| Accumulating evidence indicates that dendrites of many neuron types contain voltage-dependent conductances that could enable thin dendritic subtrees to behave as independently thresholded integrative units. In addition, empirical data suggest that axons and dendrites retain to a large extent their capacity for structural plasticity in the adult brain. Take together, these findings raise the possibility of a novel form of information storage in neural tissue. Our goal in the present work is to quantify the degree to which the combination of active dendritic processing and structural plasticity could contribute to the memory capacity of neural tissue, and to relate this structure-based capacity to morphological and biophysical parameters of individual neurons. We address this question in three stages.;First, we develop a function counting method to study the capacity of a special form of polynomial classifier, where dendritic nonlinearities are represented by a selective set of quadratic terms. We find that the capacity of this model resides both in its conventional synaptic weight values , and in the choice as to which product terms are included in the model, and which are not. Second, to more realistically assess the impact of active dendritic processing on synaptic integration, we build a detailed compartmental model of a CA1 pyramidal neuron. We find that the time-averaged firing rate of this cell under various stimulus conditions is closely approximated by a trivial arithmetic model involving a sum of independently computed branch responses. Third, we extend our previously developed function counting method to compute the capacity of a more realistic neuron-like abstraction. We find that long-term information storage in neural tissue could reside primarily in the selective mapping of synaptic contacts onto dendritic subregions.;In summary, this work, provides several new insights regarding the possible contributions of dendritic electrogenesis and structural plasticity to learning and information storage in the central nervous system. |
