A basic question about human learning is how adult-like cognitive representations arise during the course of development. One potential answer is that children bring a set of basic abilities to the task of learning, and over the course of development they recombine these basic abilities in increasingly complex and useful ways. This capacity to recombine existing cognitive representations in novel ways is often assumed in computational models;however, this capacity has not been empirically tested. The current proposal is to experimentally evaluate children's ability to form new representations in this manner. This research will therefore test a broad class of formalized developmental models which assume this ability, and potentially shed light on a basic developmental process. This work is most relevant to the Child Development and Behavior branch of NICHD's goal of understanding learning and development, since this ability would potentially explain many types of learning, across both linguistic and nonlinguistic domains. More broadly, a better understanding of children's capacity for creating new representations out of previously-learned ones will likely support improved diagnostics and treatments for language impairments - language acquisition likely relies heavily on this ability - and for cognitive impairments such as autism. The training component of this proposal is centered around developing skills for conducting experiments with infants and children. In the long term, the applicant plans to work at the intersection of computational modeling and child experimentation, testing developmental theories which have been formalized. This will enable computational modeling work that addresses and clarifies developmental theories, and fosters child experiments that test state-of-the art, formalized learning models. This goal will require fluency with both approaches to the study of cognitive development. The key skills learned during this project will involve designing, running, and analyzing infant and child experiments. This includes learning to interact with infants and children in an experimental setting, and also learning experimental designs and methods that are used with these populations. Methods in this proposal include eye-tracking and verbal responses, with age groups ranging from 12 months to three years of age. The applicant has done extensive experimental work with adults, and in the Aslin lab has already designed, run, and analyzed a pilot study (in progress) with 4-year-olds. This variety of proposed experiments will allow the applicant to develop expertise in several experimental areas. Both Dr. Aslin and the University of Rochester are exceptionally strong in experimental approaches, providing an ideal environment for learning these skills. Importantly, the proposed experiments are relevant to recent computational models, including some developed by the applicant. This means that most of the applicant's time can be spent on learning experimental skills, rather than the computational models and theories which motivate this research.
This research is relevant to public health because it tests basic capacities that likely support some of the major conceptual changes that occur in development, including language and concept learning. An understanding of these abilities may lead to improved detection and eventual treatment of cognitive impairments in these areas since the mechanism of impairment in atypical populations may become more clear with an understanding of how normal development progresses. Moreover, an understanding of how children create adult-like conceptual systems from their early conceptual repertoire may allow for improved teaching techniques that foster children's standard learning processes, both in typical and disabled populations.
|Piantadosi, Steven; Aslin, Richard (2016) Compositional Reasoning in Early Childhood. PLoS One 11:e0147734|
|Jara-Ettinger, Julian; Piantadosi, Steve; Spelke, Elizabeth S et al. (2016) Mastery of the logic of natural numbers is not the result of mastery of counting: evidence from late counters. Dev Sci :|
|Piantadosi, Steven T (2016) Efficient estimation of Weber's W. Behav Res Methods 48:42-52|
|Piantadosi, Steven T (2016) A rational analysis of the approximate number system. Psychon Bull Rev 23:877-86|
|Piantadosi, Steven T; Kidd, Celeste (2016) Extraordinary intelligence and the care of infants. Proc Natl Acad Sci U S A 113:6874-9|
|Piantadosi, Steven T; Kidd, Celeste; Aslin, Richard (2014) Rich analysis and rational models: inferring individual behavior from infant looking data. Dev Sci 17:321-37|
|Piantadosi, Steven T (2014) Zipf's word frequency law in natural language: a critical review and future directions. Psychon Bull Rev 21:1112-30|