Associative learning or Bayesian inference? Revisiting backwards blocking reasoning in adults.

Causal reasoning is a fundamental cognitive ability that enables humans to learn about the complex interactions in the world around them. However, the cognitive mechanisms that underpin causal reasoning are not well understood. For instance, there is debate over whether Bayesian inference or associative learning best captures causal reasoning in human adults. The two experiments and computational models reported here were designed to examine whether adults engage in one form of causal inference called backwards blocking reasoning, whether the presence of potential distractors affects performance, and how adults' ratings align with the predictions of different computational models. The results revealed that adults engaged in backwards blocking reasoning regardless of whether distractor objects are present and that their causal judgements supported the predictions of a Bayesian model but not the predictions of two different associative learning models. Implications of these results are discussed.

When correlation equals causation: A behavioral and computational account of second-order correlation learning in children.

We examined 2- and 3-year-old children's ability to use second-order correlation learning-in which a learned correlation between two pairs of features (e.g., A and B, A and C) is generalized to the noncontiguous features (i.e., B and C)-to make causal inferences. Previous findings showed that 20- and 26-month-old children can use second-order correlation learning to learn about static and dynamic features in category and noncategory contexts. The current behavioral study and computational model extend these findings to show that 2- and 3-year-olds can detect the second-order correlation between an object's surface feature and its capacity to activate a novel machine, but only if the children had encoded the first-order correlations on which the second-order correlation was based. These results have implications for children's developing information-processing capacities on their ability to use second-order correlations to infer causal relations in the world.

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.

Do 5-Month-Old Infants Possess an Evolved Detection Mechanism for Snakes, Sharks, and Rodents?

The four experiments reported here used the preferential looking and habituation paradigms to examine whether 5-month-olds possess a perceptual template for snakes, sharks, and rodents. It was predicted that if infants possess such a template then they would attend preferentially to schematic images of these non-human animal stimuli relative to scrambled versions of the same stimuli. The results reveal that infants look longer at a schematic snake than at two scrambled versions of that image and generalize from real snakes to the schematic image. The experiments also demonstrate that 5-month-olds show no preferential looking for schematic sharks or schematic rodents relative to scrambled versions of those images. These data add to the growing support for the view that humans, like many non-human animals, possess an evolved fear mechanism for detecting threats that were recurrent across evolutionary time.

Seeing the unseen: Second-order correlation learning in 7- to 11-month-olds.

We present four experiments with the object-examining procedure that investigated 7-, 9-, and 11-month-olds' ability to associate two object features that were never presented simultaneously. In each experiment, infants were familiarized with a number of 3D objects that incorporated different correlations among the features of those objects and the body of the objects (e.g., Part A and Body 1, and Part B and Body 1). Infants were then tested with objects with a novel body that either possessed both of the parts that were independently correlated with one body during familiarization (e.g., Part A and B on Body 3) or that were attached to two different bodies during familiarization. The experiments demonstrate that infants as young as 7months of age are capable of this kind of second-order correlation learning. Furthermore, by at least 11months of age infants develop a representation for the object that incorporates both of the features they experienced during training. We suggest that the ability to learn second-order correlations represents a powerful but as yet largely unexplored process for generalization in the first years of life.

Who is the dynamic duo? How infants learn about the identity of objects in a causal chain.

Four experiments investigated infants' and adults' knowledge of the identity of objects in a causal sequence of events. In Experiments 1 and 2, 18- and 22-month-olds in the visual habituation procedure were shown a 3-step causal chain event in which the relation between an object's part (dynamic or static) and its causal role was either consistent or inconsistent with the real-world. In Experiment 3, 22-month-olds were tested with a delayed launching causal chain in which the second object, rather than the first, was the agent of the outcome. In Experiment 4, adults were shown the same events and were asked to judge whether the first or second object in the causal chain was animate or inanimate. Experiments 1 and 2 revealed that 18-month-olds were unconstrained in the part-causal role relations they would encode, but 22-month-olds learned only those relations that were consistent with the real-world. Experiment 3 showed that 22-month-olds expect the second object in a delayed launching sequence to possess a dynamic, moving part. Experiment 4 showed that adults expect the first object of a causal chain to be animate and the second object to be inanimate. The results are discussed with regard to the developmental timetable for causal learning and the mechanisms for early concept acquisition.

Connectionist modeling of developmental changes in infancy: approaches, challenges, and contributions.

