Structured Representations of Utility in Combinatorial Domains

People can judge whether they will enjoy dishes like waffles with horseradish cream sauce or broccoli ice cream even if they have never tried them. What representations and computations support reasoning in such situations? We develop a theory of decision making in combinatorial domains. Its central claim is that utility functions can be compositionally structured: The utility of a combination is a function of its constituents’ utilities and the rules for combining them. Utilities are induced from experience by probabilistic reasoning over the structured space of utility functions. In a series of experiments, we show how this theory can capture human evaluations of novel food combinations. We first show that the theory quantitatively predicts evaluations of novel food combinations. We then report more strongly controlled experiments (using unfamiliar foods) that rule out several alternative theories. Taken together, these experiments demonstrate how compositionally structured representations of utility can support decision making in combinatorial domains.

[1]  Craig Boutilier,et al.  Elicitation of Factored Utilities , 2008, AI Mag..

[2]  J. Bentham.  An Introduction to the Principles of Morals and Legislation , 1945, Princeton Readings in Political Thought.

[3]  R. Hertwig,et al.  The description–experience gap in risky choice , 2009, Trends in Cognitive Sciences.

[4]  P. Wakker.  Additive Representations of Preferences: A New Foundation of Decision Analysis , 1988 .

[5]  D. McFadden.  Conditional logit analysis of qualitative choice behavior , 1972 .

[6]  Thomas L. Griffiths,et al.  A Rational Analysis of Rule-Based Concept Learning , 2008, Cogn. Sci..

[7]  A. Tversky.  Elimination by aspects: A theory of choice. , 1972 .

[8]  Timothy E. J. Behrens,et al.  Online evaluation of novel choices by simultaneous representation of multiple memories , 2013, Nature Neuroscience.

[9]  Gordon D. A. Brown,et al.  Decision by sampling , 2006, Cognitive Psychology.

[10]  J. Feldman.  What is a visual object? , 2003, Trends in Cognitive Sciences.

[11]  B. Scholl.  Objects and attention: the state of the art , 2001, Cognition.

[12]  W. M. Gorman.  The Structure of Utility Functions , 1968 .

[13]  A. R. Wagner,et al.  Negative patterning in classical conditioning: Summation of response tendencies to isolable and configurai components , 1972 .

[14]  R. L. Keeney,et al.  Decisions with Multiple Objectives: Preferences and Value Trade-Offs , 1977, IEEE Transactions on Systems, Man, and Cybernetics.

[15]  Y. Niv,et al.  Discovering latent causes in reinforcement learning , 2015, Current Opinion in Behavioral Sciences.

[16]  Fahiem Bacchus,et al.  Graphical models for preference and utility , 1995, UAI.

[17]  Cleotilde Gonzalez,et al.  Instance-based learning in dynamic decision making , 2003, Cogn. Sci..

[18]  Elizabeth S. Spelke,et al.  Principles of Object Perception , 1990, Cogn. Sci..

[19]  N. Kanwisher,et al.  Objects, Attributes, and Visual Attention: Which, What, and Where , 1992 .

[20]  Timothy D. Wilson,et al.  Affective Forecasting , 2005 .

[21]  Peter C. Fishburn,et al.  INTERDEPENDENCE AND ADDITIVITY IN MULTIVARIATE, UNIDIMENSIONAL EXPECTED UTILITY TIHEORY* , 1967 .

[22]  Carl E. Rasmussen,et al.  Gaussian processes for machine learning , 2005, Adaptive computation and machine learning.

[23]  Charles Kemp,et al.  The discovery of structural form , 2008, Proceedings of the National Academy of Sciences.

[24]  J. Pearce.  Similarity and discrimination: a selective review and a connectionist model. , 1994, Psychological review.

[25]  Ralph L. Keeney,et al.  Decisions with multiple objectives: preferences and value tradeoffs , 1976 .

[26]  R. Rescorla.  A theory of pavlovian conditioning: The effectiveness of reinforcement and non-reinforcement , 1972 .

[27]  R. H. Strotz.  The Empirical Implications of a Utility Tree , 1957 .

[28]  J. Stockman.  Pure Reasoning in 12-Month-Old Infants as Probabilistic Inference , 2013 .

[29]  N. Schmajuk,et al.  Stimulus configuration, classical conditioning, and hippocampal function. , 1992, Psychological review.

[30]  D. McDermott.  LANGUAGE OF THOUGHT , 2012 .

[31]  Christopher G. Lucas,et al.  A rational model of function learning , 2015, Psychonomic Bulletin & Review.

[32]  J. Neumann,et al.  Theory of Games and Economic Behavior. , 1945 .

[33]  Geoffrey E. Hinton,et al.  Visualizing Data using t-SNE , 2008 .

[34]  A. Tversky,et al.  Foundations of Measurement, Vol. I: Additive and Polynomial Representations , 1991 .

[35]  Daniel Kahneman,et al.  Predicting a changing taste: Do people know what they will like? , 1992 .

[36]  Richard S. Sutton,et al.  Reinforcement Learning: An Introduction , 1998, IEEE Trans. Neural Networks.

[37]  Chin-Hwa Kuo,et al.  Bootstrapping in a language learning environment , 2003, J. Comput. Assist. Learn..

[38]  D. Kahneman,et al.  Back to Bentham? Explorations of experience utility , 1997 .

[39]  J. Douglas Carroll,et al.  Toward a new paradigm for the study of multiattribute choice behavior: Spatial and discrete modeling of pairwise preferences. , 1991 .

[40]  Vilfredo Pareto,et al.  Manuale di economia politica , 1965 .

[41]  I. Gilboa,et al.  Case-Based Decision Theory , 1995 .

[42]  Karl J. Friston,et al.  Bayesian model selection for group studies , 2009, NeuroImage.

[43]  Noah D. Goodman,et al.  Bootstrapping in a language of thought: A formal model of numerical concept learning , 2012, Cognition.

[44]  Alessandro Moschitti,et al.  Making Tree Kernels Practical for Natural Language Learning , 2006, EACL.

[45]  Noah D. Goodman,et al.  Theory Acquisition and the Language of Thought , 2008 .

[46]  Noah D. Goodman,et al.  Learning a theory of causality. , 2011, Psychological review.

[47]  Jaap Van Brakel,et al.  Foundations of measurement , 1983 .