Memory-specific correlated neuronal activity in higher-order auditory regions of a parrot

Male budgerigars (Melopsittacus undulatus) are open-ended learners that can learn to produce new vocalisations as adults. We investigated neuronal activation in male budgerigars using the expression of the protein products of the immediate early genes zenk and c-fos in response to exposure to conspecific contact calls (CCs: that of the mate or an unfamiliar female) in three subregions (CMM, dNCM and vNCM) of the caudomedial pallium, a higher order auditory region. Significant positive correlations of Zenk expression were found between these subregions after exposure to mate CCs. In contrast, exposure to CCs of unfamiliar females produced no such correlations. These results suggest the presence of a CC-specific association among the subregions involved in auditory memory. The caudomedial pallium of the male budgerigar may have functional subdivisions that cooperate in the neuronal representation of auditory memory.

brains [26][27][28][29][30][31] . A study in zebra finches (Taeniopygia guttata), the most widely studied songbird species, reported differential expression of Zenk and Fos following conspecific song stimulation in higher-order auditory regions 32 . In male zebra finches reared with conspecific birds in communal aviaries, the density of Zenk-positive neurons, but not Fos-positive neurons, was increased in the CMM/NCM of males following exposure to zebra finch songs at post-hatching day 30 (phd 30) 32 . In contrast, in female zebra finches, who do not sing, the density of Fos-positive neurons, but not Zenk-positive neurons, was increased in the CMM/NCM 32 . In line with the IEG study, the study using pharmacological inhibitor 21 showed the formation of an auditory memory of tutor song requires the extracellular signal-regulated kinase (ERK) signaling pathway in a brain area including the CMM/ NCM of male zebra finches. The induction of Zenk is regulated by the ERK in the zebra finch auditory forebrain 33 .
Parrots learning their contact calls (CCs) may be a good animal model for human speech acquisition 34,35 . Both sexes of the budgerigar produce individually distinct CCs (Fig. 1). At any given point in time, a bird may have a repertoire of one to several different types of CCs. CC production and recognition learning of both sexes of the budgerigar are observed in the laboratory 36,37 . During pair bond formation, males learn to produce the CCs of females and thus create a new CC type [38][39][40] , and also learn to recognise the CCs of their mate 20 . CC production learning reliably occurs in males paired with females, compared to males in a small group of the same sex 38 . In parrots, few studies have investigated IEG expression in response to CCs in the CMM or NCM. However, IEG responses to CC stimuli have been observed in a region of the caudomedial pallium of the budgerigar, which has been termed 'Field L' by Brauth et al. 6,18,41,42 and 'NCM' in other studies 4,19,20 . Only one study in budgerigars has reported Zenk responses to CC stimulation in the CMM of parrots 20 . Although the caudomedial pallium in songbirds is thought to at least partly contain the neuronal substrate of memory of the tutor song, little is known about neuronal activation related to memory for mate CC in the caudomedial pallium of parrots. We paired male budgerigars with a female for 5 weeks, during which males copied the CCs of their mates. Here, we examined whether exposing male budgerigars to previously memorized CCs results in increased IEG expression, especially differential expression of Zenk and Fos, in the caudomedial pallium. Moreover, correlations of Zenk expression between regions of the caudomedial pallium have also been observed in previous 43 and latest 44 female songbird reports. In the Zenk study on the neural processing of anuran acoustic communication signals, the differential neural representation of conspecific and heterospecific signals involves both changes in mean activity levels across multiple subnuclei, and in the functional correlations among acoustically active areas, at the level of the midbrain and forebrain 45 . Since the functional network organisation of the brains of vertebrates is thought to be closely  . We randomly selected five exemplar sonograms of each CC type and used them for rating the CC Similarity Score (Score). In each of 10 pairwise comparisons, the CC Similarity Score was rated as 1 (no similarity), 2 (modest similarity), or 3 (good similarity). Only type C was rated as good similarity and had a cross-correlation value (Correlation) greater than 0.75. In cross-correlations between two CCs from different CC types, values greater than 0.75 are highly unlikely to occur by chance 38 . Type C was not present in the CC repertoire of male pre-pairing. www.nature.com/scientificreports/ related to neural processing of signal coding, we further analysed for correlations among the post-exposure IEG expressions of the CMM and two NCM subregions of the caudomedial pallium.

