Fig. 8: Transient and ramping units differently encode monotonic RP in Tu.

a (Top) Percentage of Tu units in the three major response clusters from Fig. 7k–m. (Bottom) Percentage of units in each cluster coding monotonic RP or showing a dominant response to one of the CS (test window indicated below each bar, see Supplementary Fig. 9a for test on complementary windows). Note that more than half of the units in the ramping cluster displayed monotonic RP during wait while only a quarter from the transient cluster did so at CS. RP and CS coding were tested via two-tailed t test with significance threshold at P < 0.05. Dashed lines indicate that the task-inhibited cluster was tested for reduction in firing rate (see “Methods”). b Example units illustrating (upper left) monotonic RP or (upper right) dominant CS50 activation at CS in two units from the transient cluster, and (lower left) monotonic RP during wait of a unit from the ramping cluster (displayed average firing rate). c Median firing rate at CS of the units in the transient cluster that encoded individually positive monotonic RP (left; n = 11 units) or non-monotonic (right; n = 26 units) responses (one-way ANOVA with Tukey post hoc comparisons). Units not individually coding monotonic RP still displayed a robust monotonic RP as a population. Thus, the transient cluster displayed monotonic RP by distributed coding. For ramping clusters, see Supplementary Fig. 9b. The bounds of the boxplot represent 25th to 75th percentiles. The lower and upper whiskers represent minimum and maximum values without outliers; outliers indicated by circles. * indicates P < 0.05 (see Supplementary Table 1 for P values and test details). Source data are provided as a Source Data file. d Scheme illustrating reward surprise and outcome discrimination (top). Units were categorized as coding for reward surprise if during CS50 rewarded trials their increase in firing rate from before to after US (d_50→R) was positive and bigger than that in CS100 trials (d₁₀₀). Units were categorized as coding for outcome discrimination if the firing rate at US was higher for rewarded than for unrewarded trials (d_{50→R vs. N} > 0). Tests were performed via two-tailed t test with a significance threshold at P < 0.05. (Bottom) Fraction of units from the three Tu clusters with reward surprise or outcome discrimination or both. Dashed lines indicate that the task-inhibited cluster was tested for the reduction in firing rate (see “Methods”). In the figure: n indicates the number of units.