Fig. 6: Control-related (task demand dependent) selectivity arising from time-varying excitation.

a The conceptual illustration how in the proposed spatial computing framework mixed selectivity arises from a combination of item selectivity (orange vs brown curves, given by network connectivity) and the unique pattern of local excitation (green curve showing the control-related local excitation levels over time that are unique to each cortical location and not dependent on the item held in WM). b Beta bursting reflects (is anti-correlated with) the excitation in the network. Blue curves (n = 64) correspond to sites in which neurons are selective to items cued at Sample 1 (measured between 0 and 0.6 s), red curves (n = 198) describe sites with no such selective neurons. Trials with three sequential sample cues and two test cues (monkeys correctly respond at Test 2) are shown. Dotted line corresponds to the lowest beta burst rate for selective sites when not including the Test periods. Red bar denotes times where beta burst rate in non-selective sites drops below that rate (cluster based statistic, p < 0.05). Grey rectangles denote cue presentations. c PEV in spiking accounting for the information about an item presented at Sample 1 measured in neurons on sites that are item selective (blue, n = 69) and non-selective (red, n = 411) during the first sample and delay period (0–0.6 s). Blue (selective sites) and red (non-selective sites) bars denotes times where PEV is significantly above 0 (cluster based statistic, p < 0.05). An exclusion criterion was adopted where data from two recording sessions were excluded due to the lack of correct trials for one of the task conditions, which would have introduced bias in the estimate of PEV information. Shaded regions indicate standard error of the mean in all panels. Source data are provided as Source Data file.