[Supporting Information]

BIOLOGICAL SCIENCES / NEUROSCIENCE
Functional MRI of the zebra finch brain during song stimulation suggests a lateralized response topography

Henning U. Voss^*,, Karsten Tabelow, Jörg Polzehl, Ofer Tchernichovski, Kristen K. Maul, Delanthi Salgado-Commissariat^¶, Douglas Ballon^*, and Santosh A. Helekar^¶

*Citigroup Biomedical Imaging Center, Weill Medical College of Cornell University, 516 East 72nd Street, New York, NY 10021; Weierstrass Institute for Applied Analysis and Stochastics, Mohrenstrasse 39, 10117 Berlin, Germany; Department of Biology, City College of New York, 138th Street and Convent Avenue, New York, NY 10031; and ^¶The Methodist Neurological Institute, 6560 Fannin Street, Suite 902, Houston, TX 77030

Edited by Dale Purves, Duke University Medical Center, Durham, NC, and approved May 8, 2007 (received for review December 22, 2006)

	Abstract

Top Abstract Results Discussion Methods Acknowledgements References

imaging | learning | memory

Birdsong is studied as a model of vocal learning, perception,production, and motor abnormalities of speech (1, 2). Thereare interesting parallels between song development and speechdevelopment (3) and between auditory and vocal pathways in thesongbird and human brain (4). Therefore, insights from experimentson songbirds may contribute to the understanding of auditoryand vocal function in humans. For example, minimal models ofspeech dyspraxia (5) and dysfluencies such as stuttering arebeing developed in zebra finches (6). Zebra finches are capableof learning, producing, perceiving, and discriminating complexsound patterns. Birdsong in zebra finches consists of a sequenceof distinctive sounds produced by males and is characterizedby a consistent and reproducible acoustic profile. Song is learnedby imitating the song of an adult conspecific tutor during asensitive period of development (7–10). Recently, it hasbeen shown that songbirds are able to learn recursive syntacticpatterns, presumably a simple form of grammar (11), thus extendingthe potential applicability of the birdsong model to our understandingof the biological basis of languages.

Several brain structures are required for learning, production,and perception of birdsong. It is known from electrophysiologicalstudies that song learning nuclei, such as the lateral magnocellularnucleus of the anterior nidopallium (LMAN) and X (12), playan important role in song development. In parallel with songmotor learning, auditory song selectivity gradually emergesduring development (13–17). Robust sensory responses toauditory stimuli have been recorded in the primary auditoryarea in the caudal telencephalic region (field L), the caudomedialnidopallium (NCM), the caudal mesopallium (CM), and the caudomedialventral hyperstriatum (18–21), as well as in the songnuclei HVC, LMAN, X, and nucleus interface of the nidopallium(22–28). Sensory representation of birdsong in the songnuclei and the secondary auditory areas NCM and CM is characterizedby response selectivity to song ownership, familiarity, andspecies-specific features. For instance, neurons in these structuresare more responsive to bird's own song (BOS) and tutor song(TUT) than to conspecific song (CON) and to CON compared withheterospecific song (26).

Electrophysiological recordings of multiple and single unitshave provided high temporal resolution regarding stimulus-specificsensory responses in functionally specific brain nuclei. Auditory-evokedresponses recorded on the surface of the brain have also revealedtemporal information about activation of the brain as a whole(29). However, because these methods do not provide 3D spatialresolution of relevant brain substrates, we have a poor understandingof the topography of sensory activity, both within each auditoryarea or song nucleus and more globally. Experiments examiningthe spatial patterns of song stimulation-induced up-regulationof immediate early gene products such as ZENK have been successfulin addressing this deficiency (30, 31). For example, Ribeiroet al. (32) have studied the topographic organization of songsyllables in the canary NCM. ZENK expression also shows higherresponse of neurons in NCM to CON vs. heterospecific song (33).

In this study, we examine the spatial pattern of brain activationin response to auditory stimuli by adapting the blood oxygenationlevel-dependent (BOLD) functional MRI (fMRI) method (34) inmildly sedated zebra finches. This method enables us to obtainbetter spatial resolution and localization of neural representationof birdsong than in electrophysiological recordings and to investigatethe effect of various stimuli on the same individual. The feasibilityof fMRI on songbirds has been demonstrated recently in anesthetizedEuropean starlings (35). It was shown that a sensory BOLD responseexists, is stable over time, and causes specific activationof auditory areas of the brain in response to auditory and songstimuli. In this experiment in awake, mildly sedated zebra fincheswe ask whether there are differences in the spatial distributionof stimulus-dependent activation based on species-specific stimulussaliency, ownership of song stimulus, and experience-based familiarityof song stimulus. Accordingly, we image sensory BOLD responsesto a BOS, TUT, CON, and a nonsong pure tone and determine theirdifferential spatial patterns of functional activation in thezebra finch brain.

