Fig. 2: Standard rotational dynamics incompletely describe neural activity.
From: Reach-dependent reorientation of rotational dynamics in motor cortex
a Rotational dynamics found in motor cortex activity using jPCA, traces colored according to target angle. b Examples of linear (left) and nonlinear (right) population activity. c Histograms of population variance explained in motor cortex activity by rotational dynamics (jPCA, gray) and single condition LDSs (black). d Linear dynamical systems can be decomposed into eigenvalues, describing rotational frequencies and half-lives (left); and eigenvectors, describing where the rotational planes are in neural state space. e Changing eigenvalues causes rotational frequencies to change between conditions, without affecting the location of rotational planes. f Changing eigenvectors causes rotational planes to differ between conditions, without affecting rotational frequencies.
