^BBSnline

A neural network model of memory and higher cognitive functions in the cerebrum

Vogel, David A neural network model of memory and higher cognitive functions in the cerebrum.

Short Abstract:

I first review a biologically plausible, neural network model of a small region of the brain that produces associative memory. In new work, identical copies of this local network are connected to form a network that produces serial memory, operant conditioning, refabrication of memory, and fabrication of possible future events. The network distinguishes between real and imagined events, and can respond to the absence as well as the presence of stimuli. Some of these properties are achieved in ways that seem to provide instances of self﷓awareness and imagination suggesting that consciousness may emerge as an epiphenomenon in simple brains.

Long Abstract:

I first review a neural network model of a small region of the brain that produces associative memory. The model depends, unconventionally, on disinhibition of inhibitory links rather than long-term potentiation of excitatory projections. The model may be shown to have advantages over traditional models both in terms information storage capacity and biological plausibility. The simple learning and recall algorithms are independent of network architecture, and require no thresholds, finely graded synaptic strengths, or implausible stochastic smoothness. In new work, several identical copies of this local network are connected by means of massive, weak, reciprocal, excitatory projections that appropriately control the recall of information in one region by activity in another (but reduce the information storage capacities of local networks). The result is a simple network with properties that seem analogous to serial memory, classical and operant conditioning, refabrication of memory, and fabrication of possible future events. The network distinguishes between real and imagined events, and can predicate its response on the absence as well as the presence of stimuli. Some of these properties are achieved in ways that seem to provide instances of self﷓awareness and imagination suggesting that consciousness may emerge as an epiphenomenon in simple brains. The network further suggests mechanisms by which memory consolidation might occur in real brains. Given a tentative definition of consciousness, mechanisms are demonstrated by which well rehearsed behaviors might become unconscious.

Keywords:	conditioning; memory; neural network; refabrication; R-net; sparse
Subjects:	Psychology: Cognitive Psychology Computer Science: Neural Nets Philosophy: Philosophy of Mind
ID code:	bbs00001177
Deposited by:	David Vogel on 09 April 2002