(A) Imaging slice position. The fMRI slice is a tilted coronal-horizontal slice that went through primary auditory pallium (field L), secondary auditory pallium [caudal medial nidopallium (NCM), CLM, CMM] and song control robust nucleus of the arcopallium (RA). The imaging slice is displayed in a 3D rendering volume of zebra finch brain with a coloured representation (see online edition for colour figure) of some key auditory regions and song control nuclei (Poirier et al., in preparation). The region in pink corresponds to field L; the regions in blue to song nuclei HVC and RA; the region in green to area X; and the region in yellow to the ectostriatum. These areas were delineated in a T2-weighted 3D spin-echo dataset of the head of a perfused zebra finch that was postfixed in a mixture of paraformaldehyde (4%) and Dotarem (1%), a paramagnetic MR contrast agent. (B) Comparison between in vivo anatomical MR image and in vitro histological slice in silver stain with same orientation. The darker horizontal band seen in the structural MR image corresponds to the dense fibre track that defines subregion L2a. The song control nucleus RA can be seen in both the structural MR image and in the silver stain. The border between the nidopallium and the mesopallium as determined by the lamina mesopallialis (LM) is seen in the silver stain and can be detected in the structural MR image. This landmark can be used to determine whether the observed activity was in field L or the secondary auditory area, CM. On the other hand, the actual border between NCM and the primary auditory pallium cannot be discriminated. (C) Regional analysis. The auditory regions are divided into four quadrants determined by the midline dividing the right from left hemisphere, and determined by the horizontal fibre track dividing ventral from dorsal regions. The dorsal quadrants include the dorsal part of L2a, L1 and L/L2b, while the ventral quadrants include the ventral part of L2a, L3, the medial-ventral part of L and NCM.(D) Time series BOLD signal. The curve shows the average time course of the BOLD signal for a single pixel that had a highly significant response. The first eight images are acquired during stimulation followed by eight images acquired during rest. Each time tick is approximately 5 s (see Materials and methods). The strength of the signal was the point-by-point difference between the signal during stimulation and the signal during rest as illustrated in the inset figure. It is also the area between these curves. The last points in the curve were excluded if their contribution added more noise than signal (see Materials and methods).

Anaesthesia and BOLD signal spread and strength. Left panel: the average total number of activated pixels under each anaesthetic condition. Each pixel is 97 × 97 µm. Right panel: the average strength of the signal (averaged across birds, pixels and all stimulus types) under each anaesthetic condition. The strength is measured as a percentage relative to background. The number of pixels and average strengths are calculated from all pixels showing significant BOLD activation. Significant differences are indicated by asterisks.

Stimulus and regional selectivity.(A) The average BOLD signal for the synthetic songs and for the bird's own song (BOS) relative to the BOLD signal found for conspecific song (CON) is shown for the four quadrants: left vs right hemisphere, and ventral vs dorsal region.(B) The BOLD responses to the spectrally (CON-sf) and temporally (CON-tf) filtered songs relative to the BOLD signal for CON found in the dorsal and ventral regions is shown for the three anaesthetic conditions. Error bars show the 95% confidence interval. The average strengths are calculated only for the significantly activated pixels within the ROI. Significant differences are indicated by asterisks.

Oscillograms (top row), spectrograms (middle row) and modulation spectra (bottom row) showing the spectral and temporal features found in the experimental auditory stimuli. The figure shows an example of conspecific song before (CON) and after spectral (CON-sf) and temporal filtering (CON-tf). The modulation spectrum (see online edition for colour figure) quantifies the spectrotemporal structure that is present in the sound (see Material and methods). ω_x = spectral modulations, ω_t = temporal modulations.

Average BOLD signal and anaesthetic (see online edition for colour figure). The images illustrate the typical activation patterns that were found under the three anaesthetics used in the experiment: medetomidine (top row), urethane (middle row) and isoflurane (bottom row). The signal shown here is for all sounds presented and all brain images averaged together. The left column shows the P-values of significant activated pixels, the right column shows the signal strength relative to the mean signal difference. The green line shows the dorso-ventral axis. Bilateral activation of the primary auditory pallium (field L) from the midline up to approximately 1.5 mm lateral (range: 0.8–2.3 mm) can be observed. Urethane and isoflurane show larger activation areas than medetomidine. With urethane and isoflurane, a second pole of activity is sometimes seen on the medial and ventral side of the nidopallium.

Stimulus selectivity.(A) The average responses (over all pixels) to the four different stimuli: bird's own song (BOS), conspecific song (CON), spectrally filtered CON (CON-sf) and temporally filtered CON (CON-tf), for the three different anaesthetic conditions (medetomidine, urethane and isoflurane).(B) Pair-wise differences in response between CON and the three other stimuli BOS, CON-sf and CON-tf.(C) The average responses (over all pixels) to the four different stimuli used in the experiment for medetomidine, urethane and isoflurane. Error bars show the 95% confidence interval. The average strengths are calculated only for the significantly activated pixels within the ROI. Significant differences are indicated by asterisks.

Example of regional selectivity. The BOLD difference map between spectrally filtered song and normal song (left column) and between temporally filtered song and normal song (right column) are shown for all three anaesthetic conditions for one example bird. The overall BOLD signal and significance maps for this bird were shown in Fig. 3. The colour maps on the first rows show the difference in responses in terms of percentage of the average signal and not in terms of percentage of maximum possible difference. In other words, because the signals are all below 5% (see Fig. 3), a difference of 5% would correspond to 100% modulation. The red/blue maps in the second row show the regions where the difference is above (red) and below (blue) the mean difference, which in most cases is very close to zero. The green line shows the dorso-ventral axis. These examples are typical of the data. In the medetomidine and urethane plots, one can observe the depressed response for both the temporally filtered and spectrally filtered songs in the ventral areas relative to the dorsal areas. One can also observe the overall higher responses to the temporally filtered song relative to conspecific song. The difference maps obtained under isoflurane show less selectivity but are not inconsistent with the results obtained on average. CON, conspecific song; CON-sf, spectrally filtered CON; CON-tf, temporally filtered CON.