The Brain's Concepts
Mirror Neurons, Simulation Semantics, and the Neural Theory of Language
			    George Lakoff
		  University of California, Berkeley

A discussion of recent collaborative work with Vittorio Gallese (of
the mirror neuron group in Parma) on the direct role of the
sensory-motor system in characterizing concepts - including abstract
concepts like freedom, love, or causation.

The talk is set with the Neural Theory of Language now being developed
at Berkeley. It discusses how semantics can be based on neural
simulation, with neural computational models provided via
Feldman-Narayanan structured connectionist models, and natural
language semantics provided via cognitive semantics.

There are two parts to the talk. The first contradicts the idea that 
all concepts are "abstractions" in the sense that they must be 
represented in the brain independently of the sensory-motor system. 
It traces thoroughly through one example where enough is known about 
the neuroscience, the neural computation, and the linguistics to 
argue that the brain's sensory-motor system can fully characterize 
the properties of the given concept.

The second part takes up the role of conceptual metaphors and "Cogs"
in adapting the sensory-motor system to characterize the meanings of
abstract concepts. The idea of "Cogs" is new, and central to how the
sensory-motor system can characterize the semantics of grammar.
Included is the idea of "dissociative learning" - the use of neural
inhibition to produce conceptual and linguistic generalizations from
exemplars.

The point of the talk is to cash out the idea of embodied meaning,
presented in the speaker's earlier work and underlying current
research on the neural theory of language.