We thank Rob Goldstone, Stephen Palmer, Karen Schloss, Tania Lombrozo, Charles Kemp, Noah Goodman, Eleanor Rosch, and the Computational Cognitive Science Lab for insightful discussions, and Shubin Li, Brian Tang, and David Belford for help with experiment construction, running participants and data analysis. This work was supported by grant FA9550-07-1-0351 from the AFOSR. Austerweil, J. L. & Gri ths, T. L. (2009). Analyzing human feature learning as nonparametric Bayesian inference. In D. Koller, Y. Bengio, D. Schuurmans, & L. Bottou (Eds.), Advances in Neural Information Processing Systems (Vol. 21). Cambridge, MA: MIT Press. Austerweil, J. L. & Gri ths, T. L. (2009). The e ect of distributional information on feature learning. The 31st Annual Conference of the Cognitive Science Society. Amsterdam, Netherlands. Gri ths, T. L. & Ghahramani, T. L. (2006). In nite latent feature models and the Indian bu et process. In B. Scholkopf, J. Platt, & T. Hofmann (Eds.), Advances in Neural Information Processing Systems (Vol. 18). Cambridge, MA: MIT Press. Shi rin, R. M. & Lightfoot, N. (1997). Perceptual learning of alphanumeric-like characters. In The Psychology of Learning and Motivation (Vol. 36, p. 45-82). San Diego: Academic Press. Schyns, P. G. & Murphy, G. (1994). The ontogeny of part representation in object concepts. In The Psychology of Learning and Motivation (Vol. 31, p. 305-354). San Diego: Academic Press. Wood, F., Gri ths, T. L., & Ghahramani, Z. (2006). A non-parametric Bayesian method for inferring hidden causes. In Proceedings of the 22nd Conference on Uncertainty in Arti cial Intelligence. Feature: a part of a stimulus elementary in psychological processing Feature representations are an essential part of cognitive modeling. Feature representations change with experience: People infer features useful for categorization (Schyns & Murphy, 1994) People unite features to facilitate rapid search (Shi rin & Lightfoot, 1997)