Sloan Center for Theoretical Neurobiology, The Salk Institute
Prediction and recurrent excitation in the neocortex
The neocortex is characterized by an extensive system of recurrent excitatory connections between neighboring neurons. The plasticity of these intracortical connections is known to be governed by a temporally asymmetric Hebbian learning rule. We describe how such a rule may allow the cortex to modify recurrent synapses for prediction of input sequences. A temporal difference learning rule for prediction used in conjunction with dendritic back-propagating action potentials is shown to reproduce the temporally asymmetric Hebbian plasticity observed physiologically. Results from biophysical simulations are provided showing that a network of cortical neurons endowed with such a learning mechanism can learn to predict input sequences and develop direction selective responses as a consequence of learning.