Perceptual learning performance in a forepaw-based decision-making task.

For panels B, C, D, E, F, G, H, I, J, K: n=9 WT, 11 Fmr1-/y male mice. A, Left: schema showing the behavioral setup. Right: Trail protocol and behavioral outcomes depending on the type of trial and the animal’s response. B, Total number of days spent in training for WT and Fmr1-/y mice. C, Total number of days spent in training until the criterion was met for high-salience stimuli for WT and Fmr1-/y mice. D, Total number of days spent in training until the criterion was met for low-salience stimuli for WT and Fmr1-/y mice. E, Sensitivity d’ throughout the training period for WT and Fmr1-/y mice. F, Correct choice rate for both high- and low-salience trails throughout the training period for WT and Fmr1-/y mice. G, Incorrect choice rate for both high- and low-salience trails throughout the training period for WT and Fmr1-/y mice. H, Incorrect choice rate for low-salience trials throughout the training period for WT and Fmr1-/y mice. I, Incorrect choice rate for high-salience trails throughout the training period for WT and Fmr1-/y mice. P values were computed using two-sided t-test for panels B, E, F, G, I,; Mann-Whitney test for panels C, D, H; *P < 0.05 or n.s, not significant. Created with BioRender.com.

Overall strategy and impact of prior choice during perceptual learning.

For panels A, C, D, E, F: n=9 WT, 11 Fmr1-/y male mice. A, Criterion depicting the licking strategy of the animals. B, Schema showing an example of how high impact of prior choice on the current trial affects the response during a high-salience trial (top) or low-salience trial (bottom). C, Proportion of incorrect responses in low-salience trials immediately following a correctly rewarded high-salience trial. D, Proportion of incorrect responses in high-salience trials immediately following a correctly rewarded low-salience trial. E, Proportion of correct responses in low-salience trials immediately following a correctly rewarded low-salience trial and incorrect responses in low-salience trials immediately following a correctly rewarded high-salience trial. Rates are corrected over the total number of correct and incorrect choices in low-salience trials. F, Proportion of correct responses in high-salience trials immediately following a correctly rewarded high-salience trial and incorrect responses in high-salience trials immediately following a correctly rewarded low-salience trial. Rates are corrected over the total number of correct and incorrect choices in high-salience trials. P values were computed using Mann-Whitney test for panel A,; two-sided t-test for panels C, D, E, F,; *P < 0.05 or n.s, not significant.

Tactile discrimination and categorization.

n=6 WT, 9 Fmr1-/y male mice. A, Psychometric curves for WT and Fmr1-/y mice generated based on the high-salience lick rate (rate of rightward licks) across 8 different amplitudes. Stimuli between 12–18 µm were designated as low-salience and rewarded at the left lickport, while stimuli between 20–26 µm were designated as high-salience and rewarded at the right lickport. Each amplitude was presented an average of 84 times. B, Comparison of high-salience (rightwards) responses for high-salience stimuli of 26 and 26 µm. C, Comparison of high-salience (rightwards) responses for high-salience stimuli of 26 and 22 µm. D, Comparison of high-salience (rightwards) responses for low-salience stimuli of 12 and 14 µm. E, Comparison of low-salience (leftwards) responses for low-salience stimuli of 12 and 14 µm. F, Categorization bias calculated based on the psychometric curves. G, Categorization precision computed based on the slope of the psychometric curves. H, Sensitivity d’ of the responses in all stimulus amplitudes. P values were computed using Mixed Linear Model Regression for panel A,; two-sided paired t-test for panels B-right, C-right, D, E,; two-sided t-test for panels F, G, H,; Wilcoxon signed-rank test for panels B-left, C-left,; Bonferroni correction was applied for panels B, C, D, **P < 0.01, *P < 0.05, or n.s, not significant.

Impact of categorization on across-categories and within-category discrimination.

n=6 WT, 9 Fmr1-/y male mice. Delta discrimination accuracy, calculated as the difference in the rate of high-salience licks across stimulus pairs. (A) Delta discrimination accuracy between across-category stimuli (18 µm vs 20 µm) and within-category low-salience stimulus pairs, averaged as “low salience.” B, Same as (A,) but for within-category high-salience stimulus pairs. C, Delta discrimination accuracy across all stimulus pairs within the low-salience category. D, Same as (C,) but across all stimulus pairs within the high-salience category. P values were computed using two-sided paired t-tests for panels A-left, B; Wilcoxon signed-rank test for panel A-right, and two-sided t-test for panel C,; **P < 0.01, or n.s, not significant.

Attention in perceptual decision-making under high cognitive load.

n=6 WT, 9 Fmr1-/y male mice. A, Proportion of missed trials for high-salience stimuli. B, Proportion of missed trials for low-salience stimuli. P-values were obtained using mixed-effects linear models accounting for repeated measures across stimulus amplitudes and animals for panels A, B,; Post-hoc two-sided t-test for panel B,-12µm (p = 0.06); **P < 0.01, or n.s, not significant.

Trial history integration during tactile categorization.

n=6 WT, 9 Fmr1-/y male mice. (A) Generalized linear model (GLM) coefficients from binomial regression predicting high-salience (right-lick) choices based on current stimulus amplitude, previous stimulus amplitude, previous outcome, previous choice, genotype, and their interactions. Filled circles indicate coefficients from the main-effects model; open circles indicate coefficients from the extended model including genotype × history interactions. Error bars represent ±95% confidence intervals. Dark blue lines denote statistically significant effects. (B) Psychometric functions showing the effect of the z-scored current stimulus amplitude for WT and Fmr1-/y mice. (C) Psychometric curves conditioned on the previous choice (left vs. right lick) for WT and Fmr1-/y mice. (D) Psychometric curves conditioned on previous trial outcome (reward vs. timeout). (E) Psychometric curves conditioned on the intensity of the previous stimulus (z-scored). P values were computed using Mixed Linear Model Regression. ***P < 0.001, **P < 0.01.

