Pulsed Neural Nets, chap. 14, pp. 353-377. W. Maass and C. M. Bishop, eds., Cambridge, MA, MIT Press, 1998

Hebbian learning of pulse timing in the barn owl auditory system

W. Gerstner, R. Kempter, J. L. van Hemmen, and H. Wagner

Hebbian learning refers to an unsupervised correlation-based adaptation mechanism and is usually formulated in terms of mean firing rates. In this Chapter we study learning at the spike level. The learning process is driven by the temporal correlations between presynaptic spike arrival and postsynaptic firing. To explore the effect of learning on pulse coding, we consider the example of auditory processing. The auditory system of the barn owl operates with a temporal precision in the microsecond range. After delay tuning, obtained by applying the learning rule to an integrate-and fire model of an auditory neuron, the temporal precision of spike firing is in the range of 20-30 microseconds. This level of precision, typical for the early stages in auditory processing, is necessary for sound source localization.