Interiors of neurons are occupied and organized by dynamic networks of protein polymers termed the cytoskeleton. These biomolecular networks [microtubules (MT), actin, intermediate filaments, centrioles, etc.] are coupled to membrane events and regulate cellular activities including synaptic plasticity. Models of purposeful behavior in the cytoskeleton include MT automata, in which cooperative coupling among MT subunit dipole/conformational states gives rise to computational patterns. Presently we are modelling MT automata interconnected by MAPs. learning, association and retrograde signaling. MT automata may provide a sub-neural dimension in the brain's hierarchical organization. Artificial neural nets may more closely approximate the brain by including "sub-neural" processing.These cytoskeletal networks are capable of adaptiveThe analogy between artificial neural nets and brain organization is complicated by the fact that biological neurons are not simple ffon-offtt states and are extremely complex. For example, regulation and formation of synaptic connections and their relative strengths among neurons (''synaptic plasticity") is, by itself, an adaptive'behavior which requires some form of computation. Such adaptive behavior depends on the cytoskeleton (Figure l), highly ordered parallel networks of filamentous protein polymers which dynamically organize interiors of neurons and other cells. These biomolecular networks include microtubules (MT), Microtubule associated proteins (MAPS), actin, intermediate filaments, and centrioles, as well as proteins which link to membrane proteins (ankyrin, fodrin, spectrin, etc.). Collectively, the cytoskeleton organizes cell functions such as mitosis, growth, differentiation, axoplasmic transport, trophism, synaptic formation and plasticity. Numerous models have shown
