School of Physics and Astronomy, Raymond and Beverly Sackler Faculty of Exact Sciences, Tel Aviv University, Tel Aviv 69978, Israel
Abstract:
Networks of formal binary neurons can be used to construct associative memory models. Given the fact that real neurons have more complex features we try to find if other cognitive functions can be modeled by adding new degrees of freedom to the formal neuron. We show that complex features can be represented by dynamically modifying the threshold of the neuron. We explain how this allows us to model phenomena of free associative transitions between memories, temporal segmentation of an input into its memory components, phase locking (binding) and short term memory.