Figure 1.

Ferrets are sensitive to repetitions embedded in mixtures. A, Ferrets were trained to respond to sound repetition by licking a waterspout. B, Schematic of the go/no-go task and spectrograms of repetition embedded noise stimuli from an example behavioral trial. Animals were exposed to the combination (bottom spectrogram) of the following two overlapping streams: a foreground stream containing a target sample (top); and a background stream, a nonrepeating sequence of noise samples (middle). In this example, the target sample (orange boxes, bottom) starts repeating after three random noise samples (gray boxes). The gray dashed line marks the first occurrence of the target sample (pale orange box), which is included in the random segment for analysis. The transition between random and repeating segment is marked by the orange dashed line and occurs when the target sample is first repeated. Animals were trained to withhold licking during the random segment (4–6 s). To receive a water reward, they had to lick the waterspout following repetition onset. C, Distribution of DI across behavior sessions for ferret O (n = 636; mean, 0.60; gray arrow) and ferret H (n = 504; mean, 0.64; black arrow) after training was completed. For both animals, average performance was significantly better than average performance computed after shuffling response times across trials (mean shuffled DI = 0.53 for both animals; dashed line, p < 0.0001). D, Mean reaction time relative to the onset of each noise sample slot in the random segment. Reaction times for target samples appearing in the random phase were identical to nontarget samples appearing at the same time relative to the onset of repetition, indicating that animals did not preferentially respond to the identity of the target sample. Shading indicates SEM. Only data from 250 ms noise samples are shown.