Amount of food obtained, and effects of satiety on choice. (a) Number of pellets obtained in each trial block; average of the first three postoperative sessions, 13–15 (** p < .01, main effect of lesion). The grey area indicates the expected range of options available to a rat making no omissions: consistent responding on the lever delivering the small, certain reward of a single pellet yields 10 pellets per trial block (horizontal border of the grey area); consistent responding on the lever delivering the large, uncertain reward yields an expected number of pellets that varies with the probability in force (as shown by the diagonal border of the grey area). Optimal behaviour, to maximize the expected amount of food, is to choose the small/certain lever when the large (4-pellet) reinforcer probability is less than 0.25 and to choose the large/uncertain lever when this probability exceeds 0.25. (b) As for (a), but showing data from the final baseline postoperative sessions, 22–24 (# p < .05, lesion × probability interaction; * p < .05, simple effect of lesion). (c) Effects on choice of alternating subjects between states of hunger and satiety. The error bar is twice the SED for the three-way (lesion × hunger × probability) interaction.