Our results provide insights into the 3D representation of birdsongin the zebra finch brain and clearly establish the feasibilityof fMRI in awake songbirds.

	Results

Top Abstract Results Discussion Methods Acknowledgements References

Fig. 1A shows a representative maximum intensity projectionof significantly active voxels for the whole brain in threeorthogonal views. The BOLD response is seen at similar locationsin both hemispheres, with a pronounced caudal bilateral clusterin the medial slice closest to the midline, and extending intothe slice adjacent to it. This cluster was present in 63 ofthe 64 scans performed and presumably includes parts of fieldL, NCM, and CM. The BOLD response time series within that cluster(Fig. 1B) has a characteristic shape with a sharp rise at thebeginning and a negative undershoot in the off part of the on–offstimulation block. Another significant bilateral cluster isseen in a more ventral position at the location of the midbrain.In addition to these consistently observed clusters, more variableactivations were seen in other brain areas. Activation in slices1 and 8 appeared to be most variable, presumably because ofpartial volume effects with areas outside the brain and aretherefore not used in the following. Fig. 2A shows for all stimuliaveraged activation clusters from the outer parasagittal slices2 and 7 (lateral) to the inner parasagittal slices 4 and 5 (medial).Before averaging, all data were approximately geometricallynormalized to a template brain. In the following, we first describetopographical properties of the BOLD response (location andextent) and then the amplitude of the BOLD response.

View larger version (61K): [in this window] [in a new window]   Fig. 1. Location and time trace of main BOLD response. (A) Slice midlines used to prescribe the fMRI scans (Upper Left) and maximum intensity projections of color-coded correlation coefficients R > 0.2 (P < 0.001), along three orthogonal views, of the brain of a male zebra finch stimulated with a CON (Upper Right and Lower). For demonstration, the green lines connect corresponding voxels in the three views. (B) Cluster of activated voxels in the forebrain area close to the sagittal midline (Inset) of a male zebra finch stimulated with his own song. Colors code the correlation coefficients of the response with the stimulation function, also given as numbers (R > 0.16, P < 0.005). Shown are averaged BOLD response time series with a characteristic shape as expected for this block design paradigm. The average was taken over the 16 stimulation blocks.

View larger version (54K): [in this window] [in a new window]   Fig. 2. Differential topography of activation of auditory areas in response to different stimuli. (A) Averaged functional activations depending on stimulus and sagittal slice position. The slice numbers as corresponding to Fig. 1A are given at the bottom. Functional activations are shown as average correlation coefficients. The grayscale background consists of a representative anatomical volume (smoothed and averaged over both hemispheres). (B) Comparison of averaged activation for different stimuli. Positive changes of the correlation coefficient from stimulus 1 to stimulus 2 are shown in red, and negative changes are in blue, as indicated with the color bar. For example, the stimulation with BOS yields more posterior (red area, arrow) and less anterior (blue) activation than the stimulation with TONE. (C) Example of differences in the shape of the mean BOLD response between primary and secondary auditory areas for stimulation with TUT. Using a simple clustering mechanism based on the strength of activation, the activated area segments into two (noncontiguous) clusters shown in red and green, presumably corresponding to activations in field L (green) and other areas including NCM (red). The corresponding averaged time series within the two clusters, shown in Inset in the same color as the clusters and with 66% confidence intervals, show a distinctively different time course. The anatomical overlay was redrawn from ref. 42.

Stimulus-Dependent Differentiation of Sensory BOLD Response Topography in Auditory Areas. Focusing on the medial brain slices 4 and 5 in Fig. 2A, theregion of the brain that shows most pronounced, consistent,and reproducible patterns of activation, we see distinct differencesin the distribution of the voxels activated in response to differentstimuli. In the averages over all birds, the largest contiguousarea of activation is seen with TUT and the smallest with theunfamiliar CON stimulus. The two developmentally familiar songs,namely TUT and BOS, show a widening of the caudal-most extensionof the activated area, which presumably corresponds to the secondaryauditory areas CM and NCM, and their input field L subregionsL1 and L3.

Fig. 2B shows differential profiles for all combinations ofstimuli in the two medial slices, averaged over the two hemispheres.Most evident is a shift of the activation toward more caudalregions from TONE to BOS (red area, positive change vs. bluearea, negative change from TONE to BOS); TUT shows much morepronounced activation throughout the activated region when comparedwith TONE, and greater amount of activation in the central androstral field L portion when compared with BOS. In the lattercomparison, TUT and BOS show nearly equal amounts of activationin the wider posterior caudal area that corresponds to NCM.

