The Resonant Brain: Learning, Attention, Memory, Search, and Consciousness.

Stephen Grossberg

Department of Cognitive and Neural Systems, Boston University

What is the nature of brain specialization, and how does it support adaptive behaviors in a
world filled with unexpected events? Recent neural models shed light on how this happens
by introducing new computational paradigms, such as Complementary Computing, which
clarifies the nature of brain specialization; and Laminar Computing, which clarifies why
mammalian neocortex uses laminar circuits to represent all aspects of higher intelligence.
This talk will summarize several lines of recent research that illustrate these paradigms.

One project addresses the basic question: What is an object? How does a brain learn to
recognize a complex object from multiple viewpoints while scanning a scene with eye
movements? How does a brain avoid the problem of erroneously classifying parts of different
objects together, even before it knows what the objects are? This work proposes how object
and spatial attention work together to achieve view-invariant object category learning and
recognition, and illustrates a predicted connection between the CLEARS processes of
Consciousness, Learning, Expectation, Attention, Resonance, and Synchrony.

A parallel line of research clarifies how multiple levels of brain organization, from spikes to
cognition, cooperate to realize cognitive processing properties. This model predicts how
laminar circuits of multiple cortical areas interact with primary and higher-order specific
thalamic nuclei and nonspecific thalamic nuclei to carry out predictive and attentive visual
category learning and information processing.  Predictions and recent data concerning the
interplay of gamma and beta oscillations during learning will be summarized.

How are sequences of events temporarily stored in a cognitive working memory and unitized
into learned sequential plans, or list chunks? How are they executed under variable-rate
volitional control?  Recent modeling work clarifies how variations of laminar visual cortical
circuits can realize cognitive working memory and list chunking circuits in the ventrolateral
prefrontal cortex. Properties of auditory streaming and speech perception further illustrate the
unity of cortical designs for fast learning and recognition in a changing world.

Finally, the talk will clarify why procedural memories are not conscious, and will summarize
progress towards the design of a more comprehensive system model of perception,
cognition, emotion, and action.