Noah Goodman
Fri 25 Sep 2015, 11:00 - 12:30
Informatics Forum (IF-4.31/4.33)

If you have a question about this talk, please contact: Diana Dalla Costa (ddallac)


Probabilistic models of human cognition have been widely successful at capturing the ways that people represent and reason with uncertain knowledge. In this talk I will explore how this approach extends to natural language pragmatics and semantics. I will first describe how probabilistic programming languages provide a formal tool encompassing probabilistic uncertainty and compositional structure. I will use these tools to construct a framework for language understanding that views literal sentence meaning through probabilistic conditioning and pragmatic enrichment as recursive social reasoning. I'll show that uncertainty about the world and the speaker lead to quantity implicature effects, that predict experimental results from reference games. I will then consider the effects of uncertainty about the language itself---what if the listener is unsure about the meaning of words or the topic of conversation? Lexical uncertainty leads to models of vague adjectives (``Bob is tall'') and generic language (``boys are tall''). Topic uncertainty leads to models of figurative speech (hyperbole and irony). Time permitting I'll touch on recent work on uncertainty about common ground. In all of these cases the models predict human judgements with high quantitative accuracy. Taken together this approach provides a theory of context in language understanding, a connection to quantitative behavioral data, and a bridge to our broader understanding of cognition.


Noah D. Goodman is Assistant Professor of Psychology, Linguistics (by courtesy), and Computer Science (by courtesy) at Stanford University. He studies the computational basis of human thought, merging behavioral experiments with formal methods from statistics and programming languages. He received his Ph.D. in mathematics from the University of Texas at Austin in 2003. In 2005 he entered cognitive science, working as Postdoc and Research Scientist at MIT. In 2010 he moved to Stanford where he runs the Computation and Cognition Lab. CoCoLab studies higher-level human cognition including language understanding, social reasoning, and concept learning; the lab also works on applications of these ideas and enabling technologies such as probabilistic programming languages.