A simple cluster analysis (Fig. 2C) of the caudomedial region,based on the general linear model coefficients and their estimationerrors of all 16 birds and all four stimuli, yielded no clusters6 times, one cluster 23 times, two clusters 34 times, and fourclusters once. The shape of the clusters was variable. Therewas no dependence between the number of clusters and the stimulusused. Of the 35 cases where two or more clusters were found,in 23 cases the cluster with the stronger activation coveredmore likely field L than NCM (five cases showed the oppositebehavior, i.e., weaker activation in field L; the remainingseven cases were inconclusive). All field L activations (greenarea in Fig. 2C) had a stronger undershoot of the BOLD responsein the off part of the stimulation than NCM (posterior red areasin Fig. 2C) and CM (anterior red area in Fig. 2C) activations.The onset of all but six BOLD response time series occurredearlier in field L than in NCM.

Discrimination Between Stimuli in the Forebrain as a Whole. A quantification of the response by measuring the time-averagedBOLD response amplitude relative to the mean signal intensityfor significantly activated voxels (P < 0.005) gave the followingvalues: the maximum time-averaged BOLD response amplitude foundin the whole brain was similar across all stimuli: 4.4% (TONE,CON, BOS) and 4.5% (TUT). The average of the strongest activatedvoxel in each bird (average over all 16 birds), however, wasvery interesting: 2.7% (TONE), 2.8% (CON), 2.6% (TUT), but 3.7%for BOS. Not surprisingly, this effect stems mostly from highauditory nuclei, so that in the area containing field L, NCM,and CM, the average of the strongest activated voxel in eachbird (average over all 16 birds) was 1.8% (TONE), 1.9% (CON),1.9% (TUT), and again very high, 2.7% to BOS. The maximum time-averagedindividual BOLD response amplitude in this area was 3.3% (TONE),2.7% (CON), 3.5% (TUT), and 4.4% (BOS). The standard deviationsof these measures ranged from 0.2% to 0.5% of the mean signalintensity.

A separation of the brain into medial and lateral parts revealsdifferential selectivity of the medial parts with respect tothe stimulus. Fig. 3A shows a comparison of BOLD response amplitudesof different stimuli for the two medial slices 4 and 5 and thelateral slices 2 and 7. A single-factor ANOVA test across thestimuli yielded P values of 0.0006 and 0.5 for the medial andlateral slices, respectively. In particular, the response toTONE stimulation is significantly smaller than to CON or BOSstimulation, and response to TUT is weaker than to BOS (Table 1;Wilcoxon signed rank test, P < 0.05). In contrast, in thetwo more lateral slices 2 and 7, the only significant discriminationfound was between TUT and CON stimulation. In these slices,CON yielded the strongest response amplitude but more variabletopography. To specify the medial response topography, slices4 and 5 were subdivided into three mutually exclusive regions:region 1 containing the caudal areas with NCM and field L, region2 containing the cerebellum and ventral parts of the brain,and region 3 containing the rostral forebrain. Only region 1showed a stimulus-dependent BOLD response (P = 0.05, 0.36, and0.30 for regions 1, 2, and 3, respectively; single-factor ANOVA).A two-factor ANOVA was significant with respect to the brainregion (P = 6 x 10^–6) and stimulus type (P = 0.02) butnot with respect to the interaction between brain region andstimulus type (P = 0.7).

View larger version (52K): [in this window] [in a new window]   Fig. 3. Larger mean BOLD response amplitude to BOS compared with TUT and right hemispheric bias in stimulus-dependent topographic differences. (A) Mean and standard error of the BOLD response amplitude over all activated voxels (P < 0.005) in medial slices 4 and 5 and lateral slices 2 and 7, for all 16 birds. The P values of a single-factor ANOVA test across the stimuli are P = 0.0006 for the medial and P = 0.5 for the lateral slices. (B) The same for the right medial slice vs. the left medial slices; P = 0.0005 and 0.2, respectively.

 
Finally, a comparison between the left and right hemispheresshowed that the average BOLD response is of comparable amplitude.However, a test of significant differences in dependence ofthe stimulus showed that only the right hemisphere discriminatesbetween different stimuli, in the aforementioned sense (Fig. 3B).The BOLD response in the right hemisphere represented by slice4 discriminates TONE and TUT from BOS (single-factor ANOVA acrossstimuli, right hemisphere P = 0.0005, left hemisphere P = 0.2).A two-factor ANOVA with stimuli and hemisphere (left/right)as independent factors, however, was only significant with respectto stimulus (P = 0.0003) but not with respect to hemisphere(P = 0.9) nor interaction between hemisphere and stimulus (P= 0.09). The BOLD responses in the left and right lateral slices,2 and 7, again do not discriminate between the different stimuli(single-factor ANOVA across stimuli, left hemisphere P = 0.7,right hemisphere P = 0.08). As far as topographic distributionbetween the left and right hemisphere is concerned, the distributionis variable, but on average there is a larger spatial patternof activation on the right (Fig. 2A).

