Fig. 1.

Operant conditioning task used for quantifying bird discrimination of vocalizer identity. a Bird testing chamber and sequence of events in one trial. Birds learn to peck a keypad to initiate the playback of 6 s sequences of vocalizations from one of two vocalizers. One of the vocalizer is associated with a reward (Re vocalizer) while the other vocalizer is not (NoRe). Twenty percent of trials trigger vocalizations from the Re vocalizer and 80% trigger vocalizations from the NoRe vocalizer. At any time, even during playback, a peck on the keypad triggers a new trial. Interrupting an Re stimulus systematically cancel the access to the reward. Birds maximize their food access by interrupting the NoRe vocalizer and by refraining from interrupting the Re vocalizer (Supplementary Movie 1). b Schematic of a subject performing the task. The bird quickly identifies the vocalizers and, accordingly, interrupts more the NoRe vocalizer stimuli (black dots aligned with interrupted (Int.)) than the Re vocalizer stimuli (yellow dots aligned with not-interrupted (Not-Int.)). The behavioral performance is measured by the odds ratio (OR) of interrupting the NoRe over the Re. OR is calculated at successive time windows (blue line) or over the entire test (large diamond on right y-axis). The right y-axis uses a logarithmic scale: ‘x2’ means NoRe vocalizations are twice more likely to be interrupted than Re vocalizations. Red diamonds indicate an OR value significantly different from 1. The gray area depicts the 5−95% quantiles for a random odds ratio for all trials. While locally the OR might not be significant (little blue diamond) although it is above the confidence interval for random interruption, these points contribute to the significance over the entire test (big red diamond). c Schematic of a subject in the control condition. This targeted interruption behavior is not observed when the two sets of vocalizations are identical and the reward is randomly assigned to the two vocalizers: OR is randomly varying around ×1