Functional MRI of auditory responses in the zebra finch forebrain reveals a hierarchical organisation based on signal strength but not selectivity

Abstract

Background: Male songbirds learn their songs from an adult tutor when they are young. A network of brain nuclei known as the 'song system' is the likely neural substrate for sensorimotor learning and production of song, but the neural networks involved in processing the auditory feedback signals necessary for song learning and maintenance remain unknown. Determining which regions show preferential responsiveness to the bird's own song (BOS) is of great importance because neurons sensitive to self-generated vocalisations could mediate this auditory feedback process. Neurons in the song nuclei and in a secondary auditory area, the caudal medial mesopallium (CMM), show selective responses to the BOS. The aim of the present study is to investigate the emergence of BOS selectivity within the network of primary auditory sub-regions in the avian pallium.

Methods and findings: Using blood oxygen level-dependent (BOLD) fMRI, we investigated neural responsiveness to natural and manipulated self-generated vocalisations and compared the selectivity for BOS and conspecific song in different sub-regions of the thalamo-recipient area Field L. Zebra finch males were exposed to conspecific song, BOS and to synthetic variations on BOS that differed in spectro-temporal and/or modulation phase structure. We found significant differences in the strength of BOLD responses between regions L2a, L2b and CMM, but no inter-stimuli differences within regions. In particular, we have shown that the overall signal strength to song and synthetic variations thereof was different within two sub-regions of Field L2: zone L2a was significantly more activated compared to the adjacent sub-region L2b.

Conclusions: Based on our results we suggest that unlike nuclei in the song system, sub-regions in the primary auditory pallium do not show selectivity for the BOS, but appear to show different levels of activity with exposure to any sound according to their place in the auditory processing stream.

Figure 1. Experimental auditory stimuli.

Spectrograms (top row) and oscillograms (bottom row) of an example of BOS and three temporal manipulated versions including reversed BOS, random BOS and BOS ripples. The spectrograms show that manipulations are restricted to each song separately. To obtain a better visualisation of the spectrograms, the maximum frequency shown is limited to 10 kHz (actual maximum frequency is 22 kHz).

Figure 2. Data acquisition.

Schematical representation of the auditory stimulation design. The entire paradigm was repeated 6 times with alternate presentation of the six different stimuli BOS, reversed BOS, random BOS, familiar CON, BOS ripples, and WN.

Figure 3. Visualisation of Field L2 on high resolution T₂-weighted SE images and regional analysis (see online edition for color figure).

The figure displays how the subfields L2a and L2b in the study of Vates et al. compare to the core of the darker ellipsoid region of our anatomical high resolution MR images that corresponds to the dense fibre track that defines sub-region L2. (Schematic illustration adapted from Vates et al. ; anatomical MR image from Poirier et al. [50]). By drawing lines rostral and caudal from L2, and a third perpendicular line, regional analysis could be performed in a caudal/ventral region that comprises L3 and NCM, a rostral/dorsal region that comprises CMM, a dorsal region that comprises L2b and a ventral region that comprises L2a. ABBREVIATIONS, Ch. O. = Optic Chiasm; CMM = caudal medial mesopallium; DLM = medial nucleus of the dorsolateral thalamus; FPL = lateral forebrain bundle; L2a, L2b, L3 = sub-regions of Field L; NCM = caudomedial nidopallium; Ov = nucleus ovoidalis; tOM = tractus occipitomesencephalicus; X = area X.

Figure 4. Average BOLD signal of one example bird (see online edition for color figure).

The images illustrate the typical activation pattern that was found in all experimental birds. The signal shown here is for all sounds presented and all brain images averaged together. The left panel shows the P-values of significant activated pixels, the right panel shows the signal strength relative to the mean signal difference. The three lines show the division in regions of interest conform Figure 3.

Figure 5. Stimulus and regional selectivity.

(A) The average BOLD signal and (B) the average number of pixels activated in the four regions of interest L2a, L2b, L3/NCM and CMM together with exposure to BOS, reversed BOS, random BOS, CON, BOS ripples and WN. All means are represented with their corresponding standard errors (SEM).

Figure 6. Regional selectivity.

This figure shows the average BOLD signal in the four regions of interest L2a, L2b, L3/NCM and CMM, for all sounds separately (A) and averaged (B). All means are represented with their corresponding standard errors (SEM). Asterisks (*) indicate significant differences.