Seminários em Neurociências 2011.1

Segunda-feira 18/04/2011 10:50h Sala 4 – Escola de Ciência e Tecnologia


Synaptic plasticity dynamics on hippocampal neurons during sleep: a neural network model

Wilfredo Blanco (PhD candidate, UFRN & ELSIINN)

Hippocampal electrical activity has different statistical features during non-dreaming sleep ("slow-wave" sleep, SWS) and dreaming sleep ("rapid-eye-movement" sleep, REM). SWS is marked by high amplitude and low frequency neural oscillations (delta waves, < 4Hz), while REM is characterized by faster oscillations (>5 Hz) (Gervasoni, Lin, Ribeiro, Soares, Pantoja, & L., 2004). By the same token, neurons tend to fire more during REM than during SWS in hippocampus and cortex, two important structures for learning and memory. To investigate how state-dependent variations in firing rates affect the synaptic configuration (Abraham & Robins, 2005), we studied the plasticity dynamics of an Artificial Neural Network (ANN) comprising an excitatory population of simple stochastic binary units. Inputs to the ANN were independent homogeneous Poisson processes with the mean firing rates of actual hippocampal spike trains, recorded from sleeping rats. Data were sorted from the lowest and highest rates recorded, to simulate the extreme SWS (low) and REM (high) conditions of the cycle. The ANN synaptic weights were randomly initialized and a stable Hebbian learning rule (van, Bi, & Turrigiano, 2000) was applied to update the synaptic weights over time. We observed that during SWS the synaptic configuration was largely maintained, with a small but significant global depression of synaptic weights. During REM, the synaptic configuration underwent significant restructuring, displaying a combination of depression and potentiation of the strong and weak synaptic weights, respectively. Our results suggest that the different spiking regimes across the sleep cycle promote global synaptic downscaling during SWS and synaptic homeostasis during REM.

Notícia cadastrada em: 12/04/2011 14:33
SIGAA | Superintendência de Tecnologia da Informação - | | Copyright © 2006-2023 - UFRN -