Gated Accumulation Model

The gated accumulation model is a descriptive framework for action selection that treats the process as a race among alternative action programs, each accumulating evidence until one crosses a threshold and triggers execution. It is gated because the evidence accumulation is controlled — not all information enters all accumulators, and the system can suppress or facilitate particular candidates.

The model addresses a limitation of simpler race models. In an ungated race, every candidate accumulates all available evidence, and the fastest accumulator wins. But biological systems do not work this way. The visual system does not prepare every possible reach simultaneously with equal vigor. The motor system prioritizes candidates that are relevant to current goals and suppresses those that are irrelevant. The gating mechanism implements this prioritization.

The gated accumulation model has been applied to explain data from perceptual decision-making, motor preparation, and response inhibition. It predicts that neural activity in motor and premotor cortex will reflect not only the currently favored action but the entire set of competing candidates, with their relative activation levels tracking the ongoing competition. This prediction has been confirmed in single-unit recordings from monkeys performing reach tasks.

The model connects action selection to the broader framework of evidence accumulation in decision theory, but with a crucial difference: it describes a system that must commit to action, not merely judge which option is best. The threshold is not a confidence criterion. It is a commitment mechanism.