Fitts's law and throughput

Fitts's law is an empirical relationship for pointing tasks: the time to move to a target grows with the log of the ratio of movement distance to target width, so the difficulty of an acquisition is captured by an index of difficulty measured in bits. Dividing that index by the movement time yields a throughput in bits per second — a single number that summarizes how fast and accurately a pointing interface operates, and that (under the standard ISO pointing-device methodology) is comparable across devices and studies. Applying it to BCI lets cursor decoders be benchmarked against mice, eye trackers, and each other on a common information scale.

Throughput is a more honest motor-BCI metric than raw accuracy or hit rate because it jointly penalizes slowness and imprecision and cannot be inflated by making targets huge or trials leisurely. It is complementary to information transfer rate: ITR is oriented toward discrete selection interfaces (spellers), whereas Fitts throughput is the natural currency for continuous cursor control. Reported throughputs for state-of-the-art intracortical cursors are on the order of a few bits per second, still below a hand-operated mouse but well within the range of practical use.