Neural Computation
| Hi-index | 0.00 |
We investigate the effect of synaptic compensation on thedynamic behavior of an attractor neural network receiving its inputstimuli as external fields projecting on the network. It is shownhow, in the face of weakened inputs, memory performance may bepreserved by strengthening internal synaptic connections andincreasing the noise level. Yet, these compensatory changesnecessarily have adverse side effects, leading to spontaneous,stimulus-independent retrieval of stored patterns. These resultscan support Stevens' recent hypothesis that the onset ofschizophrenia is associated with frontal synaptic regeneration,occurring subsequent to the degeneration of temporal neuronsprojecting on these areas.