In summary, areas in the two medial slices containing predominantlyNCM, CM, and field L show greater sensitivity to BOS/CON comparedwith TONE and to BOS compared with TUT. There is no significantdifference in mean BOLD response amplitudes between stimuliin the lateral slices. Activations in the right brain hemispherediscriminate better between stimuli than those in the left hemisphere,and there appears to be a more variable but wider representationof stimuli on the right side.

	Discussion

Top Abstract Results Discussion Methods Acknowledgements References

fMRI of the Adult Zebra Finch Reveals Strong Auditory Stimulus-Evoked Activation Patterns. We have demonstrated that it is feasible to obtain localizedsensory BOLD responses in awake, mildly sedated zebra finchesby using a 3.0-T MRI scanner. Widespread functional activityis seen across the forebrain and midbrain. In particular, wecould identify BOLD responses in forebrain loci that presumablycorrespond to the auditory areas NCM, CM, and field L. In addition,significant activations were found in lateral slices that containparts of the midbrain, HVC/HVC shelf, and robust nucleus ofthe arcopallium (RA)/RA cup. The latter activations, however,were spatially more variable and diffuse, and we were not ableat this point to make clear assignments to specific regions.It is worth mentioning that methods of immediate early geneexpression, which have a considerable higher resolution thanthe method used here, reveal activity after song playback insome of these areas (in particular, HVC shelf, RA cup, and nucleusmesencephalicus lateralis pars dorsalis in the midbrain) (30,36, 37).

The BOLD response amplitudes in the caudomedial brain regiondepended on stimulus type used. In particular, there were significantdifferences in the BOLD response amplitude for a pure tone versusCON and BOS activation, and the TUT response was different fromthe BOS response. We could also demonstrate that the auditoryresponses in the forebrain show high sensitivity to BOS [aswas found in electrophysiological recordings in anesthetizedbirds (22, 26, 28, 38, 39)] and TUT in areas corresponding tothe higher auditory area NCM [as was found in awake birds (40)],but that other stimuli activate the brain as well.

Familiar Song Stimuli Show Selective Differential Topography and Lateralization in Auditory Areas. The observation that stimuli with a high degree of developmentalexperience-based familiarity, namely TUT and BOS, show a widerextent of BOLD responsiveness in areas corresponding to theNCM and CM, and their input field L subregions L3 and L1, ishighly consistent with the functional circuitry and electrophysiologicaldata. The neuroanatomy of the songbird brain indicates thatNCM and CM are secondary auditory areas with immediate bidirectionalconnections with L3 and L1, respectively (41, 42). These areasare likely to be modulated by experience and learning. Indeed,unit recordings show that neuronal responses are more selectivelytuned to learned vocal sounds in NCM (20, 21) and CM (43–45),whereas the primary auditory subregions L2a and L2b are responsiveto sounds within the wider species-specific spectrotemporalrange (24, 46, 47). Measurement of long-term response habituationin NCM, by both electrophysiology and the study of the songstimulation-induced up-regulation of ZENK, has suggested thatthis area might encode the long-lasting sensory memory of theTUT (31, 40). Other experiments point to the fact that NCM andCM might be involved in short-term plasticity related to songdiscrimination (18, 19, 48). The greater representation of TUTand BOS revealed by fMRI in our experiments therefore mightreflect an important aspect of the sensory memory for thesedevelopmentally salient familiar stimuli.

The observation that better discrimination between stimuli (measuredas the mean amplitude of the BOLD response in significantlyactivated areas) is seen in the right hemisphere suggests apossible lateralization of the mechanisms underlying song perception.Bilateral asymmetry with either left-sided or right-sided dominance,or differential dual specialization, has been previously observedin songbirds in relation to both central and peripheral controlof song production (49–52). As far as song perceptionis concerned, increased neuronal responsiveness to behaviorallyrelevant song stimuli has been observed in the field L complexand HVC on the right side in starlings (53, 54). Right hemisphericspecialization is seen only in awake birds. These observationstherefore are in close agreement with our fMRI findings in awakebut mildly sedated zebra finches. Functional lateralizationof this type is of obvious significance from the perspectiveof birdsong as a model of speech, a strongly lateralized humanbehavior.

