Condition-independent signal

Around movement onset a large fraction of motor-cortex population variance is captured by a component that is essentially the same for every reach condition — the condition-independent signal. It is a stereotyped, high-amplitude change in the population state that occurs whenever a movement begins, regardless of its direction, distance, or speed, and it is frequently the single largest source of variance in the data. Because it carries no information about which movement is being made, it is invisible to analyses that only look at how responses differ across conditions.

The condition-independent signal matters for both theory and BCI. Theoretically it looks like a global timing or trigger event that shifts the whole population into its movement-generating regime, consistent with the dynamical-systems view in which such a shift launches the autonomous execution dynamics. Practically it warns that the biggest thing a naive decoder sees at movement onset may be a go-related signal rather than kinematics, which is why methods such as demixed PCA that separate condition-independent from condition-dependent components are useful before attributing variance to direction or speed.