TY - GEN

T1 - Look-Ahead Monte Carlo with People

AU - Blundell, Charles

AU - Sanborn, Adam

AU - Griffiths, Thomas L.

N1 - Funding Information:
Acknowledgements: We wish to thank Peter Dayan and Charles Sutton for several useful discussions.
Publisher Copyright:
© CogSci 2012.All rights reserved.

PY - 2012

Y1 - 2012

N2 - Investigating people’s representations of categories of complicated objects is a difficult challenge, not least because of the large number of ways in which such objects can vary. To make progress we need to take advantage of the structure of object categories – one compelling regularity is that object categories can be described by a small number of dimensions. We present Look-Ahead Monte Carlo with People, a method for exploring people’s representations of a category where there are many irrelevant dimensions. This method combines ideas from Markov chain Monte Carlo with People, an experimental paradigm derived from an algorithm for sampling complicated distributions, with hybrid Monte Carlo, a technique that uses directional information to construct efficient statistical sampling algorithms. We show that even in a simple example, our approach takes advantage of the structure of object categories to make experiments shorter and increase our ability to accurately estimate category representations.

AB - Investigating people’s representations of categories of complicated objects is a difficult challenge, not least because of the large number of ways in which such objects can vary. To make progress we need to take advantage of the structure of object categories – one compelling regularity is that object categories can be described by a small number of dimensions. We present Look-Ahead Monte Carlo with People, a method for exploring people’s representations of a category where there are many irrelevant dimensions. This method combines ideas from Markov chain Monte Carlo with People, an experimental paradigm derived from an algorithm for sampling complicated distributions, with hybrid Monte Carlo, a technique that uses directional information to construct efficient statistical sampling algorithms. We show that even in a simple example, our approach takes advantage of the structure of object categories to make experiments shorter and increase our ability to accurately estimate category representations.

KW - Markov chain Monte Carlo

KW - category representation

KW - directional judgements

UR - http://www.scopus.com/inward/record.url?scp=84941270230&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84941270230&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84941270230

T3 - Building Bridges Across Cognitive Sciences Around the World - Proceedings of the 34th Annual Meeting of the Cognitive Science Society, CogSci 2012

SP - 1356

EP - 1361

BT - Building Bridges Across Cognitive Sciences Around the World - Proceedings of the 34th Annual Meeting of the Cognitive Science Society, CogSci 2012

A2 - Miyake, Naomi

A2 - Peebles, David

A2 - Cooper, Richard P.

PB - The Cognitive Science Society

T2 - 34th Annual Meeting of the Cognitive Science Society: Building Bridges Across Cognitive Sciences Around the World, CogSci 2012

Y2 - 1 August 2012 through 4 August 2012

ER -