fMRI in Songbirds as a Reliable Research Tool. Any fMRI experiment on animals differs from natural conditions.The background noise cannot be shielded completely and may affectthe overall functioning of auditory pathways. However, by usinga block design paradigm in which only responses with stimuluson and stimulus off are compared, it is assumed that a sufficientamount of the background noise effect is subtracted out, asin auditory fMRI studies in humans that use continuous scanningparadigms. An indication that this is true to some degree isthat even pure tone stimulation at a frequency within the spectrumof the scanner noise still caused substantial BOLD activation.A caveat to our approach could be that sedation with Diazepammay enhance inhibitory activity in some parts of the brain,and we cannot exclude that additional areas may have been activatedwithout the use of Diazepam. However, this effect of Diazepam,albeit at doses higher than those that we used, has been shownto make electrophysiological neuronal responses in HVC lessvariable and more selective to BOS as compared with wakefulness(55). Further, as described in supporting information (SI) Text,SI Fig. 5, and SI Fig. 6, surface-evoked potential responsesand field potential responses in NCM to the stimuli are notsignificantly altered by Diazepam at the dose used. The restraintof the birds may cause stress and possible changes in neuromodulatorlevels, which again may affect the BOLD response. There areremaining technical challenges and room for improvements tomake fMRI in songbirds a reliable research tool. The detailsof the coupling of the BOLD response to neuronal activationin songbirds are unknown, and therefore, optimal timing anddelivery of stimuli should be further investigated. The anatomicalmapping of active areas could be further improved by the developmentof automatic registration methods suitable for the avian brain,which does not have anatomical landmarks as significant as thehuman brain. This would make the assignment of observed activationsto anatomical structures more reliable, for example, in ourcase field L and NCM, which were based to some degree on functionitself.

In conclusion, we have demonstrated that auditory fMRI of theawake zebra finch can reveal details about the 3D topographyof neural correlates of song perception. It suggests a differentialencoding underlying experience-based discrimination betweensong stimuli and a right hemispheric bias in auditory processing.Even though there is still room for improvement, the noninvasivenature of fMRI holds the promise of conducting within-subjectlongitudinal studies of the development of neural correlatesof song perception.

	Methods

Top Abstract Results Discussion Methods Acknowledgements References

View larger version (51K): [in this window] [in a new window]   Fig. 4. Experimental setup and stimulation paradigms. (A) Immobilization of mildly sedated zebra finch in a flexible Tygon plastic tube. (B) Bird in the MRI radiofrequency coil. (C) Timing of stimuli. Here the pure tone stimulus TONE is shown. It consists of 1-s on and 1-s off tones, repeated twice within a sampling interval of 4 s. The sampling interval is repeated eight times (32 s), followed by a silent block of 32-s duration. The whole on–off block of 64 s is repeated 16 times, yielding 17:04 min stimulation time per experiment. For song stimulation (CON, BOS, TUT), the song is played twice within a sampling interval of 4 s, and the gaps between the repeated song playouts are adjusted accordingly.

MRI Parameters. Images were acquired on an Excite 3.0 T scanner (GE, Waukesha,WI) with an in-house built solenoid transmit/receive coil of20-mm length and 15-mm inner diameter (Fig. 1B). BOLD-sensitivefMRI sequences of images were acquired by using a four-shot2D gradient echo EPI sequence with repetition time/echo time= 1,000/25 ms. The effective repeat time per volume was 4 s.Eight sagittal slices of 1.0-mm thickness, 4 cm field of view(FOV) (phase FOV = 0.75), flip angle 70°, and a matrix sizeof 128 x 64 (zero filled to 128 x 128) were acquired with gradientramp sampling. Slices were prescribed from right to left (Fig. 1A).The scan time per experiment was 1,024 s (256 repeats). Additionally,localizers, in-plane anatomical images, and field correctionmaps were acquired.

Paradigms. All stimuli were delivered in 16 blocks, each consisting ofa 32-s on and a 32-s off segment, totaling 1,024 s (Fig. 4C).All stimuli where normalized with respect to peak amplitudeand played out 16 times (twice per sampling interval) duringthe on segment of each block, i.e., the number of stimuli pertime was kept constant despite the different stimuli lengths.The auditory stimuli were TONE, CON, BOS, and TUT. The birdsused for our experiments were obtained from the laboratoriesof O.T. and S.A.H. The BOSs and TUTs were highly dissimilarwith the CON [similarity <20%, Sound Analysis program (55)],which was an unfamiliar song recorded from a bird in a differentcolony. Two CONs were tested before the experiment, yieldingsimilar results, and one of these songs was used in all birds.All songs were female-directed songs recorded from birds thatwere >150 days old. All birds used in this study were tutoredby a single tutor in groups of two to four. Seven differentTUTs were used, depending on the exact tutoring history. Thedurations of the TONE and CON stimuli were 1,000 and 730 ms,respectively. The mean durations of BOS and TUT were 1,233 ±406 and 1,519 ± 454 ms, respectively.

