Learning a theory of causality. Goodman, N. D, Ullman, T. D, & Tenenbaum, J. B Psychol Rev, 118(1):110–119, 2011. doi abstract bibtex The very early appearance of abstract knowledge is often taken as evidence for innateness. We explore the relative learning speeds of abstract and specific knowledge within a Bayesian framework and the role for innate structure. We focus on knowledge about causality, seen as a domain-general intuitive theory, and ask whether this knowledge can be learned from co-occurrence of events. We begin by phrasing the causal Bayes nets theory of causality and a range of alternatives in a logical language for relational theories. This allows us to explore simultaneous inductive learning of an abstract theory of causality and a causal model for each of several causal systems. We find that the correct theory of causality can be learned relatively quickly, often becoming available before specific causal theories have been learned–an effect we term the blessing of abstraction. We then explore the effect of providing a variety of auxiliary evidence and find that a collection of simple perceptual input analyzers can help to bootstrap abstract knowledge. Together, these results suggest that the most efficient route to causal knowledge may be to build in not an abstract notion of causality but a powerful inductive learning mechanism and a variety of perceptual supports. While these results are purely computational, they have implications for cognitive development, which we explore in the conclusion.
@Article{Goodman2011,
author = {Noah D Goodman and Tomer D Ullman and Joshua B Tenenbaum},
journal = {Psychol Rev},
title = {Learning a theory of causality.},
year = {2011},
number = {1},
pages = {110--119},
volume = {118},
abstract = {The very early appearance of abstract knowledge is often taken as
evidence for innateness. We explore the relative learning speeds
of abstract and specific knowledge within a Bayesian framework and
the role for innate structure. We focus on knowledge about causality,
seen as a domain-general intuitive theory, and ask whether this knowledge
can be learned from co-occurrence of events. We begin by phrasing
the causal Bayes nets theory of causality and a range of alternatives
in a logical language for relational theories. This allows us to
explore simultaneous inductive learning of an abstract theory of
causality and a causal model for each of several causal systems.
We find that the correct theory of causality can be learned relatively
quickly, often becoming available before specific causal theories
have been learned--an effect we term the blessing of abstraction.
We then explore the effect of providing a variety of auxiliary evidence
and find that a collection of simple perceptual input analyzers can
help to bootstrap abstract knowledge. Together, these results suggest
that the most efficient route to causal knowledge may be to build
in not an abstract notion of causality but a powerful inductive learning
mechanism and a variety of perceptual supports. While these results
are purely computational, they have implications for cognitive development,
which we explore in the conclusion.},
doi = {10.1037/a0021336},
language = {eng},
medline-pst = {ppublish},
pmid = {21244189},
school = {Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, USA. ngoodman@stanford.edu},
timestamp = {2011.05.10},
}
Downloads: 0
{"_id":"6g6hAgKWsn87wQrGX","bibbaseid":"goodman-ullman-tenenbaum-learningatheoryofcausality-2011","author_short":["Goodman, N. D","Ullman, T. D","Tenenbaum, J. B"],"bibdata":{"bibtype":"article","type":"article","author":[{"firstnames":["Noah","D"],"propositions":[],"lastnames":["Goodman"],"suffixes":[]},{"firstnames":["Tomer","D"],"propositions":[],"lastnames":["Ullman"],"suffixes":[]},{"firstnames":["Joshua","B"],"propositions":[],"lastnames":["Tenenbaum"],"suffixes":[]}],"journal":"Psychol Rev","title":"Learning a theory of causality.","year":"2011","number":"1","pages":"110–119","volume":"118","abstract":"The very early appearance of abstract knowledge is often taken as evidence for innateness. We explore the relative learning speeds of abstract and specific knowledge within a Bayesian framework and the role for innate structure. We focus on knowledge about causality, seen as a domain-general intuitive theory, and ask whether this knowledge can be learned from co-occurrence of events. We begin by phrasing the causal Bayes nets theory of causality and a range of alternatives in a logical language for relational theories. This allows us to explore simultaneous inductive learning of an abstract theory of causality and a causal model for each of several causal systems. We find that the correct theory of causality can be learned relatively quickly, often becoming available before specific causal theories have been learned–an effect we term the blessing of abstraction. We then explore the effect of providing a variety of auxiliary evidence and find that a collection of simple perceptual input analyzers can help to bootstrap abstract knowledge. Together, these results suggest that the most efficient route to causal knowledge may be to build in not an abstract notion of causality but a powerful inductive learning mechanism and a variety of perceptual supports. While these results are purely computational, they have implications for cognitive development, which we explore in the conclusion.","doi":"10.1037/a0021336","language":"eng","medline-pst":"ppublish","pmid":"21244189","school":"Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, USA. ngoodman@stanford.edu","timestamp":"2011.05.10","bibtex":"@Article{Goodman2011,\n author = {Noah D Goodman and Tomer D Ullman and Joshua B Tenenbaum},\n journal = {Psychol Rev},\n title = {Learning a theory of causality.},\n year = {2011},\n number = {1},\n pages = {110--119},\n volume = {118},\n abstract = {The very early appearance of abstract knowledge is often taken as\n\tevidence for innateness. We explore the relative learning speeds\n\tof abstract and specific knowledge within a Bayesian framework and\n\tthe role for innate structure. We focus on knowledge about causality,\n\tseen as a domain-general intuitive theory, and ask whether this knowledge\n\tcan be learned from co-occurrence of events. We begin by phrasing\n\tthe causal Bayes nets theory of causality and a range of alternatives\n\tin a logical language for relational theories. This allows us to\n\texplore simultaneous inductive learning of an abstract theory of\n\tcausality and a causal model for each of several causal systems.\n\tWe find that the correct theory of causality can be learned relatively\n\tquickly, often becoming available before specific causal theories\n\thave been learned--an effect we term the blessing of abstraction.\n\tWe then explore the effect of providing a variety of auxiliary evidence\n\tand find that a collection of simple perceptual input analyzers can\n\thelp to bootstrap abstract knowledge. Together, these results suggest\n\tthat the most efficient route to causal knowledge may be to build\n\tin not an abstract notion of causality but a powerful inductive learning\n\tmechanism and a variety of perceptual supports. While these results\n\tare purely computational, they have implications for cognitive development,\n\twhich we explore in the conclusion.},\n doi = {10.1037/a0021336},\n language = {eng},\n medline-pst = {ppublish},\n pmid = {21244189},\n school = {Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, USA. ngoodman@stanford.edu},\n timestamp = {2011.05.10},\n}\n\n","author_short":["Goodman, N. D","Ullman, T. D","Tenenbaum, J. B"],"key":"Goodman2011","id":"Goodman2011","bibbaseid":"goodman-ullman-tenenbaum-learningatheoryofcausality-2011","role":"author","urls":{},"metadata":{"authorlinks":{}}},"bibtype":"article","biburl":"https://endress.org/publications/ansgar.bib","dataSources":["xPGxHAeh3vZpx4yyE","TXa55dQbNoWnaGmMq"],"keywords":[],"search_terms":["learning","theory","causality","goodman","ullman","tenenbaum"],"title":"Learning a theory of causality.","year":2011}