Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!tut.cis.ohio-state.edu!ucbvax!bloom-beacon!bu-cs!slehar
From: slehar@bucasd.bu.edu (Lehar)
Newsgroups: comp.ai.neural-nets
Subject: Re: What has NN research taught us about the brain?
Message-ID: <44971@bu-cs.BU.EDU>
Date: 20 Dec 89 05:43:01 GMT
References: <9096@cs.yale.edu>
Sender: daemon@bu-cs.BU.EDU
Distribution: comp.ai.neural-nets
Organization: Boston University Center for Adaptive Systems
Lines: 89
In-reply-to: zador-anthony@CS.YALE.EDU's message of 19 Dec 89 18:20:51 GMT


PAPER TO ADDRESS THE QUESTION "WHAT HAS NN TAUGHT US ABOUT THE BRAIN?"

My vote would go to Stephen Grossberg's models.  These models are unique in
several aspects:

  1:	the models are based on behavioral data, rather than being arbitrary
	networks that perform cute tricks, these networks are models of
	behavior of an entire organism, such as in classical conditioning,
	and have made quantitative behavioral predictions on numerous occasions

  2:	the models also extend to the neurological level, where Grossberg
	makes specific predictions about certain pathways in hippocampus,
	cerebellum, visual cortex, etc. many of which have already been
	confirmed

  3:	the models reflect the dynamic nature of brain mechanisms, being
	expressed as dynamic systems and differential equations.  This is
	an aspect of cognition which is ignored by most neural models for
	simplicity.  Grossberg's models have complicated dynamics and are
	difficult to simulate and analyze, but in return they display
	a diversity of subtle dynamic behaviors unseen in other models,
	and reminiscent of the subtle complexity of natural networks, like
	Aplysia.

  4:	Grossberg has written on so many interesting and diverse fields
	including

		vision: the boundary contour system & feature contour 
		system is a network that was designed to replicate human
		visual illusions- where human vision fails, to understand
		how human vision works.  Each element of the model is
		rigorously derived from psychophysical data on illusions, 
		yet they correspond remarkably with known neural architectures
		of the visual system, while making predictions about what is
		not yet known (some recently confirmed).

		motor control: Grossberg's models begin with postulates on
		whether feedback or feedforward systems are used, and results
		in a hybrid system where initial feedback errors are used to
		learn feedforward precision.  Components of the model 
		correspond to known motor control architectures in limbs,
		spinal chord and cerebellum.

		cognition:  Grossberg has a whole lineage of models describing
		ever more complex aspects of classical conditioning, arousal,
		motivation, feedback, normalization of inputs, etc. culminating
		in high level cognitive models such as adaptive resonance
		theory, with neurological analogs in hippocampus, hypothalamus
		and cortex.
	
To me, this represents the finest in neural modeling, building models
that are inspired by behavior, making predictions about
neurophysiology, and explaining dynamic processes in the brain.

Stephen Grossberg & Dejan Todorovic NEURAL DYNAMICS OF 1-D AND 2-D
BRIGHTNESS PERCEPTION: A UNIFIED MODEL OF CLASSICAL AND RECENT
PHENOMENA Perception & Psychophysics 1987

Stephen Grossberg & Ennio Mingolla NEURAL DYNAMICS OF PERCEPTUAL
GROUPING: TEXTURES, BOUNDARIES AND EMERGENT SEGMENTATIONS Perception &
Psychophysics 1985, 38 (2) 141-171

Gail Carpenter & Stephen Grossberg A MASSIVELY PARALLEL ARCHITECTURE
FOR A SELF-ORGANIZING NEURAL PATTERN RECOGNITION MACHINE Computer
Vision, Graphics, and Image Processing 1987, 37, 54-115 Acedmic Press,
Inc.

Stephen Grossberg & Gregory Stone NEURAL DYNAMICS OF ATTENTION
SWITCHING AND TEMPORAL ORDER INFORMATION IN SHORT TERM MEMORY Memory
and Cognition 1986, 14 (6), 451-468

Stephen Grossberg & Daniel S. Levine   NEURAL DYNAMICS OF ATTENTIONALLY
MODULATED PAVLOVIAN CONDITIONING: BLOCKING, INTER-STIMULUS INTERVAL,
AND SECONDARY REINFORCEMENT   Applied Optics 1987, 26, 5015-5030

Steven Grossberg & Nestor A. Schmajuk  NEURAL DYNAMICS OF ATTENTIONALLY
MODULATED PAVLOVIAN CONDITIONING: CONDITIONED REINFORCEMENT, INHIBITION,
AND OPPONENT PROCESSING   Psychobiology 1987, 15, 195-240

Daniel Bullock and Steven Grossberg   NEURAL DYNAMICS OF PLANNED ARM
MOVEMENTS: EMERGENT INVARIANTS AND SPEED-ACCURACY PROPERTIES DURING
TRAJECTORY FORMATION  
--
(O)((O))(((O)))((((O))))(((((O)))))(((((O)))))((((O))))(((O)))((O))(O)
(O)((O))(((              slehar@bucasb.bu.edu              )))((O))(O)
(O)((O))(((    Steve Lehar Boston University Boston MA     )))((O))(O)
(O)((O))(((    (617) 424-7035 (H)   (617) 353-6425 (W)     )))((O))(O)
(O)((O))(((O)))((((O))))(((((O)))))(((((O)))))((((O))))(((O)))((O))(O)