Cortico-thalamic organization and memory

Abstract The different patterns that can be formed by the presence or absence of signals in a number of channels may be recognized by a fixed system that forms certain functions of the signals and selects the maximum. It is also possible to have a system with a memory capacity that exhibits no pattern recognition initially but which automatically forms the particular functions required for the recognition of patterns presented to it. The human cerebral cortex appears to be such a memory system and it seems probable that cortical neurons might perform the mathematical operations required for the formation of the functions. A detailed system of cortical organization is described. This is consistent with anatomical and physiological evidence and is also capable of a learning process in which each stimulus pattern arriving at the cortical afferents produces maximum activity in the cortical efferents of pyramidal neurons that have produced the most favourable response to the stimulus in the past. The selection of this maximum activity might be accomplished through the action of the non-specific cortico-thalamic and thalam -cortical fibre tracts which have the required characteristics if their terminal synapses are respectively excitatory and inhibitory. A computing system has been used to simulate the cortico-thalamic organization. After suitable training the system successfully learnt certain types of association in the presence of random imperfections, but failed to cope with other types. The results demonstrate that, subject to certain limitations, the initial effects of imperfections may be overcome by adequate training that only involves small percentage physical changes in the state of the memory.