Postprocessing. BOLD-sensitive EPI images were corrected for distortions byusing field correction maps and in-house software written inIDL (Research Systems, Boulder, CO) (56, 57). The images werethen despiked and motion was corrected by using AFNI (58). Afterstatistical modeling, as described below, the statistical parametricmaps (SPMs) were registered to a brain template that consistedof the least distorted and most symmetric (with respect to thesagittal midline) EPI scan. The 2D registration was based ona locally affine but globally smooth transformation (59) estimatedfrom the EPI data and applied to the SPMs and averaged BOLDresponse time series. All averaging of activities, as describedfor the two modeling approaches below, was done after this registrationprocess. Activations in areas with an EPI intensity baseline<20% of the maximum slice intensity were discarded. It wasnot necessary to discard scans due to bulk motion which wasalways small [unlike in our first experiment using mild anesthesia(60)]. Eye components were removed. Distortion corrected imageswere mapped to anatomical drawings (ref. 42 and B. Nixdorf andH.-J. Bischof, personal communication).

Statistical Modeling. Statistical modeling was done following two approaches: correlationanalysis and general linear modeling with spatial adaptive smoothing.Correlation analysis is easier to reproduce and turned out tobe more suitable for obtaining averaged statistical parametricmaps, whereas general linear modeling with spatial adaptivesmoothing yields a better effective resolution and renderingof individual activation areas, as well as a higher statisticalaccuracy.

Correlation analysis.  As a first and easily reproducible modeling approach, data weresmoothed slicewise with a 2D Gaussian filter (half-width 1.5voxels), voxelwise detrended by subtracting a linear fit, andtemporally smoothed by convolution with a binomial filter overthree time points. Statistical significance of activation wasdefined voxelwise by correlating the signal intensity with theon–off block stimulation function. In the BOLD responsetime series (Fig. 1B), all 16 repeated blocks were averaged.For each voxel in the average plot in Fig. 2A, a sufficientcondition for a voxel to be displayed was that at least 4 ofthe 16 birds had a correlation coefficient of R > 0.16, correspondingto P < 0.005 for each bird.

General linear model.  As a statistically more advanced approach that was used forquantification and statistics, we also fitted general linearmodels to the data that take local trends and the expected hemodynamicresponse function (HDR) into account. The statistical parametricmap for this response has been smoothed by using the propagation–separation(PS) approach (61–63) to achieve noise reduction withoutblurring the shape of the activation areas. The PS method naturallyadapts to different shapes of activation areas by generatinga spatial structure corresponding to similarities and differencesbetween time series in adjacent locations. The general linearmodel was given by y_t = ₀ + ₁h(t) + ₂t + ₃t² + _t, with y_t thesignal in one voxel at time t, _t the residual, _(.) coefficientvectors, and h(t) the expected BOLD response function, definedas the convolution of the block stimulation function with anidealized HDR. The HDR was modeled as:

with a₁ = 6, a₂ = 12, b₁ = b₂ = 0.9, and d_i= a_ib_i, i = 1, 2, c = 0.35, and t the time in seconds (64).Subsequently, PS was applied as described (63). From the smoothedmap of coefficients ₁ and its estimated standard deviation,voxelwise t scores were computed. Discrimination of the responseto the four stimuli was performed by measuring the mean BOLDresponse amplitude in percent of the mean signal amplitude withinsignificant voxels for t > 2.60 (P < 0.005) for the regionsas described in Results and subsequent application of a Wilcoxonsigned rank test for zero median of the differences (Table 1).To account for the multiple testing problem, these results wereverified by a single-factor ANOVA test with the stimulus asparameter, as described in Results. All ANOVA tests were repeated-measuretests with the individual 16 birds as repeated measure.

Cluster analysis.  The PS approach contains a simple clustering mechanism by segmentingregions of activations by the fit coefficient ₁ and its standarddeviation. Clusters were computed only within activations witha global, i.e., multiple test corrected, threshold of P = 0.05,by including voxels with a voxelwise threshold of P = 0.005.Clusters within the activated region presumably containing fieldL and NCM were counted. Clusters were evaluated with respectto their exact position relative to each other; clusters weregrouped by their position as described as rostral-dorsal (presumablyfield L) vs. caudal-ventral (presumably NCM), and these groupswere compared with amplitudes and onset latencies of the correspondingaveraged time series of activation.

All computations were performed with in-house software writtenin R (The R Project for Statistical Computing, www.r-project.org),MATLAB (Mathworks, Natick, MA), and Excel 2003 (Microsoft, Redmond,WA) on personal computers and Unix workstations.

	Acknowledgements

Top Abstract Results Discussion Methods Acknowledgements References

	Footnotes

Abbreviations: LMAN, lateral magnocellular nucleus of the anterior nidopallium; NCM, caudomedial nidopallium; CM, caudal mesopallium; CON, conspecific song; BOS, bird's own song; TUT, tutor song; BOLD, blood oxygenation level-dependent; fMRI, functional MRI; TONE, pure tone of 2-kHz frequency; EPI, echo-planar imaging; PS, propagation–separation.