Results
Contact call (CC) learning. In a representative pair, a male had five types of CCs after pairing, and one type of them showed a high 'CC Similarity Score' by visual inspection (score 3) and a 'Maximum Cross-correlation Value' (0.76; see "Methods") with the mate CC prior to pairing (Fig. 1 There was no significant interaction between Stimulus and Brain Region (F 6,36 = 0.662, P = 0.680). The data were then analysed separately for each of subregion (CMM, dNCM, vNCM, or HP) with one-way ANOVA with Stimulus again the between-subjects factor. We found a significant effect of Stimulus in Zenk expression levels in the CMM (F 2,9 = 7.788, P = 0.014) and dNCM (F 2,9 = 9.197, P = 0.007). Posthoc Bonferroni testing revealed a significant difference between the MATE and SILENCE groups in the CMM (P < 0.05, Fig. 3a). In the dNCM, significant differences were observed between the MATE and SILENCE groups (P < 0.01), and between the UNFAMILIAR and SILENCE groups (P < 0.05, Fig. 3a). One-way ANOVA revealed a significant effect of Stimulus in the vNCM (F 2,9 = 4.263, P = 0.049), but Bonferroni testing did not reveal any significant difference between groups. One-way ANOVA revealed no significant effect of Stimulus in the HP (F 2,9 = 3.956, P = 0.058). There were no significant differences between MATE and UNFAMILIAR groups in all brain regions. There was no significant correlation between the CC Similarity Score and number of Zenk-IR cells for birds of the MATE group (n = 4), UNFAMILAR group (n = 4), or SILENCE group (n = 4) in the CMM, dNCM, vNCM, or HP (Spearman's rank correlation test).

Fos expression.
As two-way ANOVA (Stimulus and Brain Region are factors) indicated significant effects of both Brain Region (F 2,27 = 5.721, P < 0.01) and Stimulus (F 2,27 = 5.312, P < 0.05), and there was no significant interaction of Stimulus and Brain Region (F 4,27 = 1.644, P = 0.192), the data for each of the three brain subregions (CMM, dNCM, and vNCM) were further analysed. One-way ANOVA indicated no significant effect of Stimulus on Fos expression levels in the CMM (F 2,9 = 0.501, P = 0.622) or dNCM (F 2,9 = 1.898, P = 0.205). A significant effect of Stimulus on Fos expression levels in the vNCM (F 2,9 = 10.058, P = 0.005) was detected. Bonferroni testing revealed significant differences in Fos expression levels in the vNCM between the MATE and SILENCE groups (P < 0.05), and between the UNFAMILIAR and SILENCE groups (P < 0.05, Fig. 3b). There were no significant differences between MATE and UNFAMILIAR groups in all brain regions. We observed that the basal expression (i.e., the expression in the SILENCE group) of Fos was very low in the vNCM (Fig. 3b). There was no significant correlation between the CC Similarity Score and number of Fos-IR cells for birds of the MATE group, UNFAMILAR group, or SILENCE group in the CMM, dNCM, or vNCM (Spearman's rank correlation test).
Correlations among brain regions in Zenk activation.  Figure 4c shows the differential pattern of Zenk expression across subregions in each of the three Stimulus groups. In order to examine the relationship between the Zenk expressions in each pair of subregions (i.e., CMM, dNCM, and vNCM), we also performed Bayesian analyses with a simple regression model, using the data for the correlation analyses mentioned above (Fig. 4). For the all three pairs of subregions in the MATE group, the 95% confidence interval (CI) for the regression coefficients did not include zero and the coefficients were positive values. These results indicate that for an increase in the Zenk expression of one of the subregions, the other subregion also increased in the expression for each pair of subregions in the MATE group. However, for the all three pairs in the UNFAMILIAR group and those in the SILENCE group, the CIs for the regression coefficients included zero. These results indicate that there was no linear relationship between the Zenk expressions in the UNFAMILIAR or the SILENCE group.
