Fig. 7: Tu units form transient and ramping task-response clusters.

a–c Same as Fig. 6a–c for Tu. After training, the task-related mean firing rate evolution ± SEM in Tu units reflected monotonic RP and reward surprise (n = 169 units). The bounds of the boxes represent 25th to 75th percentiles. The center indicates the median. The lower and upper whiskers represent the minimum and maximum values without outliers; outliers indicated by circles. * indicates P < 0.05 (see Supplementary Table 1 for exact P values and test details). Source data are provided as a Source Data file. d–f Fraction of single units showing significant task-excited, task-inhibited or no responses to either one or more trial types (CS0, CS50, CS100) at CS (d), during wait (e), and at US (f). Of note, the fraction of single units responsive to more than one trial type increased with training. g–i Fractions of selective and shared unit responses to different trial types (g) at CS, (h) during wait, and (i) at US in the Tu. The size of the spheres corresponds to the respective fraction of units. Green color indicates task-excited responses (+), while red color stands for task-inhibited responses (–). Note the shared coding of rewarded trial types during wait and at US. j Same as Fig. 6g for Tu. In the Tu, PCA-based hierarchical clustering revealed four task-response clusters. k–m Same as Fig. 6h–j for Tu. Mean firing rate ± SEM for each cluster identified in (j). According to their responsiveness at CS and during wait, single units were grouped into three functional clusters composed of units with (k) task-excited, transient activity at CS, (l) task-excited, ramping activity during waiting, and (m) task-inhibited responses. Both the transient- and ramping clusters displayed a monotonic RP (see Supplementary Fig. 8d–f). n indicates the number of units.