Infants' expectations about motion properties of objects: The role of sequence representations in generalization

2018-07-01T03:17:53Z (GMT) by Yevdokiya Yermolayeva
<p>Motion cues such as agency and self-propulsion are considered central to infants’ developing ability to distinguish animate and inanimate entities. Although infants’ ability to encode these properties and attribute them to the appropriate classes of entities has been well-documented, no consensus exists regarding how infants represent these motion properties and generalize between them. The present work sought to address these issues: it expands on prior research on infants’ generalization about motion, proposes a new account of generalization, and rigorously evaluates the account using experimental and observational methods.</p> <p>The first set of studies presented herein examined infants’ ability to generalize about stopping, which is a motion property that can distinguish animates from inanimates (only the former can stop abruptly without an external force) and has not been studied previously. In particular, the studies examined infants’ ability to form expectations about stopping based on other motion properties typically attributed to animates and inanimates, namely, agency and recipiency. In Experiments 1 and 2, infants were habituated to a causal event and shown test events in which the agent and the recipient stopped gradually or abruptly. The results of Experiment 1 showed that 12- and 16-month-old infants looked longer at an agent that stopped gradually than one that stopped abruptly, indicating the presence of expectations for the manner of stopping of agents but not recipients. The results of Experiment 2 showed that 14-month-old infants did not have expectations about either agents or recipients. This U-shaped developmental trajectory may have been due to the different types of information that 12- and 16-month-olds encoded. When causal role information was removed in Experiment 3, 12-month-olds’ behavior remained similar to their behavior in Experiment 1, which implied that they may have attended to something other than causal roles during habituation. Finally, the results of Experiment 4 showed that the addition of a self-propulsion cue that correlated with agency did not induce 14-month-old infants to generalize about stopping. Taken together, these studies demonstrate that the development of expectations about stopping emerges within the second year of life.</p> <p>The second set of studies was designed to test a new theoretical account of the foundation of infants’ generalization. The account posited that infants’ generalization about motion is dependent on their ability to learn sequential motion information from their environment. Experiment 5 tested the prediction that infants’ experience with motion sequences constrains their learning such that they learn sequences that are consistent with their experience but not those that are inconsistent. The results of the experiment confirmed the prediction and demonstrated a developmental change in the emergence of experience-based constraints: whereas 12-month-olds showed similar learning regardless of sequence type, 16-month-olds only learned sequences consistent with their experience. The Observational Study examined the theoretical account further by quantifying infants’ daily experience with motion. Specifically, the study examined which regularities in infants’ environments are sufficient to give rise to generalization: regularities in the frequency of individual events, in the frequency of event pairs, or in the transitional probability within event pairs. The results suggested that the frequency of event pairs provides the information that agrees most closely with infants’ behavioral patterns</p> <p>Taken together, the studies reported herein make significant contributions to the existing research on infants’ generalization. Specifically, they provide new evidence regarding the extent of these abilities and they clarify the mechanism that enables infants to form such expectations. Furthermore, the studies demonstrate that the regularities in infants’ environments are sufficient to support generalization about animate and inanimate properties, which implies that specialized mechanisms may not be necessary for the emergence of the animate/inanimate distinction</p>



In Copyright