Extended Data Fig. 1: Sex differences in motivation to perform a value-based decision-making task in mice.
From: A neural substrate of sex-dependent modulation of motivation
a. To quantify sex differences in the distribution of trial initiation latencies, we fit each animal’s data with a shifted inverse gaussian distribution. Top: probability distribution function for the shifted inverse gaussian. Middle: Cumulative distribution of trial initiation latencies (shaded line) and fit (dashed line) for example females (orange) and males (green). Bottom: Probability distribution function of trial initiation latencies (shading) and fit (dashed line) for example females (orange) and males (green) for 1–10 seconds. b. Parameter estimates for each mouse. Translucent circles are individual animals, crosses are mean and SEM across males or females. Comparisons between males and females were performed with 2-sided Wilcoxon rank sum tests. μ: Z = −4.44, p = 9.04 × 10−6; λ: Z = −5.46, p = 4.65 × 10−8; θ: Z = 5.19, p = 2.06 × 10−7. c. Histograms of trial initiation latencies binned in log-space for males and females following rewarded and unrewarded trials. Top panels are the mean and SEM across animals, and the bottom panels show the histograms separately for each animal. Trials preceded by reward or no reward are plotted separately. d. Trial initiation latencies binned as short (<1 s), medium (1–10 s) and long (>10 s) for previously rewarded and unrewarded trials for males and females. Males had more short trials than females (2-sided Wilcoxon rank sum tests, reward: Z = −3.90, p = 9.7 × 10−5; unrewarded: Z = −6.01, p = 1.9 × 10−9). Females had more medium trials than males (2-sided Wilcoxon rank sum tests, rewarded: Z = 4.20, p = 2.6 × 10−5; unrewarded: Z = 5.62, p = 1.9 × 10−8). Females had more long trials only following unrewarded outcomes (2-sided Wilcoxon rank sum tests, rewarded: Z = 1.37, p = 0.17; unrewarded: Z = 4.21, p = 2.5 × 10−5). e. Daily fluctuations in weight affected value modulation of trial initiation latencies similarly in males and females (mixed effects regression of latency with sex, weight, relative chosen value, trial number and their interactions as fixed effects; see Supplementary Table 5 for details. Significance of coefficients was assessed with F tests: weight: F(1,55.95) = 8.87, p = 0.004; sex:weight: F(1,55.95) = 13.03, p = 6.55 × 10−4; relative_chosen_value:weight: F(3,135.8)=28.57, p = 2.20 × 10−14; sex:relative_chosen_value:weight: F(3,135.8) = 1.60, p = 0.19. For each mouse, sessions were binned in terciles of weight and trials were divided into quantile bins of relative chosen value. Trial initiation latencies were then averaged for each bin. n = 35 females and 36 males. One male maintained constant weight throughout the experiment and was excluded from this plot. f. Sessions were divided based on weight, and trial initiation latencies were binned in 4 quantile bins of relative chosen value and averaged for each animal. g. Average trial initiation latency versus relative chosen value in 30 min bins for males and females. Trials were binned based on time in session as well as in quantiles of relative chosen value and averaged for each mouse. Trial number significantly affected trial initiation latencies but there was no effect of trial on the interaction between relative chosen value and sex (see Supplementary Table 5 for details; trial: F(1,66.20) = 392.34, p = 1.60 × 10−29; sex:trial: F(1,66.20) = 4.08, p = 0.05; sex:trial:relative_chosen_value: F(3,100.02) = 1.45, p = 0.23). h. Following unrewarded trials, only females modulated their trial initiation latencies based on the upcoming stay versus switch decision. There was a significant effect of sex, stay/switch and a significant interaction between sex and stay/switch, reflecting greater modulation of trial initiation latency by whether or not the upcoming trial was a stay or switch trial in females compared to males (mixed-effects regression: latency ~ sex + stay + sex:stay + (1 + stay|subject); sex: F(1,424000) = 857.46, p = 6.67 × 10−188; stay: F(1,37.27) = 13.02, p = 8.99 × 10−4; sex:stay: F(1,23373) = 9.54, p = 0.002). Females were significantly slower than males to initiate stay and switch trials (2-sided Wilcoxon rank sum tests; switch male vs. female: Z = 4.55, p = 5.3 × 10−6; stay male vs. female: Z = 4.83, p = 1.34 × 10−6). In females, trial initiation latencies for switch trials were slower than stay trials. There was no difference between trial types in males (2-sided Wilcoxon signed rank tests. Female: Z = 3.52, p = 4.29 × 10−4; Male: Z = 1.26, p = 0.21). ** p < 0.01, *** p < 0.001; n = 37 males, 35 females. b–h. Error bars or shading are standard error of the mean.
