Mereth Labs
← The research queue
Pure engineering

Feedback latency

The Latency Problem

A corrective signal only helps inside the window the brain can still act on — hundreds of milliseconds, not days.

A student does maths homework on Tuesday, turns it in on Wednesday, and gets it back graded on Friday. By the time the correction arrives the brain has moved on and the error has set. The loop is simply too long for the brain to learn from.

A latency problem with a known target

Control theory is blunt about this: a corrective signal that arrives long after the action has little authority over the next one. The brain learns fastest when the gap between attempt and signal is short enough that the attempt is still in working memory. That window is hundreds of milliseconds, not days.

tfeedback:3 days  ⟶  300 ms t_{\text{feedback}} : 3\ \text{days} \;\longrightarrow\; 300\ \text{ms}

Cutting that latency by seven orders of magnitude is an engineering job, not a pedagogical mystery. Instrument the attempt, evaluate it, return the signal at once. Every piece is mechanical and already within reach.