Constraints on the design of neuromorphic circuits set by the properties of neural population codes

Abstract In the brain, information is encoded, transmitted and used to inform behaviour at level of timing action potentials distributed over population neurons. To implement neural-like systems in silico, emulate neural function, interface successfully with neuromorphic circuits need encode a way compatible that by populations neuron brain. facilitate cross-talk between engineering neuroscience, this review we first critically examine summarize emerging recent findings about how neurons transmit information. We effects on encoding readout for different features activity, namely sparseness representations, heterogeneity properties, correlations among neurons, timescales (from short long) which maintain it consistently time. Finally, elaborate these facts constrain design coding circuits. focus primarily implications designing communicate as case essential artificial biological use codes. However, also discuss implementation or emulation computation.