The Brain’s Cognitive Dynamics: The Link between Learning, Attention, Recognition, and Consciousness

Abstract

The processes whereby our brains continue to learn about a changing world in a stable fashion throughout life are proposed to lead to conscious experiences. These processes include the learning of top-down expectations, the matching of these expectations against bottom-up data, the focusing of attention upon the expected clusters of information, and the development of resonant states between bottom-up and top-down processes as they reach a predictive and attentive consensus between what is expected and what is there in the outside world. It is suggested that all conscious states in the brain are resonant states, and that these resonant states trigger learning of sensory and cognitive representations when they amplify and synchronize distributed neural signals that are bound by the resonance. Thus, processes of learning, intention, attention, synchronization, and consciousness are intimately bound up together. The name Adaptive Resonance Theory, or ART, summarizes the predicted link between these processes. Illustrative psychophysical and neurobiological data have been explained and quantitatively simulated using these concepts in the areas of early vision, visual object recognition, auditory streaming, and speech perception, among others. It is noted how these mechanisms seem to be realized by known laminar circuits of the visual cortex. In particular, they seem to be operative at all levels of the visual system. Indeed, the mammalian neocortex, which is the seat of higher biological intelligence in all modalities, exhibits a remarkably uniform laminar architecture, with six characteristic layers and sublamina. These known laminar ART, or LAMINART, models illustrate the emerging paradigm of Laminar Computing which is attempting to answer the fundamental question: How does laminar computing give rise to biological intelligence? These laminar circuits also illustrate the fact that, in a rapidly growing number of examples, an individual model can quantitatively simulate the recorded dynamics of identified neurons in anatomically characterized circuits and the behaviors that they control. In this precise sense, the classical Mind/Body problem is starting to get solved. It is further noted that many parallel processing streams of the brain often compute properties that are complementary to each other, much as a lock fits a key or the pieces of a puzzle fit together. Hierarchical and parallel interactions within and between these processing streams can overcome their complementary deficiencies by generating emergent properties that compute complete information about a prescribed form of intelligent behavior. This emerging paradigm of Complementary Computing is proposed to be a better paradigm for understanding biological intelligence than various previous proposals, such as the postulate of independent modules that are specialized to carry out prescribed intelligent tasks. Complementary computing is illustrated by the fact that sensory and cognitive processing in the What processing stream of the brain, that passes through cortical areas V1-V2-V4-IT on the way to prefrontal cortex, obey top-down matching and learning laws that are often complementary to those used for spatial and motor processing in the brain’s Where/How processing stream, that passes through cortical areas V1-MT-MST-PPC on the way to prefrontal cortex. These complementary properties enable sensory and cognitive representations to maintain their stability as we learn more about the world, while allowing spatial and motor representations to forget learned maps and gains that are no longer appropriate as our bodies develop and grow from infanthood to adulthood. Procedural memories are proposed to be unconscious because the inhibitory matching process that supports their spatial and motor processes cannot lead to resonance. Because ART principles and mechanisms clarify how incremental learning can occur autonomously without a loss of stability under both unsupervised and supervised conditions in response to a rapidly changing world, algorithms based on ART have been used in a wide range of applications in science and technology.