2.11.3 Implications for computing

Fig. 1.6 also reveals some secrets of the effective biological computing.

  • biology makes the ”weighted summing” of neuron’s synaptic inputs simultaneously, in one single operation making multiplication and integration, furthermore selecting the time window and its effective synaptic inputs

  • the heavily used neuronal information ”is stored, as it should be, in every circuit” [144]

  • ”information stored directly at a synapse can be retrieved directly” [144]

  • part of the information (in ’volatile’ memory) is stored only for the period when it might be needed (a real temporary cache)

  • ”Computing” is much shorter than ”Delivering”

  • asynchronous; the operation time varies (no pipelining)

  • ”Send only information that is needed, and send it as slowly as possible” [144]

  • using voltage temporal gradients enables transferring more information and functionality; for example, synchronization [63]

  • for simulation: there is no need to send and integrate spike shapes, only the time of arrival/sending