Correlations among brain regions in Fos activation. Figure 5 gives the Fos data corresponding to  Figure 5c shows the differential pattern of Fos expression across subregions in each of the three Stimulus groups. In contrast to Zenk expression, pairwise correlations between Fos expressions were significant only in CMM versus vNCM in the UNFAMIL-IAR group. No ordered relationship, which forms a nested triangle, was observed in any of the three groups. In order to examine the relationship between the Fos expressions in each pair of subregions (i.e., CMM, dNCM, and vNCM), we also performed Bayesian analyses with a simple regression model, using the data for the correlation analyses mentioned above (Fig. 5). For the all three pairs in each of the three conditions (the MATE, the UNFAMILIAR and the SILENCE groups), the CIs for the regression coefficients included zero. These results indicate that there was no linear relationship between the Fos expressions for the MATE, the UNFAMILIAR or the SILENCE group.

Discussion
In the present study, male budgerigars mimicked the acoustic structure of the CCs of their female mates within 5 weeks after being paired (Fig. 1). These behavioural results are consistent with previous findings by Hile et al. 38,40 and thus confirm that male budgerigars learn the CCs of their mates. We investigated neuronal activation based on Zenk and Fos expression in response to female CCs in the caudomedial pallium of male budgerigars. In the CMM/dNCM, CC stimulation led to significantly increased Zenk expression when compared to silence. A trend to increase was observed in the vNCM, while no increase was detected in the hippocampus (Fig. 3a). In contrast, increased Fos expression in response to CC was observed in the vNCM but not in the CMM or dNCM (Fig. 3b). The differential pattern of IEG expression suggests a functional difference between the CMM/dNCM pair and the vNCM. For both IEGs, we found no significant differences in mean expression levels between the MATE group (mate CC stimuli) and the UNFAMILIAR group (unfamiliar CC stimuli) in any brain region (Fig. 3a, b). However, only in the MATE group did Zenk (but not Fos) expression reveal significant positive correlations in neuronal activation among subregions (Fig. 4). In the UNFAMILIAR group, in contrast, we observed no significant correlations of neuronal activation among the subregions. In addition, only the mate CC stimuli induced an across-subjects pattern of Zenk expression that was consistent across subregions (Fig. 4c). These results suggest that the correlated neuronal activity among these subregions is associated with CC memory, because the context difference between the two stimuli consisted of being related or not related to this memory.
Interestingly, our findings can be interpreted as the memory of the CC not being represented by the amount of subregion activity alone, as estimated by IEG expression, but by the correlations in activity among these subregions. A design focusing only on IEG expression within the subregions would not have detected these interrelationships. Whether a correlation is observed could depend on the presence of neuronal activation that www.nature.com/scientificreports/ is synchronous across subregions 46 . By observing significant correlations of Zenk expression between subregions (Fig. 4), it can be inferred that the functional connectivities 45 between these subregions lead to synchronous activity enhancements in response to CC stimuli 47 . CC memory formation may require changes in the functional connectivities or circuits among the subregions which are constituted by Zenk-positive neurons 12,48 . This is consistent with Hebb's law and cell assemblies 49 , which posits that memory formation is due to plastic changes in synaptic connections in memory-related neural circuits 50 . We do not know any direct projections among the subregions in the caudomedial pallium of the budgerigar. Further research is necessary to investigate anatomical pathways within the caudomedial pallium, although audio-vocal pathways have been found in this species (discussed below). A study of gene expression profiles in avian brains indicated that the mesopallium cell populations containing the CMM and nidopallium cell populations containing the NCM constitute functional columns with the same physiological functions, similar to the layered structure of the mammalian cerebral cortex 51 . The CMM and NCM are structurally separated by the lamina mesopallium ventrale (LMV) in the budgerigar, as in other avian species (see Fig. 2a, b). The dNCM of budgerigars examined in this study is the region located between the LMV and The second-ranked bird The third-ranked bird The fourth-ranked bird   www.nature.com/scientificreports/ Field L2. There are differences in the orientation of Field L2 across avian species, with Field L2 having a more vertical orientation and the CMM shifted more rostrally in songbirds compared to parrots 52 . In songbirds, there is limited NCM space between the LMV and Field L2. The vNCM of the budgerigar are anatomically located on the ventral side of Field L complex, while the CMM/dNCM are located on the dorsal side of Field L complex. The vNCM, but not the CMM/dNCM, exhibited Fos expression