Figure 4: D1R modulation of neuronal sample selectivity.

(a) Mean normalized firing rates across sample-selective BS neurons (P<0.01 in any task epoch) that were recorded either with D1R inhibition (SCH23390, n=33) or with D1R stimulation (SKF38393, n=42). Data are presented for preferred and non-preferred sample numerosities in SCH23390 and SKF38393 trials. (b) Absolute difference in normalized firing rates compared with control conditions for preferred and non-preferred sample numerosities shown for each neuron in a. Inset: modulation indices quantifying the degree to which D1R inhibition or D1R stimulation modified firing rates either for the preferred stimulus (positive values) or the non-preferred stimulus (negative values). (c) Sliding window receiver operating characteristic (ROC) analysis quantifying the discriminability of the preferred versus non-preferred sample numerosity, that is, sample-coding strength, for the BS neurons in a and b in control conditions, with D1R inhibition and with D1R stimulation (mean across neurons). Horizontal bars above the curves denote the time bins where sample-coding strength was significantly modulated by the level of D1R activation (Kruskal–Wallis test, P<0.01). (d–f) Same conventions as in a–c for sample-selective NS neurons (P<0.01 in any task epoch) that were recorded either with D1R inhibition (SCH23390, n=18) or with D1R stimulation (SKF38393, n=12). Error bars, s.e.m. across neurons; ***P<0.001.