These experiments were conducted as follows...

SUBJECTS:
Electrophysiological recordings were carried out in 6 adult zebra finches (Taeniopygia guttata) from the
MIT zebra finch breeding facility. Birds ranged from 120-400 days post-hatch in age. 
Craniotomies were performed and headposts implanted on birds in order to prepare birds 
for head-fixation and electrophysiological recordings.


HEAD FIXATION AND E-PHYS RECORDINGS:
In advance of recording, birds were gradually habituated to the recording setup over the course of 3 days.
Neuropixel probes (Phase 3a [Jun et al., 2017]) were used together with OpenEphys software 
(https://open-ephys.org/) for data acquisition, aloing with the PXIe system (National Instrument 
NI-PXIe-1071). Extracellular signals were amplified and filtered (0.3-10 kHz) to obtain action potential 
data. Craniotomies were made so that the center of the cranial aperture was centered above NCM (see 
surgery protocol), and the probe was lowered approximately 2000 𝜇m below the surface of the brain, until 
auditory responses were visible up to the 200th contact from the bottom of the probe (approximately 
1500 𝜇m from the probe tip).

Experiments took place on an optical table, surrounded by a home-made sound- reduction enclosure.


EXPERIMENTAL STRUCTURE:
This experiment involved 2 songs (consisting of 5 syllables each). The songs were played during two 
separate phases of the experiment. Each song was played with shuffled inter-bout intervals, consisting 
of 50, 100, 150, 200, 250, 300, 400, 500, 750, 1000 ms. Over the course of the experiment, each of 
these intervals was repeated 10 times. 


AUDITORY STIMULATION:
Auditory stimuli were synthesized using custom MATLAB code and stimulus components from the 
Tchernikovskii lab zebra finch song archive [Tchernichovski et al., 2004], 
http://ofer.sci.ccny.cuny.edu/songs. Perturbations were introduced by inserting harmonic stacks 
(in the case of harmonic stack substitution), eliminating all sound (in the case of silence 
substitutions), or inserting spectrally matched noise (in the case of noise substitution). All 
substitutions were made using cosine-squared ramps to minimize edge artifacts. Spectrally-matched 
noise was generated by randomizing the phase of the standard, unperturbed birdsong. Sound amplitudes 
were matched so that substitutions (aside from silence) were as loud as the standard song piece that 
they replaced.

Sounds were played through a PyleDriverPro speaker, with SPL of 75 dB at 20 cm. This intensity was 
matched to the sound level of an adult zebra finch singing at the same distance.