in response to CC stimulation. Further, the budgerigar vNCM is characterised by lower basal expression of Fos than that of Zenk 42 (Fig. 3b). Based on these observations, the vNCM may be structurally or functionally different from the CMM/dNCM. The CMM/dNCM and the vNCM may cooperatively constitute a posterior auditory column 51 of the budgerigar. A previous study suggested a role of the CMM in CC perception in the budgerigar 20 , and the involvement of the CMM/NCM in song perception in budgerigars has also been reported 19 . A large body of evidence, much of it based upon gene expression studies, suggests that the caudomedial pallium of songbirds is involved in memory as well as perception of song 10 . In songbird studies, a representation of the auditory memory of the tutor song has been distinguished from that of the motor memory of the bird's own song (BOS) 8 . In adult zebra finches, IEG expression in the NCM was positively correlated with the strength of song learning in the males that were exposed to the tutor song, not in those to the BOS or to novel conspecific song 53 . By contrast to the NCM, IEG expression in the HVC (used as a proper name), which is a songbird nucleus involved in song production and sensorimotor learning, was positively correlated with the strength of song learning in the males exposed to the BOS, not in those to the tutor song or to novel song 10 . In the present study, the correlated neuronal activity among subregions of the caudomedial pallium was suggested to be memory-related in male budgerigars. Whether the correlated IEG expression is related to the auditory memory or to the motor memory could be tested by exposing male subjects to the bird's own call. Otherwise, another test could be considered. Male budgerigars learn to recognise their mate CC during the pairing period of 1 month 20 . In the present study, we did not conduct the preference test 37 for mate CC after the pairing period. If the correlated IEG expression is related to the auditory memory, the IEG expression is correlated with the strength of CC learning, measured as a preference score for the mate CC.
The design of this study has a limitation in that only one type of CC stimulus can be used for a given subject. However, using a cellular compartment analysis of temporal activity based on data from fluorescence in situ hybridization (catFISH), such as Arc catFISH, it may be possible to provide a subject with two CC stimuli The second-ranked bird The third-ranked bird The fourth-ranked bird   Fig. 4 but pertaining to the Fos-immunopositive cell counts. Unlike the results revealed by Zenk expression (Fig. 4), there was no linear relationship between subregions for the MATE, the UNFAMILIAR or the SILENCE group with Bayesian analysis.  54 . In the future, it will be necessary to use optogenetic methods 55 to deepen our understanding of the mechanism of CC memory formation, since these methods can identify memory circuits by detecting IEG expression. Activation of NMDA receptors contributes to the activation of ERKs and Zif268 (Zenk), which are essential components of a signaling cascade required for the LTP and long-term memory formation 23 . A recent study in rodents indicated a strong association between NMDA receptor activation and Zif268 (but not Fos) expression in the hippocampus during spatial learning 25 . Several studies in songbirds have examined the role of signaling cascade in song production, sensorimotor learning, and adult song plasticity. The NMDA receptors and ERK in the caudomedial pallium of the zebra finch contributes to the induction of Zenk and are essential for song learning in young birds 21,33 . The NMDA receptor subunits NR2A and NR2B show higher and lower expression, respectively, in most nuclei of the song system relative to surrounding brain subdivisions in which nuclei of the song system are located 56 . Basham et al. 57 inactivated the lateral magnocellular nucleus of the anterior nidopallium, a song system nucleus, via injections of an NMDA receptor antagonist on days that birds were exposed to tutor song. Song learning was significantly impaired in birds that received the experimental treatment, although it is not clear whether NMDA receptor antagonist infusions affected sensorimotor integration or the formation of auditory memory 10 . The three avian groups of vocal learners have overlapping but distinct cerebral nuclei for vocal production and learning, commonly called the 'song system' . In the budgerigar, the NMDA receptor subunits NR2A and NR2B show an expression pattern similar to that of songbirds in most nuclei of the song system 58 . These receptor subunits do not seem to be differentially expressed in the caudomedial pallium compared to surrounding brain subdivisions 56 . Further confirmation of the involvement of NMDA receptors 58 , ERK and Zenk in CC learning in parrots will be needed. In visual imprinting in chicks, NMDA receptors in higher visual regions of the telencephalon are also implicated in learning. The neural circuits responsible for visual imprinting in chicks reside in the pallial layers of the telencephalon 59,60 . A subsequent study reported that NR2B-containing NMDA receptors (NR2B/NR1) in this region were essential for imprinting 61 .