Connectionist models have been applied to many phenomena in infant development including perseveration, language learning, categorization, and causal perception. In this article, we discuss the benefits of connectionist networks for the advancement of theories of early development. In particular, connectionist models contribute novel testable predictions, instantiate the theorized mechanism of change, and create a unifying framework for understanding infant learning and development. We relate these benefits to the 2 primary approaches used in connectionist models of infant development. The first approach employs changes in neural processing as the basis for developmental changes, and the second employs changes in infants' experiences. The review sheds light on the unique hurdles faced by each approach as well as the challenges and solutions related to both, particularly with respect to the identification of critical model components, parameter specification, availability of empirical data, and model comparison. Finally, we discuss the future of modeling work as it relates to the study of development. We propose that connectionist networks stand to make a powerful contribution to the generation and revision of theories of early child development. Furthermore, insights from connectionist models of early development can improve the understanding of developmental changes throughout the life span.

Expectations about single event probabilities in the first year of life: The influence of perceptual and statistical information.

Recent evidence suggests that infants can generate expectations about future events from a sample of probabilistic data. However, little is known about the conditions that support the development of this ability. Three experiments tested the prediction that 8- and 12-month-olds respond to base rates as well as perceptual cues when they generate expectations from a sample of probabilistic data. Results revealed that 12-month-olds were sensitive to the statistical and perceptual properties of the evidence depending on the distribution of high-to-low base rate items in the sample. Specifically, 12-month-olds focused on perceptual features of the evidence when a sample was large and more skewed (e.g., 6:1), whereas they attended to statistical properties when the sample was smaller and less skewed (e.g., 4:1). In contrast, eight-month-olds always focused on the perceptual features of the evidence. Neither group generated expectations from a small, less skewed sample (e.g., 2:1). These results suggest that the ability to generate expectations about future events is mediated by specific features of the available evidence and undergoes significant change during the 1st year of life.

Does causal action facilitate causal perception in infants younger than 6 months of age?

Previous research has established that infants are unable to perceive causality until 6» months of age. The current experiments examined whether infants' ability to engage in causal action could facilitate causal perception prior to this age. In Experiment 1, 4½-month-olds were randomly assigned to engage in causal action experience via Velcro sticky mittens or not engage in causal action because they wore non-sticky mittens. Both groups were then tested in the visual habituation paradigm to assess their causal perception. Infants who engaged in causal action - but not those without this causal action experience - perceived the habituation events as causal. Experiment 2 used a similar design to establish that 4½-month-olds are unable to generalize their own causal action to causality observed in dissimilar objects. These data are the first to demonstrate that infants under 6 months of age can perceive causality, and have implications for the mechanisms underlying the development of causal perception.

Exploring the neural correlates of goal-directed action and intention understanding.

Because we are a cooperative species, understanding the goals and intentions of others is critical for human survival. In this fMRI study, participants viewed reaching behaviors in which one of four animated characters moved a hand towards one of two objects and either (a) picked up the object, (b) missed the object, or (c) changed his path halfway to lift the other object. The characters included a human, a humanoid robot, stacked boxes with an arm, and a mechanical claw. The first three moved in an identical, human-like biological pattern. Right posterior superior temporal sulcus (pSTS) activity increased when the human or humanoid robot shifted goals or missed the target relative to obtaining the original goal. This suggests that the pSTS was engaged differentially for figures that appeared more human-like rather than for all human-like motion. Medial frontal areas that are part of a protagonist-monitoring network with the right pSTS (e.g., Mason and Just, 2006) were most engaged for the human character, followed by the robot character. The current data suggest that goal-directed action and intention understanding require this network and it is used similarly for the two processes. Moreover, it is modulated by character identity rather than only the presence of biological motion. We discuss the implications for behavioral theories of goal-directed action and intention understanding.

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.

Attention to Multiple Cues During Spontaneous Object Labeling.

It is well established that 2-year-olds attribute a novel label to an object's global shape rather than local features (i.e., parts). Although recent studies have found that younger infants also attend to global rather than local features when given a label, the test stimuli in these experiments confounded parts and shape by varying both or neither. With infants (16- and 24-month-olds) and adults, this experiment disentangled shape and parts with appropriate test objects. Results showed a clear development of a strategy incorporating multiple cues. Across three age groups, there was an increase in generalizing labels to objects matching the exemplar's local and global features (parts, base, and shape), and a decrease to objects matching in only one local feature. We discuss these results in terms of a learned flexibility in using multiple cues to predict lexical categories.

Developing without concepts.