To whom correspondence should be addressed. E-mail: hev2006{at}med.cornell.edu

Author contributions: D.B. and S.A.H. contributed equally tothis work; H.U.V., D.S.-C., D.B., and S.A.H. designed research;H.U.V., K.K.M., and S.A.H. performed research; H.U.V., K.T.,and J.P. contributed new reagents/analytic tools; H.U.V., K.T.,and J.P. analyzed data; and H.U.V., K.T., J.P., O.T., K.K.M.,D.S.-C., D.B., and S.A.H. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

This article contains supporting information online at www.pnas.org/cgi/content/full/0611515104/DC1.

© 2007 by The National Academy of Sciences of the USA

	References

Top Abstract Results Discussion Methods Acknowledgements References

• Marler, P. (1970) Am Sci58, 669–673.[Web of Science][Medline]
• Konishi, M. (1965) Z Tierpsychol22, 770–783.[Medline]
• Doupe, AJ & Kuhl, PK. (1999) Annu Rev Neurosci22, 567–631.[CrossRef][Web of Science][Medline]
• Jarvis, ED. (2004) Ann NY Acad Sci1016, 749–777.[CrossRef][Web of Science][Medline]
• Scharff, C & White, SA. (2004) Ann NY Acad Sci1016, 325–347.[CrossRef][Web of Science][Medline]
• Helekar, SA, Espino, GG, Botas, A & Rosenfield, DB. (2003) Behav Neurosci117, 961–969.[CrossRef][Web of Science][Medline]
• Immelmann, K. (1969) in Bird Vocalizations ed. Hinde, RA. (Cambridge Univ Press, Cambridge, UK,) pp. 61–77.
• Konishi, M & Nottebohm, F. Hinde, RA. (1969) Bird Vocalizations (Cambridge Univ Press, Cambridge, UK,) pp. 29–48.
• Price, P. (1979) J Comp Physiol Psych93, 260–277.[CrossRef][Web of Science]
• Clayton, NS. (1987) Anim Behav35, 714–721.[CrossRef][Web of Science]
• Gentner, TQ, Fenn, KM, Margoliash, D & Nusbaum, HC. (2006) Nature440, 1204–1207.[CrossRef][Medline]
• Brainard, MS & Doupe, AJ. (2002) Nature417, 351–358.[CrossRef][Medline]
• Doupe, A. (1997) J Neurosci17, 1147–1167.[Abstract/Free Full Text]
• Doupe, A & Solis, M. (1997) J Neurobiol33, 694–709.[CrossRef][Web of Science][Medline]
• Solis, M & Doupe, AJ. (1997) J Neurosci17, 6447–6462.[Abstract/Free Full Text]
• Volman, S. (1993) J Neurosci13, 4737–4747.[Abstract]
• Nick, TA & Konishi, MJ. (2005) J Neurobiol62, 469–481.[CrossRef][Web of Science][Medline]
• Chew, SJ, Mello, C, Nottebohm, F, Jarvis, E & Vicario, DS. (1995) Proc Natl Acad Sci USA92, 3406–3410.[Abstract/Free Full Text]
• Gentner, TQ & Margoliash, D. (2003) Nature424, 669–674.[CrossRef][Medline]
• Stripling, R, Volman, SF & Clayton, DF. (1997) J Neurosci17, 3883–3893.[Abstract/Free Full Text]
• Terleph, TA, Mello, CV & Vicario, DS. (2006) J Neurobiol66, 281–292.[CrossRef][Web of Science][Medline]
• Doupe, A & Konishi, M. (1991) Proc Natl Acad Sci USA88, 11339–11343.[Abstract/Free Full Text]
• Bonke, B, Bonke, D & Scheich, H. (1979) Cell Tissue Res200, 101–121.[Web of Science][Medline]
• Müller, C & Leppelsack, H. (1985) Exp Brain Res59, 587–599.[Web of Science][Medline]
• Williams, H & Nottebohm, F. (1985) Science229, 279–282.[Abstract/Free Full Text]
• Margoliash, D. (1986) J Neurosci6, 1643–1661.[Abstract]
• Margoliash, D & Fortune, E. (1992) J Neurosci12, 4309–4326.[Abstract]
• Vicario, D & Yohay, K. (1993) J Neurobiol24, 488–505.[CrossRef][Web of Science][Medline]
• Espino, GG, Lewis, C, Rosenfield, DB & Helekar, SA. (2003) Neuroscience122, 521–529.[CrossRef][Web of Science][Medline]
• Mello, CV & Clayton, DF. (1994) J Neurosci14, 6652–6666.[Abstract]
• Bolhuis, JJ, Zijlstra, GO, den Boer-Visser, AM & Van der Zee, EA. (2000) Proc Natl Acad Sci USA97, 2282–2285.[Abstract/Free Full Text]
