Abstract
Neural computations support stable behavior despite relying on many dynamically changing biological processes. One such process is representational drift (RD), in which neurons' responses change over the timescale of minutes to weeks, while perception and behavior remain unchanged. Generally, RD is believed to be caused by changes in synaptic weights, which alter individual neurons' tuning properties. Since these changes alter the population readout, they require adaptation of downstream areas to maintain stable function, a costly and non-local problem. Here we propose that much of the observed drift phenomena can be explained by a simpler mechanism: changes in the excitability of cells without changes in synaptic weights. We show that such excitability changes can change the apparent tuning of neurons without requiring adaptation of population readouts in downstream areas. We use spike coding networks (SCN) to show that the extent of these tuning shifts matches experimentally observed changes. Moreover, specific decoders trained on one excitability setting perform poorly on others, while a general decoder can perform close to optimal across excitability changes if trained across many days. Our work proposes a simple mechanism without synaptic plasticity that explains experimentally observed RD, while downstream decoding and, by extension, behavior remain stable.