Tuesday: Discussion of Chalmers and consciousness.

There are two basic schemas on the table to account for human cognitive behavior.

Symbolic Approach (embedded in a formal machine, human or silicon)

idealization in Plato's Heaven.
mathematics
computer science

1. control time by simplifying to serial order
2. control states by simplifying to a discrete set

Symbols, symbol structures (static)
plus general control mechanisms (discrete time)
that are isolated from the world.

very useful - for math and for programming, even logic
but implies a disciplined, reliable execution mechanism which must be supplied apriori (that is, external to the formal system itself)

Evidently people can do formal, symbolic systems. What does this tell us? In what sense is this the kind of mechanism that underlies human cognition?

A. One response (eg, Chomsky and Newell) is the assertion that ``All cognitive mechanisms simple ARE symbolic ones''. On this `uniformitarian' view, the research questions become:

• (1) What set of symbols is employed?
• (2) what rules do they follow? and
• (3) what control structure runs this system?

Chomsky and Newell endorse the first two problems but brush off the third one.

B. My answeris that the way in which symbolic systems underlie human cognition is: ``Symbolic mechanisms serve as a kind of attractor for effective cognitive mechanisms'' - an ideal set of properties toward which language, logic, mathematics and computation tend over time. This means both in the evolution of our cognitive system (ie, natural language, an innate symbol-like skill of Homo sapiens) and in its culturally evolved refinements over only the last couple millenia (ie, geometry, artithmetic, logic, programming languages, religious dogmas, science, etc) there is an unmistakable tendency toward more symbolic formulation, toward greater formalization of world-models in terms that are serial-order, discrete models. In this case, the research questions are:

• (1) How are the less symbol-like functions (eg, gross motor control, orientation, pattern recognition, etc) achieved?
• (2) How are cognitive symbol-like units (eg, categories, concepts, syntactic structures, etc) implemented and manipulated in a dynamically functioning nervous system (in those special cases where they actually are implemented - eg, long division, constructing theoretical arguments, writing and debugging computer programs, etc)?
• (3) WHY is it that dynamical control mechanisms tend toward symbolic ones through evolutionary and cultural history?

Dynamic Approach (embedded in a body, in continuous time)

account directly for control - no matter how limited the effectiveness of the control. dynamical models work in space and time,
in the brain, body and environment

learning in carefully `designed' networks
eg, IAC, ART, apparent motion model (consciousness?)

emergence of `design' from blind, purposeless processes given some selection mechanism

Implies there will be lots of adhoc systems specialized for various tasks.
Eg, perception of visual forms (sometimes moving)
layout of 3D space, spatial periodicities, control of limbs, of speech, temporal periodicities, generalized force dynamics, etc.

Language, the architypal symbol-like system: phonology, lexicon, syntax

Description of everyday life events in terms of force dynamics

Consciousness. Yes, there is a `hard' problem - a nonfunctional residue

but the residue has no influence on anything but my `feels' - seems like a cognitive epiphenomenon. Seems like a little problem, not a big one, though apparently it is a nonfunctional case.