Figure 5.

A: Event-locking approach. top: raster diagram of a simulated neuronal response whose trial-to-trial timing is linked closely to go-cue onset. Spike trains for individual trials are sorted according to RT, shortest RT at top. Time intervals used in the analysis are labeled below. Colored circles show times on individual trials of go-cue onset (cyan), neuronal response onset (red) and movement onset (olive). bottom: trial-to-trial values reflecting cue-to-response (cyan) and response-to-movement (olive) time intervals are plotted relative to the associated single trial RT. Cue-to-response intervals exhibit no relation to the trial-to-trial variations in RTs (i.e., slope = 0), whereas response-to-movement intervals covary strongly with RT (slope = 1).

B-C: Activity of exemplar GPi units whose responses were categorized as time-locked to the go-cue (B) and movement onset (C). B top: trial-averaged SDF (black) and sorted raster (shortest RTs at top). Estimated times of response onset (red circles) relative to the times of go-cue (cyan) and movement (olive) onset. B bottom: similar to panel A, cue-to-response intervals (cyan) exhibit no relation to trial-to-trial variations in RTs, whereas response-to-movement intervals (olive) covary strongly with RTs (legend details results from linear regressions). C: a response categorized as movement-locked because cue-to-response intervals covaried closely with RTs while response-to-movement intervals did not (following conventions of panel B).

D: Temporal linkage of responses to movement declined post-MPTP. Movement-locked responses were the most common form of event locking prior to MPTP, whereas cue-locked responses became most common post-MPTP. The prevalence of intermediate locking (significant relations for both cue-response and response-movement intervals) did not change following MPTP.

E: Event locking indices (ELIs). ESIs for individual neuronal responses were distributed smoothly ranging from more cue-locked (negative values) to more movement-locked (positive values). Prior to MPTP, the distribution centered on a median ESI of 0.30, supporting a greater prevalence of locking to movement. Following MPTP, the distribution shifted leftward to favor more cue-locked responses (median −0.28; p<0.001, Kolmogorov-Smirnov test).