In this study, we have shown that the caudomedial pallium (i.e., the posterior auditory column 51 ) of the budgerigar may contain the neuronal substrate of CC memory, the basis of CC learning. However, the details of how auditory information processed through the posterior auditory column is used for CC learning remain unknown. In male songbirds that are learning vocalisations, there is continual interaction between the song system and caudomedial pallium, similar to the interaction between Broca's and Wernicke's areas in human infants acquiring speech and language. Although the budgerigar learns both songs and CC, their song system has been termed a vocal control system because research exploring the neural basis of vocal learning has mainly been performed with CC learning. Unlike songbirds, the vocal control system of budgerigars employs two structural connections, one from the caudomedial pallium and another from the anterior auditory pathway (see Fig. 2b) 4,18,[62][63][64][65] . The basorostralis (B, auditory B) located in the rostral telencephalon receives input from the intermediate nucleus of the lateral lemniscus and constitutes the anterior auditory pathway that includes the NFl (the lateral frontal neostriatum, former name) as a projection field of the auditory B 4, 7,18,62,66 (Fig. 2b). Information from the NFl is conveyed to the adjacent HVl (the lateral ventral hyperstriatum, former name) and to the central nucleus of the lateral nidopallium 67 (NLC), of the vocal control system 7,18 . The NLC receives auditory input from the caudal pallium via the NFl, which also receives axonal input from the auditory area in the caudal nidopallium. Plummer and Striedter 7 suggested that auditory information from the anterior auditory pathway plays an important role in CC learning because the destruction of HVl/NFl interferes with CC learning of budgerigars but does not disturb the production of learned CCs. The HVl/NFl is a large area which has been characterised using electrophysiological and tracing methods; nevertheless, its boundaries are less clearly delineated than those of the vocal control nucleus, NLC 7,18,62 . IEG studies have indicated that nuclei showing singing-driven Zenk expression are embedded within the HVl/NFl, and these nuclei are termed the lateral nucleus of the anterior mesopallium and lateral nucleus of the anterior nidopallium (LAN, see Fig. 2b) 4,68 . The LAN corresponds to the songbird's nuclear interfacialis of the nidopallium (NIf) 69,70 . The songbird's NIf is an important nucleus that mediates auditory and vocal representations 70,71 and is embedded in the caudomedial pallium, whereas the LAN of the budgerigar is embedded in the anterior auditory system 68,72 . Recently, Chakraborty et al. 72,73 reported that the vocal control system of parrots, unlike the song system of songbirds, has a duplicated anterior vocal control system (shell song system) which was thought to have evolved from the adjacent motor learning circuit 68 in parrots 72,73 . A recent review suggests that the shell song system which does not exist in songbirds characterises an adult CC learning in parrots as open-ended learners 74 . Taken together with our results, the anatomy of the two auditory input systems from the caudomedial pallium and anterior auditory pathway to the vocal system, as well as that of the shell song system 72,73 , suggests a mechanism for organising a dynamic functional connectivity during CC learning; these structures may function in an integrated manner as the source of the highly developed CC learning ability found in budgerigars.
