Fig. 6.

Behavioral generalization across call types. Familiar subjects (a and b) were tested in the all-call-type tests with the same vocalizers as in previous single-call-type tests. Naive subjects (c and d) performed the discrimination between vocalizers they had never heard before. a−c One bird subject performance at discriminating two female vocalizers across five call types (color coded as in Fig. 4a) and multiple renditions within each call type (all-call-type test; a n = 313 and c n = 214 renditions). Each dot corresponds to a stimulus. The behavioral performance as measured by the odds ratio (OR, see Fig. 1) is given by the blue line. Red diamonds indicate an OR value significantly different from 1. The gray areas depict the 5−95% and 1−99% quantiles for a random odds ratio for all trials. The insert shows the OR calculated for each call type across the entire test (diamonds) with the 5−95% confidence intervals for a random interruption behavior (bars). b−d Behavioral performance of all subjects (b n = 13; d n = 7) in the all-call-type test (effect of vocalizer, LRT on GLME: b χ²₁ = 397.2, p = 0, d χ²₁ = 55, p = 1.2068×10⁻¹³; effect of call type, LRT on GLME: b χ²₆ = 247.7, p = 0, d χ²₆ = 172.4, p = 0). Black diamonds and error bars correspond to the estimates of the GLME coefficients and their 95% confidence intervals. Filled circles indicate OR significantly above chance tested by exact test as described in Methods. Call type is color coded as described in Fig. 4a. The category All corresponds to the behavioral performance (OR) when all stimuli are considered, irrespective of the call type. OR performance for the subjects shown in a and c is identified by a small green dot in b and d