Probabilistic Models of Cognitive Development (09w5100)
Organizers
Tom Griffiths (University of California Berkeley)
Fei Xu (University of British Columbia)
Description
The Banff International Research Station will host the "Probabilistic Models of Cognitive Development" workshop next week, May 24 - May 29, 2009.
This workshop aims to capitalize on a major new research direction in research on formal models of human cognition. The question of how people come to know so much about the world on the basis of their limited experience has been at the center of the study of the mind since it was first asked by Plato. This question takes a modern form in the nature-nurture debate, which has guided the study of cognitive development from infants to middle childhood over the last few decades. Nativists, favoring strong innate constraints provided by nature, have emphasized competences found in young infants (e.g., constraints on word learning, early understanding of the physical world) whereas empiricists, who focus on the role of experience and nurture, have emphasized learning mechanisms (e.g., keeping track of frequencies and correlations). However, this debate has been hard to resolve without formal tools for evaluating what might plausibly be learned from experience, and what kind of constraints are necessary to support the inferences that children make.
In recent years, several researchers in the cognitive sciences have argued that the nature-nurture framework may have set up a false dichotomy. A more fruitful and productive research strategy may be to find principled ways of combining prior constraints with statistical information in the input. In particular, a number of researchers have begun to use the principles of Bayesian statistics to establish a formal framework for investigating empirical phenomena in development and building computational models of developmental processes. The technical advances that have been made in the use of probabilistic models over the last twenty years in statistics, computer science, and machine learning have made this research enterprise possible, providing psychologists with a set of mathematical and computational tools that can be used to build explicit models of psychological phenomena. By analyzing the conclusions that a rational learner might draw from the data provided by experience, Bayesian models can be used to investigate how nature and nurture contribute to human knowledge.
This workshop aims to capitalize on a major new research direction in research on formal models of human cognition. The question of how people come to know so much about the world on the basis of their limited experience has been at the center of the study of the mind since it was first asked by Plato. This question takes a modern form in the nature-nurture debate, which has guided the study of cognitive development from infants to middle childhood over the last few decades. Nativists, favoring strong innate constraints provided by nature, have emphasized competences found in young infants (e.g., constraints on word learning, early understanding of the physical world) whereas empiricists, who focus on the role of experience and nurture, have emphasized learning mechanisms (e.g., keeping track of frequencies and correlations). However, this debate has been hard to resolve without formal tools for evaluating what might plausibly be learned from experience, and what kind of constraints are necessary to support the inferences that children make.
In recent years, several researchers in the cognitive sciences have argued that the nature-nurture framework may have set up a false dichotomy. A more fruitful and productive research strategy may be to find principled ways of combining prior constraints with statistical information in the input. In particular, a number of researchers have begun to use the principles of Bayesian statistics to establish a formal framework for investigating empirical phenomena in development and building computational models of developmental processes. The technical advances that have been made in the use of probabilistic models over the last twenty years in statistics, computer science, and machine learning have made this research enterprise possible, providing psychologists with a set of mathematical and computational tools that can be used to build explicit models of psychological phenomena. By analyzing the conclusions that a rational learner might draw from the data provided by experience, Bayesian models can be used to investigate how nature and nurture contribute to human knowledge.
The Banff International Research Station for Mathematical Innovation and Discovery (BIRS) is a collaborative Canada-US-Mexico venture that provides an environment for creative interaction as well as the exchange of ideas, knowledge, and methods within the Mathematical Sciences, with related disciplines and with industry. The research station is located at The Banff Centre in Alberta and is supported by Canada's Natural Science and Engineering Research Council (NSERC), the US National Science Foundation (NSF), Alberta's Advanced Education and Technology, and Mexico's Consejo Nacional de Ciencia y Tecnología (CONACYT).