• Ribeiro, S, Cecchi, GA, Magnasco, MO & Mello, CV. (1998) Neuron21, 359–371.[CrossRef][Web of Science][Medline]
• Mello, CV, Vicario, DS & Clayton, DF. (1992) Proc Natl Acad Sci USA89, 6818–6822.[Abstract/Free Full Text]
• Ogawa, S, Lee, T, Kay, A & Tank, D. (1990) Proc Natl Acad Sci USA87, 9868–9872.[Abstract/Free Full Text]
• Van Meir, V, Boumans, T, De Groof, G, Van Audekerke, J, Smolders, A, Scheunders, P, Sijbers, J, Verhoye, M, Balthazart, J & Van der Linden, A. (2005) NeuroImage25, 1242–1255.[CrossRef][Web of Science][Medline]
• Mello, CV & Ribeiro, S. (1998) J Comp Neurol393, 426–438.[CrossRef][Web of Science][Medline]
• Velho, TA, Pinaud, R, Rodrigues, PV & Mello, CV. (2005) Eur J Neurosci22, 1667–1678.[CrossRef][Web of Science][Medline]
• Margoliash, D. (1983) J Neurosci3, 1039–1057.[Abstract]
• Volman, SF. (1996) J Comp Physiol A178, 849–862.[Medline]
• Phan, ML, Pytte, CL & Vicario, DS. (2006) Proc Natl Acad Sci USA103, 1088–1093.[Abstract/Free Full Text]
• Fortune, ES & Margoliash, D. (1992) J Comp Neurol325, 388–404.[CrossRef][Web of Science][Medline]
• Vates, GE, Broome, BM, Mello, CV & Nottebohm, F. (1996) J Comp Neurol366, 613–642.[CrossRef][Web of Science][Medline]
• Sen, K, Theunissen, FE & Doupe, AJ. (2001) J Neurophysiol86, 1445–1458.[Abstract/Free Full Text]
• Hsu, A, Woolley, SM, Fremouw, TE & Theunissen, FE. (2004) J Neurosci24, 9201–9211.[Abstract/Free Full Text]
• Gill, P, Zhang, J, Woolley, SM, Fremouw, T & Theunissen, FE. (2006) J Comput Neurosci21, 5–20.[CrossRef][Web of Science][Medline]
• Grace, JA, Amin, N, Singh, NC & Theunissen, FE. (2003) J Neurophysiol89, 472–487.[Abstract/Free Full Text]
• Cousillas, H, Leppelsack, HJ, Leppelsack, E, Richard, JP, Mathelier, M & Hausberger, M. (2005) Hear Res207, 10–21.[CrossRef][Web of Science][Medline]
• Mello, C, Nottebohm, F & Clayton, D. (1995) J Neurosci15, 6919–6925.[Abstract/Free Full Text]
• Nottebohm, F. (1971) J Exp Zool177, 229–261.[CrossRef][Web of Science][Medline]
• Hartley, RS & Suthers, RA. (1990) J Neurobiol21, 1236–1248.[CrossRef][Web of Science][Medline]
• Floody, OR & Arnold, AP. (1997) Horm Behav31, 25–34.[CrossRef][Medline]
• Halle, F, Gahr, M & Kreutzer, M. (2003) J Neurobiol56, 303–314.[CrossRef][Web of Science][Medline]
• George, I, Cousillas, H, Richard, JP & Hausberger, M. (2005) J Comp Neurol488, 48–60.[CrossRef][Web of Science][Medline]
• George, I, Vernier, B, Richard, JP, Hausberger, M & Cousillas, H. (2004) Behav Neurosci118, 597–610.[CrossRef][Web of Science][Medline]
• Cardin, JA & Schmidt, MF. (2003) J Neurophysiol90, 2884–2899.[Abstract/Free Full Text]
• Chang, H & Fitzpatrick, JM. (1992) IEEE Trans Med Im11, 319–329.[CrossRef]
• Voss, HU, Watts, R, Ulug, AM & Ballon, D. (2006) MRI24, 231–239.
• Cox, RW. (1996) Computers Biomed Res29, 162–173.[CrossRef]
• Periaswamy, S, Weaver, JB, Healy, DM Jr, Rockmore, D, Kostelec, PJ & Farid, H. (2000) Proc SPIE Int Soc Opt Engineer4119, 1066–1075.
• Voss, HU, Maul, K, Ballon, D, Tchernichovski, OT & Helekar, SA. (2006) Proc Int Soc Magn Reson Med14, 2132.
• Polzehl, J & Spokoiny, V. (2001) J R Statist Soc Ser C50, 485–501.[CrossRef]
• Polzehl, J & Spokoiny, V. (2006) Probab Theory Relat Fields135, 335–362.[CrossRef]
• Tabelow, K, Polzehl, J, Voss, HU & Spokoiny, V. (2006) NeuroImage33, 55–62.[CrossRef][Web of Science][Medline]
• Glover, GH. (1999) NeuroImage9, 416–429.[CrossRef][Web of Science][Medline]