We evaluate the heterogeneity hypothesis by considering the developmental time course and the mechanism of acquisition of exemplars, prototypes, and theories. We argue that behavioral and modeling data point to a sequential emergence of these three types of concepts within a single system. This suggests that similar or identical underlying cognitive processes - rather than separate ones - underpin representation acquisition.

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.

Does women's greater fear of snakes and spiders originate in infancy?

Previous studies with adult humans and non-human animals revealed more rapid fear learning for spiders and snakes than for mushrooms and flowers. The current experiments tested whether 11-month-olds show a similar effect in learning associative pairings between facial emotions and fear-relevant and fear-irrelevant stimuli. Consistent with the greater incidence of snake and spider phobias in women, results show that female but not male infants learn rapidly to associate negative facial emotions with fear-relevant stimuli. No difference was found between the sexes for fear-irrelevant stimuli. The results are discussed in relation to fear learning, phobias, and a specialized evolved fear mechanism in humans.

New perspectives on the effects of action on perceptual and cognitive development.

This special section was motivated by a resurgence in the view that it is impossible to investigate perceptual and cognitive development without considering how it is affected by, and intertwined with, infants' and children's action in the world. This view has long been foundational to the field, yet contemporary investigations of the effects of acting on cognition and perception have been limited. The research showcased in this section indicates that this trend is changing as researchers consider anew the ways in which cognition derives structure from action. The work presented here illustrates the breadth of these potential effects across ages and domains of development, and it highlights the breadth of methods that can be recruited to investigate them. This new research focus provides insight for the mechanisms by which action affects perception and cognition and at the same time reveals that much remains to be learned.

Producing and processing self-propelled motion in infancy.

Three experiments investigated 5- through 8-month-olds' ability to encode self-propelled and caused motion and examined whether processing of motion onset changes when crawling begins. Infants were habituated (Experiments 1 and 2) or familiarized (Experiment 3) with simple causal and noncausal launching events. They then viewed the caused-to-move and self-propelled objects from the events both stationary and side-by-side, and their preferential looking to the objects was assessed. Results revealed that 5- and 6-month-olds displayed a different pattern of looking than did 8-month-olds. More notably, noncrawling 7-month-olds and 7-month-olds with crawling experience also demonstrated such a differential pattern. These data suggest that processing of motion onset changes in concert with the commencement of self-locomotion. Findings are discussed in reference to the mechanisms underlying infants' ability to recognize self-propelled motion and the scope of the relationship between action production and action perception in infancy.

Developing object concepts in infancy: an associative learning perspective.

We present a domain-general framework called constrained attentional associative learning to provide a developmental account for how and when infants form concepts for animates and inanimates that encapsulate not only their surface appearance but also their movement characteristics. Six simulations with the same general-purpose architecture implement the features of the theory to model infant behavior in learning about objects' motion trajectory, their causal role, their onset of motion, and the initial mapping between a label and a moving object. Behavioral experiments with infants tested novel hypotheses generated by the model, showing that verbal labels initially may be associated with specific features rather than overall shape. Implications of the framework and model are discussed in relation to the mechanisms of early learning, the development of the animate-inanimate distinction, and the nature of development in the first years of life.

Do infants possess an evolved spider-detection mechanism?

Previous studies with various non-human animals have revealed that they possess an evolved predator recognition mechanism that specifies the appearance of recurring threats. We used the preferential looking and habituation paradigms in three experiments to investigate whether 5-month-old human infants have a perceptual template for spiders that generalizes to real-world images of spiders. A fourth experiment assessed whether 5-month-olds have a perceptual template for a non-threatening biological stimulus (i.e., a flower). The results supported the hypothesis that humans, like other species, may possess a cognitive mechanism for detecting specific animals that were potentially harmful throughout evolutionary history.

Language is not just for talking: redundant labels facilitate learning of novel categories.

In addition to having communicative functions, verbal labels may play a role in shaping concepts. Two experiments assessed whether the presence of labels affected category formation. Subjects learned to categorize "aliens" as those to be approached or those to be avoided. After accuracy feedback on each response was provided, a nonsense label was either presented or not. Providing nonsense category labels facilitated category learning even though the labels were redundant and all subjects had equivalent experience with supervised categorization of the stimuli. A follow-up study investigated differences between learning verbal and nonverbal associations and showed that learning a nonverbal association did not facilitate categorization. The findings show that labels make category distinctions more concrete and bear directly on the language-and-thought debate.