Page 53 - Artificial Intelligence in the Age of Neural Networks and Brain Computing

40 CHAPTER 2 Mind, Brain, Autonomous Agents, and Mental Disorders

during source segregation; speaker normalization that enables language learning
from adults after a critical period of babbled sounds by a child; cognitive-
emotional dynamics that direct motivated attention toward valued goals; and adap-
tive sensory-motor control circuits, such as those that coordinate predictive smooth
pursuit and saccadic eye movements, and coordinate looking and reaching move-
ments. Brain regions that are functionally described include visual and auditory
neocortex; specific and nonspecific thalamic nuclei; inferotemporal, parietal, pre-
frontal, entorhinal, hippocampal, parahippocampal, perirhinal, and motor cortices;
frontal eye fields; supplementary eye fields; amygdala; basal ganglia; cerebellum;
and superior colliculus.
ART does not, however, describe many spatial and motor behaviors. These pro-
cesses typically use different matching and learning laws. ART is thus not “a theory
of everything.”.

6. VECTOR ASSOCIATIVE MAPS FOR SPATIAL
REPRESENTATION AND ACTION
Complementary spatial/motor processes in the Where-stream often use inhibitory
matching and mismatch-based learning to continually update spatial maps and
sensory-motor gains that can effectively control our changing bodies throughout
life. Inhibitory matching can take place between representations of where we
want to move and where we are now (Fig. 2.6), so that when we arrive at where
we want to be, the match equals zero. Inhibitory matching by the vector associative
map, or VAM, matching rule thus cannot solve the stability-plasticity dilemma
[21,22]. That is why spatial and motor representations cannot support conscious
qualia. Instead, spatial maps and motor gains experience catastrophic forgetting as
they learn how to accurately control our changing bodies throughout life.
Together these complementary processes create a self-stabilizing perceptual/
cognitive front end in the What-stream for learning about the world and becoming
conscious of it, while it intelligently commands more labile spatial/motor processes
in the Where-stream that control our changing bodies.

7. HOMOLOGOUS LAMINAR CORTICAL CIRCUITS FOR ALL
BIOLOGICAL INTELLIGENCE: BEYOND BAYES
The second computational paradigm is called laminar computing [1,13,23,24].
Laminar computing describes how the cerebral cortex is organized into layered cir-
cuits whose specializations support all higher-order biological intelligence. Indeed,
the laminar circuits of cerebral cortex seem to realize a revolutionary computational
synthesis of the best properties of feedforward and feedback processing, digital and
analog processing, and data-driven bottom-up processing and hypothesis-driven