Table 3. Overview.

	Relationships between GPi and VLa Hypothesized versus observed
	Gating [15; 41]	Rebound [18; 22]	Observed
Incidence of increase- versus decrease-type responses	Inverted. GPi increases proportional to VLa decreases	Direct. GPi increases proportional to VLa increases	Direct. Similar rates of increases found in GPi and VLa (see Table 1)
Timing of movement-related responses	GPi leads VLa by about 2 ms	GPi leads VLa by about 100 ms	GPi lags VLa by about 43 ms (see Fig 3A)
GPi-VLa spike-spike CCFs	Negative CCF peaks at short latency in subpopulation of pairs	CCF peaks negative at short latency followed by positive at long latency.	CCF peaks are rare (at noise rate). No net bias toward negative or positive peaks (see Fig 4A–4F and Table 2)
GPi-VLa FR correlations (NCs)	Negative NCs are larger and/or more common	Positive NCs are larger and/or more common	Significant NCs are rare w/ no net bias toward positive or negative relationships (see Fig 4G–4H and Table 2)
VLa FR following burst in GPi firing	VLa FR decreases at short latency (about 2 ms)	VLa FR decreases following burst onset	Long-lasting (>200 ms) VLa decrease w/ pause part of GPi burst-pause complex (see Fig 5)
VLa FR following pauses in GPi firing	VLa FR increases at short latency (about 2 ms)	VLa FR increases at short latency

Abbreviations: CCF, cross-correlation function; FR, firing rate; GPi, globus pallidus-internus; NC, noise correlation; VLa, ventrolateral anterior nucleus; w/, with