In summary, we have found evidence suggesting that the neuronal representation of auditory (CC) memory in male budgerigars is a functional connectivity between the subregions of the caudomedial pallium. The most important result is that exposure to the mate, but not unfamiliar CCs, induced correlated Zenk responses among subreigons in the caudomedial pallium of male budgerigars. The correlated activities among these subregions as revealed by Zenk expression and induced by stimuli differing in relatedness to CC memory suggest organisation of functional connectivity among the subregions. The caudomedial pallium of the male budgerigar may therefore be functionally subdivided and play an important role in the neuronal representation of CC memory as cooperativity among subdivisions.

Methods
Subjects. All experimental procedures were performed in accordance with Japanese law and approved by the Animal Experiments Committee of Japan Women's University (Permit Number: II07-13). The 24 budgerigars (12 males and 12 females) in the present experiment had been used in a previous study 37 that only assessed female behaviours. All birds were obtained from a local supplier and kept in a controlled environment suitable for breeding (23 ± 3 °C and a light/dark cycle of 14:10 h light:dark) throughout the study, as described previously 37 . Hile et al. 38 developed a behavioural paradigm to elicit vocal imitation reliably and fast (within an average of 2.1 weeks) in adult male budgerigars. We used this paradigm in the present experiment. Males and females were acquired from different breeders and were housed in separate rooms, so that females were unfamiliar with males prior to the study. Birds were initially singly housed, and their vocalisations were recorded. After recording of all birds was completed, pairing was initiated. As described by Hile et al. 38 , pairs were prevented from seeing other pairs in the same room by means of wooden barriers; however, they were in vocal contact with other pairs. Hile et al. 38 reported the shared calls emerged within 3 days to 5 weeks after pairing. After the pairing period of 5 weeks, each bird was isolated from the mate and kept in an individual cage, and their vocalisations were recorded again. To assess the development of pair bonds, we performed behavioural observations on days 2 and 12 after the pairs were placed into breeding cages. As reported in the previous study 37 , (1) the birds showed significantly more allopreening (a behaviour associated with pair bonding in the budgerigar) on day 12 than on day 2, and (2) six of 12 females laid eggs by day 24 after the start of pairing.
Recording and analysis of CCs. At any given point in time, a bird has a repertoire of one to several different patterns or types of CC 36,38 . We recorded and analysed the CC repertoires of all birds. The methods used to record and analyse CCs were similar to those described in previous studies 20,37,38 . We recorded at least 80 CCs per bird at each time point. After classifying CCs into types, we paired birds dissociatively with respect to CCs, that is, pair members did not have similar CCs at this time (see below). The CCs at 5 weeks after pairing were compared with those of his assigned female (see below). Call classification. According to our previous study 37 , the vocalisations were displayed as sonograms (fast Fourier transformation, FFT, size: 256, temporal resolution 3 ms) using the sound analysis software Avisoft SASLab Pro (Avisoft Bioacoustics, Berlin, Germany). Farabaugh et al. 36 classified CCs on the basis of their frequency modulation patterns on sonograms. This procedure proved reliable for classifying CCs within individuals; average interobserver agreement for four independent observers was 93% in our pilot analysis conducted using 880 CCs (80 CCs recorded from each of 11 males). CCs of each bird were classified into types by an observer (one of four observers in our pilot analysis mentioned above) naïve to the identity of the birds. Of all CC types for each bird, one type occurs most frequently and is therefore termed 'the dominant CC' . We determined this CC for each bird.
Call similarity. Prior to pairing, all the CC types of each male were compared with those of his mate by visual inspection of sonograms. A blinded observer rated the degree of similarity as 1 (no similarity), 2 (modest similarity), or 3 (good similarity). All comparisons were rated as 1. We determined whether a male had imitated one of his paired female's CC types. Each male had several CC types (range: 5-31), while the number of CC types of each female ranged from one to five. CC types of a male at 5 weeks after pairing were compared to those of his mate prior to pairing by means of visual inspection of sonograms (Fig. 1). We randomly selected 5 exemplars of each CC type from each female prior to pairing. In the same way, 5 exemplars were randomly selected for each CC type from each male at 5 weeks after pairing. If there were fewer than 5 exemplars of a given type, all of the exemplars of that type were used. A blinded observer rated the degree of similarity as above. The highest score of all pairwise comparisons for 5 weeks after pairing was determined and defined as 'the CC Similarity Score' for each bird pair.
