Fig. 2

Nlgn3^−/y and Nlgn3^R451C mice display normal premise stimulus pair learning. a Optimized visual stimuli used for premise stimulus pairs. b Trials and c errors to criterion during discrimination learning for the 4 premise stimulus pairs (stage 2). Data presented as medians ± 95% CI. b Mice required more trials to reach criterion on C⁺D⁻ and D⁺E⁻ than A⁺B⁻ (C⁺D⁻: p < 0.001, aMD = 488.552, 95% CI = 370.371, 606.732; D⁺E⁻: p < 0.001, aMD = 407.035, 95% CI = 288.854, 525.215). c Mice made more errors to criterion on pairs B⁺C⁻, C⁺D⁻, and D⁺E⁻ than A⁺B⁻ (B⁺C⁻: p = 0.038, aMD = 51.241, 95% CI = 2.94, 99.540; C⁺D⁻: p < 0.001, aMD = 203.035, 95% CI = 154.735, 251.334; D⁺E⁻: p < 0.001, aMD = 153.379, 95% CI = 105.080, 201.679). d Effect of genotype on the likelihood of responding correctly to trials (pseudorandom first-presentation) during premise pair learning relative to WT (represented by dotted line; 1 = no difference to WT). A⁺B⁻ (Nlgn3^R451C: p < 0.001, OR = 1.69, 95% CI = 1.314, 2.165; Nlgn3^−/y: p = 0.009, OR = 1.379, 95% CI = 1.083, 1.755). B⁺C⁻ (Nlgn3^R451C p = 0.042, OR = 1.416, 95% CI = 1.013, 1.979). e Effect of stimulus pair on the likelihood of responding correctly to trials during premise pair learning relative to A⁺B⁻ (represented by dotted line; 1 = no difference to A⁺B⁻). See Additional file 1: Table S3 for statistics. # denotes significant difference between stimulus pairs (^#p < 0.05; ^##p < 0.01). * denotes significant difference between genotypes (*p < 0.05, **p < 0.01)