******* Language, Cognition, and Computation Lecture Series *******

 
Title                       Grounding conceptual processing in modality-specific systems: Current evidence and issues

Speaker                 Lawrence W. Barsalou

Affiliation             Department of Psychology, Emory University

Date                      Thursday, February 17, 2005

Time                      2:00pm

Location                E15-070 (Bartos Theater)


Abstract

Accumulating evidence implicates the brain's modality-specific systems in conceptual processing. When people represent knowledge about a category, multi-modal simulations of its members become active and affect task performance. Examples of such findings will be reviewed from cognitive psychology and cognitive neuroscience. At the time this research was performed, it was not widely accepted that modality-specific systems participate in conceptual processes. Researchers holding this hypothesis therefore attempted to assess it primarily in demonstration experiments. Now that the presence of modality-specific processing in conceptual tasks is becoming well established, however, demonstration experiments are likely to have diminishing returns. Instead, it is important to begin establishing the specific roles that modality-specific systems play in conceptual processing. Issues that are increasingly likely to be raised include: (1) Do multi-modal simulations play roles in implementing basic symbolic functions, such as argument binding, relations, productivity, recursion, and abstract concepts? Or are these functions implemented with amodal symbols, or with some other mechanism? (2) What are the relations between multi-modal simulations and language? Do linguistic symbols control simulations, and do simulations ground the compositional meanings that result from combining words? (3) How does the brain represent abstract concepts such as truth? Do modality-specific systems play a role? (4) What relations do multi-modal simulations have with learning systems in the brain, such as association areas? What sorts of computational theories that implement simulations in neural net mechanisms are possible and plausible? For modality-specific approaches to develop further, it will become increasingly important for researchers to address such issues effectively. Preliminary ideas and empirical evidence, where available, will be offered.

*******************************************************************