Theta and beta synchrony coordinate frontal eye fields and anterior cingulate cortex during sensorimotor mapping

Abstract

The frontal eye fields (FEFs) and the anterior cingulate cortex (ACC) are commonly coactivated for cognitive saccade tasks, but whether this joined activation indexes coordinated activity underlying successful guidance of sensorimotor mapping is unknown. Here we test whether ACC and FEF circuits coordinate through phase synchronization of local field potential and neural spiking activity in macaque monkeys performing memory-guided and pro- and anti-saccades. We find that FEF and ACC showed prominent synchronization at a 3-9 Hz theta and a 12-30 Hz beta frequency band during the delay and preparation periods with a strong Granger-causal influence from ACC to FEF. The strength of theta- and beta-band coherence between ACC and FEF but not variations in power predict correct task performance. Taken together, the results support a role of ACC in cognitive control of frontoparietal networks and suggest that narrow-band theta and to some extent beta rhythmic activity indexes the coordination of relevant information during periods of enhanced control demands.

Figure 1. Experimental paradigm and sample traces of simultaneously recorded activity in ACC and FEF.

(a) Schematic of the memory-guided saccade task and pro-/anti-saccade task. (b) The traces show the multiunit activity, raw LFP signal (0.5–125 Hz), theta band-pass-filtered signal (3–9 Hz), and beta band-pass-filtered signal (12–30 Hz) in a trial of memory-guided saccade task. (c) Same as in a in another memory-guided saccade task trial. The vertical green dashed lines indicate the 500 ms time intervals aligned on stimulus onset.

Figure 2. LFP power in FEF and ACC during the memory-guided saccade task.

(a) Average time-frequency spectra of the FEF LFP power across eight target locations in the memory-guided saccade task. (b) Average time frequency spectra of ACC LFP power across eight target locations in the memory-guided saccade task. The dashed lines demarcate the time of the onset and offset of the target stimulus. The black boxes on top of each graph demarcate the delay period.

Figure 3. Increased theta and beta coherence between ACC and FEF.

(a) Time-frequency spectrum of the WPLI-debiased coherence between the FEF and ACC in memory-guided saccade task for the population of ACC-FEF channel pairs (n=674). The white contour shows the area in which the subsequent analyses were performed (see Methods). The dashed lines demarcate the time of the onset and offset of the target stimulus. (b) WPLI-debiased FEF-ACC coherence spectrum of the individual monkeys in the delay period across all recording pairs (n=674). (c) Theta-band (3–9 Hz) time course of the ACC-FEF WPLI-debiased phase synchronization. (d) Beta-band (12–30 Hz) time course of the ACC-FEF WPLI-debiased phase synchronization. (e) Comparison of WPLI-debiased coherence between baseline and delay period of the contra- and ipsiversive memory-guided saccades (***P<0.001, t-test). Error bars indicate s.e.m. (f) Comparison of the overall Granger-causality influence of ACC over FEF (GACC→FEF−GFEF→ACC) between baseline and delay period of the contra- and ipsiversive memory-guided saccades (***P<0.001, **P<0.01, t-test, n=275). Error bars indicate s.e.m. (g) Comparison of beta-band WPLI-debiased coherence between baseline and delay period of the contra- and ipsiversive memory-guided saccades (**P<0.01, *P<0.05; t-test). Error bars indicate s.e.m. (h) Comparison of the beta-band overall Granger influence of ACC over FEF (GACC→FEF−GFEF→ACC) between baseline and delay period of the contra- and ipsiversive memory-guided saccades (***P<0.001; t-test). Error bars indicate s.e.m.

Figure 4. Correct task performance is dependent on field-field coherence but not on LFP power.

(a) Shown are the comparison of theta- (columns 1 and 2) and beta-band (columns 3 and 4) WPLI-debiased coherence between correct and error memory-guided saccades in the delay period of the memory-guided saccade task (400–1,100 ms following target stimulus onset) and the preparatory period (400–1,100 ms following fixation onset) of the anti-saccade task. (b) Same as in a, but show normalized ACC theta- and beta-band power. (c) Same as in a but for normalized FEF power. *P<0.05; **P<0.01; ***P<0.001, t-test. Error bars indicate s.e.m.

Figure 5. Percentage of units with significant spike-field coupling in theta and beta band.

(a) Percentage of the ACC-unit with FEF-LFP pairs showing significant changes in phase locking across the theta and beta frequency range. Comparison between baseline and delay of contraversive saccades (left), comparison between baseline and delay of ipsiversive saccades (middle), and comparison between the contra- and ipsiversive saccades in the delay period (right). (b) Same as in a, but now depicted the percentage of the FEF-unit with ACC-LFP pairs showing significant changes in phase locking across the theta and beta frequency range. Statistical testing was performed using two-sided permutation tests, such that chance level is 2.5%.

Figure 6. Pairwise phase consistencies (PPCs) across delta and theta band.

(a) PPC spike-field coherence spectrum of the population of the ACC-unit with FEF-LFP pairs across the delta and theta frequency range. Comparison between baseline and delay of contraversive saccades (left), comparison between baseline and delay of ipsiversive saccades (middle), and comparison between the contra- and ipsiversive saccades in the delay period (right). (b) PPC spike-field coherence spectrum of the population of the FEF-unit with ACC-LFP pairs across the delta and theta frequency range. Comparison between baseline and delay of contraversive saccades (left), comparison between baseline and delay of ipsiversive saccades (middle), and comparison between the contra- and ipsiversive saccades in the delay period (right). It should be noted that the same significant differences between ipsi- and contraversive trials were seen even after we compared the contra- versus ipsiversive conditions using a permutation test as described in the Methods section. Error bars denote s.e.m. in all panels. *P<0.05, paired t-test. The rose plots on the side of each graph show the histogram of the coupling angles of the population of the ACC/FEF units.

Figure 7. Pairwise phase consistencies (PPCs) across beta band.

(a) PPC spike-field coherence spectrum of the population of the ACC-unit with FEF-LFP pairs across the beta frequency range. Comparison between baseline and delay of contraversive saccades (left), comparison between baseline and delay of ipsiversive saccades (middle), and comparison between the contra- and ipsi-versive saccades in the delay period (right). (b) PPC spike-field coherence spectrum of the population of the FEF-unit with ACC-LFP pairs across the beta frequency range. Comparison between baseline and delay of contraversive saccades (left), comparison between baseline and delay of ipsiversive saccades (middle), and comparison between the contra- and ipsiversive saccades in the delay period (right). Error bars denote s.e.m. in all panels. *P<0.05, paired t-test.

Figure 8. Illustration of recorded brain area locations and summary of main interareal ACC-FEF modulations observed in this study.

(a) ACC and FEF recording locations (in red shading) shown on a rendering of a semi-inflated macaque brain. (b,c) Illustration of main interareal effects in the theta (b) and beta (c) band with the thickness of connections indicating the strength or prevalence of the effects. LFP-LFP coherence (top row) was modulated during the delay in >75% of LFP-LFP pairs in both frequencies (with increased coherence in the largest majority). Granger causality (middle row) increased during the delay for both ACC to FEF, and FEF to ACC directions (more pronounced in theta band), but the ACC to FEF Granger-causal flow was stronger than FEF to ACC Granger-causal flow at both theta and beta frequencies. Spike-LFP coherence (bottom row) increased for both directions during the delay in the theta band, but was different between delay and baseline merely in one beta frequency bin (at 22 Hz) for ACC spike to FEF LFP sites for contraversive saccades. Reduced interareal modulation prior to error commission was evident in both frequencies across different measures and is described in the text.