Spectrogram cross-correlation analysis. We also conducted spectrogram cross correlation, using Avisoft Correlator version 2.2 (Avisoft Bioacoustics), and verified that 'CC Similarity Score' was a reliable measure. The correlation coefficient is a value ranging from − 1 to + 1. A value of 1 means that the two sonograms are identical. A value of 0 means that there is no similarity between the sonograms. The mean value of the correlation within the same type classified by visual inspection was 0.75 ± 0.01. This is significantly higher than that among different types, which was 0.54 ± 0.02 (P < 0.01, Mann-Whitney U-test). CC types of a female prior to pairing were compared to those of her mate at 5 weeks after pairing, using the exemplar sonograms selected for visual inspection mentioned above. For each male, we determined the three highest correlation values in all of the comparisons between sonograms of male CCs at 5 weeks after pairing and those of his mates' prior to pairing. We defined this averaged value as 'the Maximum Cross-correlation Value' . The two methods (visual inspection and cross correlation by software) produced similar results, with higher scores by visual inspection having larger cross correlation values (mean values ± SEM: 0.75 ± 0.036 in the group with similarity score 3; 0.60 ± 0.030 in the group with similarity score 2; 0.48 ± 0.006 in the group with similarity score 1). There was a significant difference in cross correlation values among three groups with different scores by visual inspection (analysis of variance [ANOVA], F 2, 444 = 21.819, P < 0.0001). There were significant differences in all three comparisons of a post-hoc Scheffe test (similarity score 3 vs similarity score 2, 3 vs 1, and 2 vs 1).
CC playback. At 6 weeks after pairing, males were placed in a cage in a sound-attenuating chamber equipped with a speaker (AS-5; Kenwood Corp., Tokyo, Japan) for at least 12 h prior to the start of stimulus presentation. Then, each male was exposed to a recording of his mate CC (group MATE; n = 4), a CC of a novel adult female (group UNFAMILIAR; n = 4) or kept in silence (group SILENCE; n = 4). We used the female's pre-pairing CC type that a male had imitated as mate CCs. For group UNFAMILIAR, we prepared four CCs from four females that were unfamiliar to the males; these recordings had been made before pairing in previous pairing experiments using different birds from those used in the present study. www.nature.com/scientificreports/ All playback was performed in the morning, as described previously 20 . Lights were switched off 15 min before the onset of playback. During playback, each bird was exposed to the repetition of a single CC (0.5 CC per sec) at regular intervals for 30 min. CCs were broadcasted at the peak value of 80 dB SPL (NA-14 Sound Level Meter; Rion Corp., Tokyo, Japan; A-weighting, slow response), measured 20 cm away from the speaker. The birds remained in darkness for 1 h after the end of CC playback and were then sacrificed. We monitored vocal behaviour of 24 birds during 30 min of playback. We adopted the criterion similar to that in a previous study 75 ; in brief, we excluded cases in which songs were produced or more than 5% of CCs broadcasted from the speaker were produced by the subject. None of the tested 12 males vocalised the song, nor produced CCs more than 5% of those broadcasted.
