Control-theoretic BCI model

A control-theoretic model treats the whole BCI as a feedback loop with identifiable blocks: a plant (the decoder plus the effector dynamics, for example a cursor that integrates decoded velocity), a controller (the user's sensorimotor system generating neural commands), and a feedback path (the sensory return, usually visual). Given such a description one can apply the standard machinery — loop gain, phase and gain margins, bandwidth, disturbance rejection — to reason about why a BCI overshoots, oscillates, or feels sluggish, rather than treating these as idiosyncratic quirks.

This view yields concrete predictions. Excessive decoder gain or added loop delay reduces stability margin and provokes oscillation; heavy output smoothing buys noise rejection at the cost of bandwidth and lag; the achievable precision is set by the interaction of decoder noise and the user's corrective bandwidth, not by either alone. The obvious caveat is that the human controller is nonlinear, time-varying, and itself adaptive, so linear time-invariant analyses are local approximations; they are valuable for design intuition and for bounding behaviour, not as exact models of the brain.