Inferring prior probabilities from Bayes-optimal behavior

Liam Paninski

To appear,
Neural Information Processing Systems 2005

We discuss a method for obtaining a subject's a priori beliefs from his/her behavior in a psychophysics context, under the assumption that the behavior is (nearly) optimal from a Bayesian perspective. The method is nonparametric in the sense that we do not assume that the prior belongs to any fixed class of distributions (e.g., Gaussian). Despite this increased generality, the method is relatively simple to implement, being based in the simplest case on a linear programming algorithm, and more generally on a straightforward maximum likelihood or maximum a posteriori formulation, which turns out to be a concave maximization problem (with no non-global local maxima) in many important cases. We explore the variability of the methods and emphasize the importance of regularization of the problem, via mathematical analysis and numerical examples. We close by briefly discussing an interesting connection to recent models of neural population coding.
Reprint (190K, pdf)  |  Liam Paninski's home