Cerebral cortical communication overshadows computational energy-use, but these combine to predict synapse number

TL;DR

This paper investigates the energy constraints of neural computation and communication, revealing that communication costs dominate and predicting synapse numbers based on energy efficiency and cortical processes.

Contribution

It introduces an energy-constrained computational function distinguishing between neural computation and communication, providing new insights into energy partitioning and synapse prediction.

Findings

Communication costs are 35 times higher than computation in the brain.

The energy efficiency of neural computation differs from ideal values by a factor of 10^8.

Predicted number of synaptic transmissions needed to fire a neuron is around 2000-2500.

Abstract

Darwinian evolution tends to produce energy-efficient outcomes. On the other hand, energy limits computation, be it neural and probabilistic or digital and logical. Taking a particular energy-efficient viewpoint, we define neural computation and make use of an energy-constrained, computational function. This function can be optimized over a variable that is proportional to the number of synapses per neuron. This function also implies a specific distinction between ATP-consuming processes, especially computation \textit{per se} vs the communication processes including action potentials and transmitter release. Thus to apply this mathematical function requires an energy audit with a partitioning of energy consumption that differs from earlier work. The audit points out that, rather than the oft-quoted 20 watts of glucose available to the brain \cite{sokoloff1960metabolism,sawada2013synapse}, the fraction partitioned to cortical computation is only 0.1 watts of ATP. On the other hand at 3.5 watts, long-distance communication costs are 35-fold greater. Other novel quantifications include (i) a finding that the biological vs ideal values of neural computational efficiency differ by a factor of $10^8$ and (ii) two predictions of $N$, the number of synaptic transmissions needed to fire a neuron (2500 vs 2000).