Figure 3

Rats change their behaviour to a familiarity driven recency-based strategy after probe trials. (A) Block 1 (probes excluded): The recency D2 (M = − 0.13, SD = 0.09) significantly differed from zero (t₍₇₎ = − 4.03, p = 0.005, d = − 1.42, CI 95% − 2.41, − 0.39). The context D2 score (M = − 0.02, SD = 0.22) and side bias (M = − 0.04, SD = 0.17) did not differ from zero (t₍₇₎ = − 0.31, p = 0.76, d = − 0.11; t₍₇₎ = − 0.72, p = 0.50, d = − 0.25; respectively). (B) Angular histogram before probe angles (M = − 109.4°, SD = 85.4°, n = 8), after probe angles (M = 106.0°, SD = 50.2°, n = 8), 20 bins of 18°. The before probe angles were uniformly distributed around the circle (Rayleigh-test: Z = 0.87, p = 0.43), but the after probe angles were not (Rayleigh-test: Z = 3.71, p = 0.02). This implies that there was significant directionality to the data after probe trials, in the theoretical direction for a coherent Recency^familiar strategy (see Fig. 1F). Also, the before and after probe angles did not have a common mean direction (Watson–Williams F test: F_1,14 = 10.40, p = 0.006), suggesting a change in behaviour after experiencing probes in block 1. (C) Average before/after probes context D2 score from trials only when the test was made in the 1st context (before: M = − 0.01, SD = 0.43; after: M = 0.44, SD = 0.12). The before probes context D2 did not differ from chance (t₍₇₎ = − 0.09, p = 0.93, d = − 0.03), however the after probes context D2 score did (t₍₆₎ = 10.23, p < 0.001, d = 3.87, CI 95% 1.61, 6.11) and from the before D2 score (t₍₆₎ = − 2.63, p = 0.04, d = − 0.99, CI 95% − 1.89, − 0.05). (D) Average before/after probes context D2 scores from trials only when test was made in the 2nd context (before: M = − 0.10, SD = 0.45; after: M = − 0.12, SD = 0.23). Both did not differ from zero (t₍₇₎ = − 0.61, p = 0.56, d = − 0.22; t₍₇₎ = − 1.48, p = 0.18, d = − 0.52; respectively) nor from each other (t₍₇₎ = 0.11, p = 0.92, d = 0.04).