Eight male zebra finches were trained to recognize the songs of other zebra finches using a Go/NoGo operant conditioning paradigm (Gess et al., 2011). All animals were handled according to Columbia University Animal Care and Use guidelines. For each bird, two songs were selected from a group of 15 as Go stimuli and two songs were selected as NoGo stimuli. Sounds were presented through a free field speaker located directly above the bird. Each bird

was trained on a different set of four songs. Birds reached a performance level of 80% correct after Osimertinib 1,500 to 10,000 trials, after which we tested their abilities to recognize the Go and NoGo songs when they were part of auditory scenes. Auditory scenes were interleaved with trials containing only the song or only the chorus. Positive and negative outcomes for hits and false alarms were the same during testing with auditory scenes as they were during training with songs, and chorus-alone trials were reinforced randomly. Each bird performed at least 3,300 trials during behavioral testing (100 per distinct stimulus), and all testing trials were included for computing psychometric functions. Behavior and physiology

experiments were performed sequentially rather than simultaneously because (1) the low Selleckchem Alpelisib yield of simultaneous physiology and behavior would have limited the surveying of neurons in multiple auditory areas and sampling of neurons throughout the volume of each area; (2) higher-level AC BS neurons were sparse firing and difficult to isolate, further decreasing the yield of simultaneous physiology and behavior experiments; (3) higher-level AC BS neurons were responsive to

only a subset of songs, and not necessarily those that birds were trained to discriminate; and (4) in the time during which BS neurons were isolated, birds were unlikely to perform a sufficient number PD184352 (CI-1040) of trials to obtain meaningful results. Sequential behavior and physiology experiments allowed for accurate characterization of psychometric functions and high yields of well-isolated neurons at multiple stages of the auditory pathway. Behavioral and electrophysiologic experiments were performed with the same set of song, chorus and auditory scene stimuli. The songs were from 15 unfamiliar zebra finches. The zebra finch chorus was created by superimposing the songs of seven unfamiliar zebra finches that were not included in the library of individual songs. To remove energy troughs from the chorus, we applied a time-varying scaling function that was inversely proportional to the RMS energy, averaged over a sliding 50 ms window. This was done so that chorus amplitude troughs did not influence the detection of each song differently by allowing “dip listening” (Howard-Jones and Rosen, 1993). Each song was 2.0 s in duration. For both behavioral training and electrophysiology, each individual song was flanked by 0.25 s of zebra finch chorus, resulting in total durations of 2.5 s.