IEGs analysis. One hour after the end of exposure to the stimulus, the birds were administered an overdose of sodium pentobarbital (Nembutal, Abbott, Japan) and subsequently perfused intra-cardially with saline and a Zamboni fixative (4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS) containing 10.5% of a saturated picric acid solution). After fixation, dissected brains were divided into two coronal parts. They were cryoprotected in PBS containing sucrose and frozen and maintained at − 20 °C until cut. Frontal frozen brain sections were processed for Zenk-IR or Fos-IR cell nuclei by immunocytochemistry, as described previously 19,76 . We used polyclonal antibodies against Egr-1 (C-19, SC-189, dilution 1:15,000) or c-Fos (K-25, SC-253, dilution 1:15,000) raised in rabbit (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Adjacent sections were Nissl stained to enable identification of anatomical markers. Image analysis. In songbirds and parrots, Field L2 in the caudal telencephalon (analogous to the mammalian primary auditory cortex) receives input from the thalamic auditory nucleus ovoidalis (Ov) and constitutes a posterior auditory pathway (system) that includes Fields L1 and L3, the NCM, and CMM 4, 41,77 (Fig. 2b). In budgerigars, Jarvis and Mello 4 observed song-induced ZENK expression in the NCM and Fields L1 and L3. As the authors could not distinguish Nissl boundaries between the different nidopallial fields outside of L2, they designated the entire region that showed hearing-induced expression in the nidopallium as the budgerigar NCM, which surrounds the presumed Fields L1 and L3 (Fig. 2a, b). We were unable to distinguish the boundaries between the different nidopallial fields outside of L2 with Nissl staining, but we identified L2 and the LMV, which is a distinct boundary between the CMM and NCM 50 (see also Fig. 2a, b). We have followed Jarvis and Mello's 4 nomenclature and used the terms CMM and NCM 19,20,77 . In the present study, we sampled two regions within the NCM at dorsal and ventral levels (dNCM and vNCM), and CMM, comparable with those in our previous papers 19,20,76 .
We quantified the number of IEGs-IR cell nuclei (cells) in the CMM, dNCM, and vNCM. As a nonauditory-related control region (see Fig. 3a, b), we also investigated Zenk expression in the HP, as described previously 19,20,76 . The counting frames (290 × 450 μm) in the CMM, dNCM and vNCM were placed adjacent to the midline (Fig. 2a). In all cases, the counting frames were positioned by an experimenter blinded to the experimental condition of each animal. Photomicrographs of the counting frames were captured by a digital camera with 8-bit intensity resolution (256 levels of grey). We took a total of four samples from both hemispheres (two from each hemisphere) per region (CMM, dNCM, vNCM, and HP) per subject. For each region, the mean number of IEGs-IR cells were calculated for the dNCM and HP at 1.00 and 1.06 mm caudal to coordinates zero (level 'a' in Fig. 2a, b), and for the CMM and vNCM at 2.00 and 2.06 mm caudal to coordinates zero (level 'b' in Fig. 2a,  b). Image analysis was performed in a semiautomated manner with a PC-based system equipped with KS400 version 3.0 software (Carl Zeiss Vision, Oberkochen, Germany), as described in detail previously 76 . Counting was performed blinded to the experimental history of the subjects.
Statistical analysis of IEGs expressions. IEG expression was analysed with analysis of variance (ANOVA) with factor Group (MATE, UNFAMILIAR, or SILENCE) and factor Brain Region. Subsequently, we used one-way ANOVAs for individual brain regions. Post-hoc comparisons were performed using Bonferroni tests. Summary of replications: 12 animals per IEG, four animals per group, and four 290 × 450 μm counting fields per brain region (two on each side) per animal. The relationship between the IEG expressions in each pair of subregion was examined by a Pearson's product-moment correlation analysis. The data were log transformed before statistical analysis to satisfy the assumptions of the parametric tests and analysed using StatView version 5 (SAS Institute Inc., Cary, NC, USA). Level of significance was set at P < 0.05.
In order to examine the relationship between the IEG expressions in each pair of subregions, we used a simple regression model by Bayesian analysis. Bayesian analysis was employed because maximum likelihood methods require large data sets 78 . For a regression coefficient, we calculated the 95% confidence interval (CI). If the CI of a regression coefficient (e.g. the coefficient showing the effect of dNCM on CMM) does not include zero (the value in our null hypotheses), it indicates that there is a significant linear relationship between the two variables (dNCM and CMM in this example).

Data availability
The datasets generated and/or analysed during the current study are available from the corresponding authors on request.