CORTICAL THOUGHT THEORY: A WORKING MODEL OF THE HUMAN GESTALT MECHANISM (BRAIN, CEREBRAL CORTEX, COMPUTER ARCHITECTURE, PERCEPTION)

A new unified theory of human brain function called Cortical Thought Theory (CTT) was developed which integrates the disciplines of Artificial Intelligence, neurophysiology, perceptual psychology, and theory of computation, to develop the theoretical constraints which determine the form of the solution of the computing architecture which the human brain uses to process information. The theoretical necessity of using a computing machine which uses reasoning primitives of analogy as opposed to reasoning primitives of deduction in order to accomplish the real time integration of a human-adult-sized knowledge base into the semantic analysis process is discussed. The assertion is made that the primitives of analogy must perform input classifications and comparisons like the human brain if human-like reasoning is to be performed. It was shown that the human gestalt mechanism is probably a singular mechanism of classification and comparison which is used in all domains and at all levels of abstraction of human information processing in the cortex. This gestalt mechanism is central to the operation of human memory access and human inferencing. Most significantly, it was shown that the cardinality of the gestalt feature vector set is two.;The theory also accounts for various types of human learning.;This new computing architecture was implemented by simulation and used to process audio (speech) and visual (human face images) inputs. The results were shown to be psychologically compatible with human speech and image perception. A variety of human perceptual phenomena are accounted for by this new architecture. One of these was the prediction of a new class of audio-illusions which have since been synthesized and verified as true human audio-illusions. A complete human-like information processing architecture capable of multiple levels of abstract human inferencing was developed which accounts for the human characteristic of direct memory access to the desired information or inference. A natural language example is presented which illustrates how the sentence, "John shot the buck," can be unambiguously interpreted using a seven abstraction level CTT computing architecture.