Neuroscience criteria for guiding silicon hardware design

Identify neuroscience-derived criteria that should inform the design of efficient silicon-based neuromorphic hardware, resolving trade-offs between biological plausibility and silicon efficiency.

Background

The thesis argues that biological plausibility and silicon efficiency are distinct optimization goals, and that hardware should leverage silicon’s strengths rather than strictly mimic biology.

Determining which neuroscientific principles are most relevant for improving hardware efficiency remains an open question, with implications for architecture, learning rules, and material choices.