Figure 6.

Single switches. (A) Superimposed gratings were presented for 30, 100, or 500 ms (see dark gray box) before switching to either vertical or horizontal orientation (here and in all subsequent plots these switches are marked in black and gray, respectively). The intertrial interval, in which a blank stimulus was shown, was set to a minimum of 5 s (200 ms were recorded when a new stimulus started to obtain baseline activity, see white boxes). (B) Population tuning in response to the 6 stimulus conditions; switches to vertical along top row, horizontal switches at bottom. Same conventions as in previous figures (cf. Figs 2F, 3, 4). Temporal structure of stimulation is indicated at bottom (cf. A). Green lines and boxes mark the time window used to calculate effect size in C and maps in E (50–90 ms after switch). Insets: Bimodal distribution representing the constituent orientations of the superimposed gratings (time averages 50–150 ms after response onset). (C) Effect size. Data points specify average modulation depth (see Materials and Methods) dependent on different presentation times of the superimposed gratings (x-axis). Bars depict standard error (80 stimulus repetitions). Positive values indicate higher activity at vertically tuned pixels (blue icon) and negative values higher activity at horizontal pixels (red icon). When testing for differences between timing conditions, pooling vertical and horizontal conditions (sign inverting the latter), we found significant differences between the 30- and the 100-ms condition (pairwise two-tailed t-test: 30 vs. 100 ms P< 0.001, n = 160), but no significant differences between the 100- and 500-ms conditions (P = 0.79). (D) Time courses of global activity (unfiltered spatial averages across imaging frames) in response to the 6 stimulus conditions. Shaded areas depict the temporal structure of stimulus conditions (cf. A). (E) Correlation of the cortical activity patterns with a standard VH orientation map (see Materials and Methods and Supplementary Fig. S1). Positive values indicate similarity with activity pattern representing vertical orientation (icon on top), negative values indicate correlations to the horizontal map. On the right of each graph the 30% most active pixels across the imaged cortex are shown for each condition (time averages of 4 imaging frames, 50–90 ms after switches). Colors code preferred orientation at each pixel location (same conventions as in Fig. 2D). Upper maps visualize population tuning to a switch to vertical (outlined black), lower maps show switches to horizontal (outlined gray). Dotted lines show correlations with the additive map instead of the standard subtractive VH orientation map (see Materials and Methods) as control.