Page 52 - Artificial Intelligence in the Age of Neural Networks and Brain Computing

5. Adaptive Resonance Theory 39

These matching and learning laws have been articulated as part of ART, which
has been systematically developed since it was first reported in 1976 [14,15].
ART is a cognitive and neural theory of how the brain autonomously learns to attend,
recognize, and predict objects and events in a changing world. ART is currently the
most highly developed cognitive and neural theory available, with the broadest
explanatory and predictive range. Central to ART’s predictive power is its ability
to carry out fast, incremental, and stable unsupervised and supervised learning in
response to a changing world. ART specifies mechanistic links between processes
of Consciousness, Learning, Expectation, Attention, Resonance, and Synchrony
(the CLEARS processes) during both unsupervised and supervised learning. I have
predicted that all brains that can solve the stability-plasticity dilemma do so using
these predicted links between CLEARS processes. Indeed, my 41-year-old prediction
that “all conscious states are resonant states” is consistent with all the data that I know,
and has helped to explain many data about consciousness, as will be briefly noted
below.
ART hereby contributes to functional and mechanistic explanations of such
diverse topics as 3D vision and figure-ground perception in natural scenes; optic
flowebased navigation in natural scenes toward goals around obstacles and spatial
navigation in the dark; invariant object and scenic gist learning, recognition, and
search; prototype, surface, and boundary attention; gamma and beta oscillations dur-
ing cognitive dynamics; learning of entorhinal grid cells and hippocampal place
cells, including the use of homologous spatial and temporal mechanisms in the
medial entorhinal-hippocampal system for spatial navigation and the lateral stream
for adaptively timed cognitive-emotional learning; breakdowns in attentive vigi-
lance during autism, medial temporal amnesia, and Alzheimer’s disease; social
cognitive abilities such as the learning of joint attention and the use of tools from
a teacher, despite the different coordinate systems of the teacher and learner; a uni-
fied circuit design for all item-order-rank working memories that enable stable
learning of recognition categories, plans, and expectations for the representation
and control of sequences of linguistic, spatial, and motor information; conscious
speech percepts that are influenced by future context; auditory streaming in noise

=
“novel events are arousing.” The vigilance parameter r determines how bad a match will be
tolerated before a burst of nonspecific arousal is triggered. This arousal burst triggers a
memory search for a better-matching category, as follows: Arousal resets F 2 by inhibiting Y.
(D) After Y is inhibited, X is reinstated and Y stays inhibited as X activates a different
category, that is represented by a different activity pattern Y*,at F 2 . Search continues until a
better matching, or novel, category is selected. When search ends, an attentive resonance
triggers learning of the attended data in adaptive weights within both the bottom-up and top-
down pathways. As learning stabilizes, inputs I can activate their globally best-matching
categories directly through the adaptive filter, without activating the orienting system.
Adapted with permission from G. Carpenter, S. Grossberg, Normal and amnesic learning, recognition, and
memory by a neural model of cortico-hippocampal interactions, Trends in Neurosciences 16 (1993) 131e137.