Perceptual learning duration for the different training phases and correct choice rates.

n=9 WT, 11 Fmr1-/y male mice for panels C-F. A, (Left) Example trace of 100 consecutive trials from an early (mixed) training session, showing the rates of correct (top) and incorrect (bottom) choices. (Right) Licking patterns from the same session for high-salience (top) and low-salience (bottom) trials. B, Same as in (A), but from a late training session of the same animal. C, Total number of days spent in the training phase where stimulus delivery is done in blocks of high- or low-salience stimuli. D, Total number of days spent in training with high-and low-salience stimuli are delivered in a pseudorandom manner. E, Correct choice rate for high-salience trails throughout the training period for WT and Fmr1-/y mice. F, Correct choice rate for low-salience trails throughout the training period for WT and Fmr1-/y mice. P values were computed using two-sided t-test for panels C, D, E, F; n.s, not significant.

Perceptual learning trajectories during training with mixed low- and high-salience trials.

n=7 WT, 9 Fmr1-/y male mice for panels. A, (Top) Average learning trajectories for WT and Fmr1-/y mice that completed the training phase, during mixed-trial training, calculated from the correct-choice rate for each training session (day) until each mouse reached the learning criterion. Shaded areas indicate 95% confidence intervals. (Bottom) Heat map of individual performance across mixed high- and low-salience sessions, showing the correct-choice rate for each session up to each mouse’s learning criterion. B, Slope of the learning trajectory for each mouse, calculated using a linear regression model. C, Performance (correct choice rate) on the first day of the mixed-trial training phase. D, Intra-individual variability in performance across training days, assessed as the standard deviation of the correct choice rate. E, Slope of the learning trajectory during the intermediate learning stage, defined as the middle 3 days of training for each animal. F, Performance (correct choice rate) during the intermediate learning stage. P values were computed using linear mixed-effects model for panel A, Mann-Whitney test for panel B, and two-sided t-test for panels C, D, E, F; n.s, not significant.

Prior strength and attention during perceptual learning.

n=9 WT, 11 Fmr1-/y male mice. A, Strength of the prior build for high-salience trials, calculated as the proportion of correct high-salience and incorrect low-salience responses following a correct high-salience response. Rates are corrected over the rate of overall correct high-salience and incorrect low-salience responses. B, Strength of the prior build for low-salience trials, calculated as the proportion of correct low-salience and incorrect high-salience responses following a correct low-salience response. Rates are corrected over the rate of overall correct low-salience and incorrect high-salience responses. C, Within-genotype comparisons of the proportion of missed high- and low-salience trials. D, Proportion of missed high-salience trials. E, Proportion of missed low-salience trials. P values were computed using two-sided t-test for panel A,; Mann-Whitney test for panels B, D, E; Wilcoxon signed-rank test for panel C-left; Paired t-test for panel C-right,; **P < 0.01, or n.s, not significant.

Perceptual learning performance during the last three days of training, for mice that were tested in tactile discrimination.

n=6 WT, 9 Fmr1-/y male mice. A, Sensitivity d’ throughout the last three days of the training period for WT and Fmr1-/y mice. B, Proportion of correct choices for high-salience trails throughout the last three days of the training period. C, Proportion of correct choices for low-salience trails throughout the last three days of the training period. D, Proportion of incorrect choices for high-salience trails throughout the last three days of the training period. E, Proportion of incorrect choices for low-salience trails throughout the last three days of the training period. F, Criterion depicting the licking strategy of the animals. G, Proportion of correct responses in high-salience trials immediately following a correctly rewarded high-salience trial. H, Proportion of correct responses in low-salience trials immediately following a correctly rewarded low-salience trial. I, Proportion of incorrect responses in high-salience trials immediately following a correctly rewarded low-salience trial. J, Proportion of incorrect responses in low-salience trials immediately following a correctly rewarded high-salience trial. K, Proportion of correct responses in high-salience trials immediately following a correctly rewarded high-salience trial and incorrect responses in high-salience trials immediately following a correctly rewarded low-salience trial. Rates are corrected over the total number of correct and incorrect choices in high-salience trials. L, Proportion of correct responses in low-salience trials immediately following a correctly rewarded low-salience trial and incorrect responses in low-salience trials immediately following a correctly rewarded high-salience trial. Rates are corrected over the total number of correct and incorrect choices in low-salience trials. M, Proportion of missed high-salience trials. N, Proportion of missed low-salience trials. P values were computed using two-sided t-test for panels A, B, C, D, E, G, H, I, J, K, L, N; Mann-Whitney test for panel F, M,; n.s, not significant.

Sensitivity, correct, and incorrect choices during categorization of high- and low-salience stimuli.

n=6 WT, 9 Fmr1-/y male mice. A, Sensitivity d’ for high-salience stimuli during categorization. B, Sensitivity d’ for low-salience stimuli during categorization. C, Correct responses for high-salience stimuli during categorization. D, Correct responses for low-salience stimuli during categorization. E, Incorrect responses for high-salience stimuli during categorization. F, Incorrect responses for low-salience stimuli during categorization. P values were computed using two-sided paired t-test for panel A,; two-sided t-test for panels B, C, D, E, F,; *P < 0.05, or n.s, not significant.