Fig. 4
Changes in noise correlations of visual responses during WM. a MGS task with visual probes. The task was the same as the regular MGS task, except that brief (200 ms) visual probes flashed on screen during both the fixation and delay periods. b Changes in noise correlations of visual responses for an example pair of neurons with overlapping RFs. RF profiles (green and blue), probes (black squares), and WM location are shown in visual space relative to the fixation point (cross). All four probes and WM locations were presented at the Peak location. Bar graphs indicate change in noise correlation (IN–OUT) for the four visual probe locations; for all four probe locations shown, the noise correlation decreased during memory IN. c Changes in noise correlations of visual responses for an example pair of neurons with partially overlapping RFs. Conventions as in b. The two probes and WM locations were presented at the In-flank location. Bar graphs indicate change in noise correlation (IN–OUT) for the two visual probe locations: the noise correlation increased during memory IN compared to OUT. d For neurons with overlapping RFs (blue), noise correlations of visual responses decreased when remembering a Peak location (n = 17 pairs, p = 0.022). For neurons with partially overlapping RFs (red), noise correlations of the visual responses increased when remembering an In-flank location (n = 56 pairs, p = 0.022). e Changes in noise correlation as a function of visual probe location: for overlapping pairs of neurons, remembering a target in the Peak location decreases the noise correlation for probes appearing at Peak RF locations. Probe location (x-axis) here and in (f) is plotted according to the geometric mean of the two normalized RF profile responses at that location \(\left( {\sqrt {{\rm{RF1}}\,{\rm{response}} \ast {\rm{RF2}}\,{\rm{response}}} } \right)\). Neuron pairs are plotted in a 0.3 wide bin (x-axis); areas in bold had a significant effect of memory location on noise correlation (p < 0.05). f Changes in noise correlation as a function of visual probe location: for partially overlapping pairs of neurons, remembering a target in the In-flank location increases the noise correlation for probes appearing at In-flank RF locations
