Dentritic computation in the brain

Abstract

Discrete messages formed by action potentials are processed by cell bodies of neurons which act as time/space integrators of synaptic excitation elicited by this potentials. Data from neurophysiological research suggest that dendritic potential waveforms may be processed in dendro-dendritic circuits largely independent from cell bodies. The spatio-temporal arrangement and spatial frequency of dendro-dendritic synapses allow the implementation of convolution and deconvolution algorithms in such dendro-dendritic circuits. Cross correlation with periodic delta functions implemented under the form of periodic ramification of dendritic trees allows the recovery of periodic waveforms processed in dendro-dentritic networks. Considering the size of dendritic circuits which may attain lengths less than a few hundred microns and a conduction velocity which may attain two to four meters per second or more produces an expectation for the frequency of dendritic circuit much above the 1KHz limit for the frequency of cell body and axon signals. Under these conditions coding of neuronal sequenties of action potentials can only be understood taking into account structural and functional active characteristics of dendritic processes.