Figure 3.

Ongoing cortical encoding of current orientation with sequences of short 30-ms stimulus periods. Frames resolve population tuning around stimulus switch-time (green line) in 10-ms steps (−20 to +40 ms, averages over 384 repetitions of each switch, same conventions as in Fig. 2F). The icons on top depict switch conditions. Black traces to the right of each frame show time average: 0 to +40 ms. The first four conditions on the left are characterized by a change in horizontal orientation (either turned on or off). Each stimulus has different orientation content after the switch, except for the switch to blank (4th frame). The last four conditions cover corresponding changes in vertical orientation. Note that responses before the switch in 30-Hz sequences can partly include responses to previous stimuli (first dashed line Fig. 2F), particularly visible in columns #3, when the switch occurred from blank that naturally leads to a low amplitude of activity. Tuning curves obtained with voltage-sensitive dye imaging (Sharon and Grinvald 2002) are generally broader than for spike recordings (Benucci et al. 2009) and hence, superimposed gratings caused per se relatively flat distributions. Also our stimuli were not of highest contrast and evoked lower modulation depth, which in turn produces higher sensitivity to noise (Grabska-Barwinska et al. 2009). Thus, responses to superimposed gratings did not reveal a clear bimodal distribution when averaged over brief time intervals (but see Fig. 6 for longer time averages). All switches P < 0.001 in comparison to flat response; except H to VH: P = 0.97; after fitting for all residuals P > 0.99; except VH to H: res. P = 0.87; B to V: res. P = 0.98; H to B: res. P = 0.73; V to B: res. P = 0.26 (see Materials and Methods and Supplementary Tables S1–S3 for details).