A) Schematic of the Bandit 100:0 task. B) Example of mouse behaviour, each datapoint represents the average choice proportion of 10 trials in a rolling window (F red M blue) with the grey trace indicating the true reward probability in the Bandit 100:0 task. C, D, E) Accuracy, win-stay and lose-stay probability of females vs males (n = 62F / 74M). F) Logistic regression predicting the current choice from the previous 5 choices. P-values were the result of unpaired 2-tailed t-tests, Bonferroni corrected.

A) Schematic of the fixed ratio 1 (FR1) task with example traces of a single female and male trial (F red, M blue). B, Accuracy over 24 hours, binned in 1hr bins. C, D) Accuracy and win-stay probability of females vs males (n = 62F / 74M). E) Significant correlations of the same metrics between the Bandit 100:0 and FR1 task. F) PCA analysis of 14 behavioural features from Bandit 100:0 and FR1 tasks cluster male (blue) and female (red) mice. G) Confusion matrix predicting sex from our behavioural data using accuracy, win-stay, total pokes and retrieval time from Bandit 100:0 and FR1 tasks. P-values were the result of unpaired 2-tailed t-tests, Bonferroni corrected.

A) Model of win-stay vs. accuracy in the Bandit task, varying poke reward probabilities. B) Fit between Bandit 100:0 behavioural data overlayed with simulated model data. C) Schematic of Bandit 80:20 task. D) Example of mouse behaviour, each datapoint represents the average choice proportion of 10 trials in a rolling window (F red, M blue) with the grey trace indicating the true reward probability in the Bandit 80:20 task. E, F) Accuracy and win-stay probability of females vs males (n = 62F/74M). G) Fit of Bandit 80:20 behavioural data overlayed with simulated model data. P-values were the result of unpaired 2-tailed t-tests, Bonferroni corrected.

A) Schematic of the progressive ratio 1 task. B) Example of pellets earned over 48 hours, showing “runs” of pellets at increasing poke requirements. C, D) Accuracy and median break point between males and females (n = 62F, 74M). E) Demand curves displaying average consumption of male vs. female mice at each FR (price), with the alpha (price at 50% consumption) and slope of the curve. F) The average number of runs in male and female mice. G, H) Modeled parameters alpha and slope between males and females. P-values were the result of unpaired 2-tailed t-tests, Bonferroni corrected.

A) The distribution of win-stay % on the Bandit 100:0 task (F red, M blue). Dashed lines represent the mean win-stay % of each sex. B) Schematic of sampling procedure for choosing between 4 and 60 mice per group for empirical power analyses. C) Comparison of empirical vs. analytic power for detecting a sex-difference in win-stay % at different sample sizes.