Figure 1. Neurexin1α disruption leads to deficits in value-based selection of actions.

(A) Schematic of trial structure wherein mice perform repeated self-initiated trials with contrasting reward volumes associated with each port. Animals were tested at four relative reward ratios across high (P_rew = 0.75) and low (P_rew = 0.4) reinforcement rates. See Materials and methods for details. (B) Both probability of reinforcement and volume contrast modulate the probability at which mice select the large reward option. Nrxn1α KOs (blue, n = 10) select the high benefit alternative at a lower rate than their WT littermates (gray, n = 11) across reward environments (three-way RM ANOVA). (C and D) For both WT and KO animals, the relative magnitude of rewarded outcome has a significant effect on the stay-probability for that alternative. (E) The relative reward-stay (RRS), which quantifies the relative tendency of animals to repeat choices after specific outcomes, was sensitive to relative magnitude of rewards but not reward probability. In comparison to WT littermates, Nrxn1α KOs less dynamically alter their choice behavior after large reward outcomes than small reward outcomes (three-way RM ANOVA). (F and G) The RRS is a significant predictor of session performance for both WT and KO mice at both rates of reinforcement. Note RRS is a better predictor of task performance at high reinforcement rates, reflecting the preponderance of unrewarded outcomes in low reinforcement conditions. All data represented as mean ± SEM.

Figure 1—figure supplement 1. Additional Behavioral Analyses in Nrxn1a KO mice.

(A) Schematic of visual discrimination task. Mice acquired a simple goal-direction contingency over repeated sessions. (B) Task engagement was measured as the total number of registered trial initiations. Nrxn1α KO mice exhibit no difference in task engagement from wild-type littermates during acquisition of visual discrimination. (C) Performance was measured as the proportion of trial initiations that resulted in the selection of the lit port. Nrxn1α mice exhibit no deficit in visual discrimination as compared to wild-type littermates. (D) Nrxn1α KO animals exhibit extended choice latencies throughout the last 3 days of task acquisition. (E–H) Logistic regression coefficients for Nrxn1α wild-type and knockout mice (ΔReward = 12 μL). The influence of past choice and reward outcome is heavily discounted after the t−1 trial. (I) There is a trend toward lower adaptability measures in Nrxn1α knockout mice in the relative reward reversal paradigm (see Figure 1A). (J) Nrxn1α KO animals exhibit similar performance to wild-type controls in extra-dimensional set shift where reward target switches from visual cue to ego-centric spatial cue. (K) Nrxn1α KO animals exhibit similar performance to wild-type controls in egocentric spatial reversal task. (B–D and I–K analyzed by two-way RM ANOVA). All data represented as mean ± SEM.