Task and functional similarities of PRR and PMd. A, Each monkey performed a planned-reach/saccade task, in which he held a central illuminated fixation spot with his hand and eyes as a peripheral target appeared and disappeared. After a planning period, the fixation spot disappeared, and he made either a reach or saccade to the remembered target. B, Recorded areas. Dot size corresponds to the number of cells (single and multiunit) identified in each region. C, Summary of ensemble tuning properties for each area. Top row, PMd chambers. Bottom row, PRR chambers. Left, Planning period. Right, Movement period. Each bar represents the percentage of cells tuned for ipsi-saccades, contra-saccades, ipsi-reaches, and contra-reaches, in that order, combined for both monkeys.
PRR and PMd have different oscillatory properties. A, Beta-band LFP activity in both brain areas emerges during the planning period. The spectrogram is averaged over both monkeys, and over PRR and PMd. B, However, differentiating by region (and averaging over the planning period) shows significantly stronger beta-band activity in PRR than PMd (p < 10−14, central lines are 1 percentile around the median, outer bands are 10th percentile). C, Spike-spike cross-correlation across all pairs of spikes revealed more periodicity in PRR spikes (blue) than PMd spikes (black) during the planning period.
Amid power asymmetry, partial spike-LFP coherence reveals selective cross-cortical interactions. A, A neural simulation (I-F neurons) of a brain-wide common driver, driving PRR LFPs at greater amplitude than PMd LFPs, generates significant Granger causality, even if all signals are completely synchronous and at the same phase (Δ = difference between GC in either direction, GC(LFPPRR,LFPPMd) − GC(LFPPMd,LFPPRR)). Therefore, Granger causality does not distinguish an asymmetric common drive from a case in which PRR's spikes have a selective interaction directly with PMd (right column); displayed here as an interaction shared between a subset of PRR cells and PMd, with an antisynchronous phase; * indicates significant Granger causality, p value arbitrarily small as number of simulations increase. B, The same is true for simple spike-LFP coherence (* indicates significant simple spike-LFP coherence). However, partial spike-LFP coherence (C) between PRR spikes and PMd, with respect to PRR LFPs, can indicate whether PRR spikes interact with PMd selectively and independently of the common drive (* indicates significant partial spike-LFP coherence).
Partial coherence reveals a selective, antiphase relationship between PRR spikes and PMd LFPs. A, LFP-LFP coherence between PRR and PMd (inner and outer bands are first and 10th percentile around the median). Rose plot represents almost no phase offset between LFPs (−0.12 rad, mean PRR phase lag vs PMd). B, Spikes (SPRR) cohere with LFPs within PRR (LFPPRR), at a small phase lag (0.46 rad). PMd spikes also cohere to a small extent with LFPs in PMd (−0.15 rad). C, Cross-cortical spike-LFP coherence also suggests a relationship between PRR spikes and PMd LFPs (left plot; mean phase-lag of 1.05 rad). The residual difference (in the complex plane) between the cross-cortical spike-LFP coherence, C, and the cross-cortical spike-LFP coherence that would have been predicted from plots A and B, produces the partial coherence, D. The relative phase of this interaction is nearly π radians (i.e., antiphase; rose plot; mean, 3.15 rad). Like the overall beta-band effect, PRR-PMd spike-LFP coherence disappears at movement onset (coherogram, bottom, representing PRR-PMd partial spike-LFP coherence only).
Partial spike-LFP coherence by brain state. Insets at right, Summaries of eye and hand positions during each state: yellow represents idleness; orange represents target presentation; green represents movement; black represents fixation; red represents planning. All differences are significant (p < 10−5, K-S test), except between idleness and target presentation. The idleness state is behaviorally similar to the planning and fixation states, in that the monkey's body and eyes are unmoving, and he sees no stimuli. Yet, unlike fixation and planning states, the idleness state does not elicit strong anti-synchronous beta-band activity.
Partial spike-LFP coherence by planned action. Reaches and saccades (these data are Monkey L only, who used the contralateral arm to the implant site) elicited significantly different parietofrontal partial spike-LFP coherence (K-S test, p < 10−5, corrected for multiple comparisons; inner and outer lines are first and 10th percentile around the median), although differences between ipsilateral and contralateral directions are not significant (p > 0.1, K-S test).
↵aEach monkey took part in a reach-saccade task with a planning period. Monkey L used the contralateral (right) arm to the recording chamber. Monkey R used the ipsilateral (also right) arm. For Monkey R, some trials without a planning period (rows 2 and 3) were interleaved with conventional planning trials, but these trials were excluded from analysis. Any trials without an instructed effector (5th column) were decision-making trials, in which the animal was free to decide on which effector to use.