they were able to discover the novel cue. Additional experiments showed that infants did not have a preexisting preference for specific test items and that transitional probability information was necessary to acquire the novel cue. Results suggest one way learners can discover relevant linguistic structure amid the multiple overlapping properties of natural language.

Make the first move: How infants learn about self-propelled objects

In 3 experiments, the author investigated 16- to 20-month-old infants' attention to dynamic and static parts in learning about self-propelled objects. In Experiment 1, infants were habituated to simple noncausal events in which a geometric figure with a single moving part started to move without physical contact from an identical geometric figure that possessed a single static part. Infants were then tested with an event in which the parts of the objects were switched. In Experiments 2 and 3, infants were habituated and tested with identical events except that the part possessed by each object during habitation was switched relative to the first experiment. Results of the experiments revealed that 16-month-olds failed to encode the relation between an object's part and its onset of motion, 18-month-olds were unconstrained in the relations involving self-propulsion that they would encode, and 20-month-olds were constrained in the relations they would encode. The results are discussed with regard to the developmental trajectory of learning about motion properties and the mechanism involved in early concept acquisition.

A secret agent? How infants learn about the identity of objects in a causal scene

Four experiments examined the role of correlations between dynamic and static parts on 12- to 16-month-olds' ability to learn the identity of agents and recipients in a simple causal event. Infants were habituated to events in which objects with a dynamic or static part acted as an agent or a recipient and then were tested with an event in which the part-causal role relations were switched. Experiment 1 revealed that 16-month-olds, but not 12-month-olds, associate a dynamic part with the role of agent and a static part with the role of recipient. Experiment 2 showed that 12- and 16-month-olds do not associate a static part with the role of agent or a dynamic part with the role of recipient. Experiment 3 demonstrated that 14-month-olds will learn the relations presented in Experiment 1 and Experiment 2. Experiment 4 revealed that 12-month-olds were able to discriminate the two geometric figures in the events. The results are discussed with respect to infants' developing ability to attend to correlations between dynamic and static cues and the mechanism underlying early object concept acquisition.