Cognitive science began by focussing on the representation and processing of symbolic information; and it became natural to view thought as logical theorem-proving. This may be called logicist cognitive science. Yet, in parallel, there was a traditional of machine learning, which focussed not on representation, but on learning from experience. It become natural to view learning, and the use of knowledge gained via learning, as Bayesian inference. How do the probabilistic and logicist viewpoints on cognitive science inter-relate? One suggestion is that the overwhelming majority of cognition (not merely learning) deals with uncertain inference, which is best modelling from a probabilistic standpoint. I shall review some data from the psychology of reasoning that suggests that people even carry out "deductive" reasoning using probabilistic methods. Yet, representation, particularly symbolic language-like representation, can currently only be understood from a logical standpoint. The challenge of defining probabilistic inference over symbolic representations---and fusing the contributions of probability and logic, is thus a major challenge, and the theme of many of the talks in this summer school. I also note that the human ability to learn both by experience, and from instruction, also creates a substantial empirical puzzle for cognitive science.
Further reading:
Chater, N., Tenenbaum, J. B., & Yuille, A. (2006). Probabilistic models of cognition: Conceptual foundations. Trends in Cognitive Sciences, 10, 287-291.
Chater, N., Tenenbaum, J. B., & Yuille, A. (2006). Probabilistic models of cognition: Conceptual foundations. Trends in Cognitive Sciences, 10, 292-293.
Oaksford, M., & Chater, N. (2007). Bayesian Rationality. Oxford: Oxford University Press.
Russell, S. & Norvig, P. (2002). Artificial Intelligence: A Modern Approach. Englewood Cliffs, NJ: Prentice-Hall.
Chapter 1 of the Chater, N. & Oaksford, M. (in press) The Probabilistic Mind, Oxford: Oxford University Press.
">http://www.ipam.ucla.edu/publications/gss2007/co_intro_chapter_v5.doc
Presentation (PowerPoint File)
Copyright. All Rights Reserved.