The role of adaptation in neural coding

The concept of 'neural coding' supposes that neural firing patterns in some sense represent some external correlate, whether sensory, motor, or structural knowledge about the world. While the implied existence of a one-to-one mapping between external referents and neural firing has been useful, the prevalence of adaptation challenges this. Adaptation provides neural responses with dynamics on timescales that range from milliseconds up to many seconds. These timescales are highly relevant for sensory experience in the natural world, in which local statistical properties of inputs change continuously, and are additionally altered by active sensing. Adaptation has a number of consequences for coding: it creates short-term history dependence; it engenders complex feature selectivity that is time-varying; and it can serve to enhance information representation in dynamic environments. Considering how to best incorporate adaptation into neural models exposes a fundamental dichotomy in approaches to the description of neural systems: ones that take an explicitly `coding' perspective versus ones that describe the system's dynamics. Here we discuss the pros and cons of different approaches to the modeling of adaptive dynamics.