Quantal synaptic failures enhance performance in a minimal hippocampal model

• DOI: 10.1088/0954-898x_15_1_004
• Publication Date: 2015-08-18T16:04:02Z
• Authors: D W Sullivan; W B Levy
• Keywords: Neurons; Models, Neurological; Hippocampus; Electrophysiology; 03 medical and health sciences; 0302 clinical medicine; Memory; Synapses; Animals; Humans; Learning; Computer Simulation

Despite the fact that animals are not optimal, natural selection is an optimizing process can readily control small bits and pieces of organisms. It for this reason we need to explain certain parameters as found in Nature (e.g., number neurons their average activity) fully understand biological basis cognition. In sense, failure quantal synaptic transmission problematic because incurs information loss at each synapse which seems like a bad thing processing. However, recent work based on information–theoretic analysis single neuron suggests such losses be tolerated lead energy savings. Here study computational simulations hippocampal model function rate. We find actually enhances some indices performance when required solve hippocampally dependent task transverse patterning or it learn simple sequence. Adding random failures recurrent CA3-to-CA3 excitatory connections results more robust parametric settings. Not only part but there notable increase sequence length memory capacity. Also, combined with additional allows lower activity settings while still remaining compatible learning task. Indeed, tended towards numbers (nearly 5×104 simulations), was possible achieve rates (55–85%) minimally setting these appropriately high were successful learning. The interpreted terms previous research demonstrating randomization during training enhance by facilitating implicit state-space search interconnected neurons.