Skip to main content
NCBI home page
Journal of Neurophysiology logoLink to Journal of Neurophysiology
. 2016 Sep 14;116(6):2831–2840. doi: 10.1152/jn.00622.2016

Social context differentially modulates activity of two interneuron populations in an avian basal ganglia nucleus

Sarah C Woolley 1,✉
PMCID: PMC5168002  PMID: 27628208

Basal ganglia activity can be modulated by behavioral context. In zebra finches, the output of a basal ganglia circuit dedicated to singing is influenced by the social context in which a bird sings. However, it is unknown whether other cell types within the nucleus show similar social modulation of activity. This study provides a novel description of the socially modulated activity of two additional basal ganglia cell types, fast-spiking interneurons and external globus pallidal neurons, during singing.

Keywords: basal ganglia, electrophysiology, globus pallidus, songbird, striatum

Abstract

Basal ganglia circuits are critical for the modulation of motor performance across behavioral states. In zebra finches, a cortical-basal ganglia circuit dedicated to singing is necessary for males to adjust their song performance and transition between spontaneous singing, when they are alone (“undirected” song), and a performance state, when they sing to a female (“female-directed” song). However, we know little about the role of different basal ganglia cell types in this behavioral transition or the degree to which behavioral context modulates the activity of different neuron classes. To investigate whether interneurons in the songbird basal ganglia encode information about behavioral state, I recorded from two interneuron types, fast-spiking interneurons (FSI) and external pallidal (GPe) neurons, in the songbird basal ganglia nucleus area X during both female-directed and undirected singing. Both cell types exhibited higher firing rates, more frequent bursting, and greater trial-by-trial variability in firing when male zebra finches produced undirected songs compared with when they produced female-directed songs. However, the magnitude and direction of changes to the firing rate, bursting, and variability of spiking between when birds sat silently and when they sang undirected and female-directed song varied between FSI and GPe neurons. These data indicate that social modulation of activity important for eliciting changes in behavioral state is present in multiple cell types within area X and suggests that social interactions may adjust circuit dynamics during singing at multiple points within the circuit.

NEW & NOTEWORTHY

Basal ganglia activity can be modulated by behavioral context. In zebra finches, the output of a basal ganglia circuit dedicated to singing is influenced by the social context in which a bird sings. However, it is unknown whether other cell types within the nucleus show similar social modulation of activity. This study provides a novel description of the socially modulated activity of two additional basal ganglia cell types, fast-spiking interneurons and external globus pallidal neurons, during singing.

basal ganglia circuits are critical for the learning, plasticity, and performance of motor behaviors and are conserved across vertebrates. Activity within these circuits appears to be dependent not just on the movement to be executed but also on the behavioral context, such as reward or cue presentation (Berke 2011; Gage et al. 2010; Garenne et al. 2011; Gdowski et al. 2001; Kawagoe et al. 2003; Lau et al. 2010; Lee and Assad 2003). Dissecting the role of particular cell types within basal ganglia circuits is crucial to revealing how these circuits integrate information about external cues to modulate behavior.

Songbirds offer an excellent opportunity to study the role of basal ganglia circuits in regulating motor output. Both developmental song learning and later adult song plasticity are dependent on a specialized cortical-basal ganglia-thalamic loop, the anterior forebrain pathway (AFP), dedicated to a single behavioral function: singing (Ali et al. 2013; Bottjer et al. 1984; Kao et al. 2005; Ölveczky et al. 2005; Scharff and Nottebohm 1991; Warren et al. 2011). The basal ganglia nucleus within the AFP, area X, shows considerable homology with the mammalian basal ganglia, including the presence of multiple neuron types with molecular signatures and activity patterns similar to those described in mammalian striatum and pallidum (Fig. 1; Carrillo and Doupe 2004; Farries and Perkel 2002; Goldberg and Fee 2010; Goldberg et al. 2010; Pfenning et al. 2014; Reiner et al. 2004). Moreover, it is possible to study the activity within these circuits as an animal produces the same behavior in different states. For example, zebra finches naturally sing the same sequence of vocal elements, known as a motif, as they perform a self-initiated (“undirected”) song when alone and a courtship song when presented with a female (“female-directed” song; Jarvis et al. 1998; Kao and Brainard 2006; Woolley and Doupe 2008; Zann 1996).

Fig. 1.

Fig. 1.

Open in a new tab

Song system. A: diagram of connections between the anterior forebrain pathway and motor pathway with reference to similar structures in mammals. HVC, proper name; RA, robust nucleus of the arcopallium; LMAN, lateral magnocellular nucleus of the anterior nidopallium; DLM, medial nucleus of the dorsolateral thalamus. B: connections among different striatal (shaded circles) and pallidal cell types (open circles) in the avian basal ganglia nucleus area X. Dashed lines indicate hypothesized connections, lines ending in arrows indicate excitatory connections, and lines ending in solid circles are inhibitory connections. FSI, fast-spiking interneuron; SN, medium spiny neuron; GPe, external globus pallidus neuron; GPi, internal globus pallidus neuron.

Investigation of neural activity under different social contexts allows us to study the same motor sequence with subtle changes in performance. For example, previous work found that the output neurons of area X (internal pallidal or GPi neurons) exhibit higher, more variable firing during undirected than during female-directed singing (Hessler and Doupe 1999; Woolley et al. 2014; Wooley and Kao 2015). Such changes to GPi firing with social context contribute to variation in neural activity in the cortical output of the AFP, the lateral magnocellular nucleus of the anterior nidopallium (LMAN; Kojima et al. 2014), highlighting the importance of such changes to neural circuit dynamics. Moreover, similar socially modulated changes are not apparent in the activity of inputs to area X from the cortical nucleus HVC (Woolley et al. 2014), indicating that these differences may arise in area X. However, the mechanisms by which these context-dependent differences in GPi activity are generated remain largely unknown.

One possibility is that other cell types in area X, including parvalbumin-positive fast-spiking interneurons (FSI) and external pallidal neurons (GPe), contribute to the modulation of basal ganglia output. An important step in understanding how different neuron types shape the output of the basal ganglia is to reveal the degree to which behavioral context modulates activity of distinct cell types. To this end, I investigated electrophysiological activity in FSI and GPe neurons in male zebra finches as they transitioned between singing female-directed and undirected song. I found that similar features of activity were modulated by social context for both cell types but that the magnitude and direction of changes to the patterns and rates of firing were distinct between FSI and GPe neurons. The details of FSI and GPe differences in signaling mode lend insight into their potential roles in circuit dynamics and the rapid adjustment of motor performance in different behavioral states.

MATERIALS AND METHODS

Animals

Adult male zebra finches (>120 days old; n = 10) were born and raised in a breeding colony at the University of California, San Francisco. Birds were housed with same-sex conspecifics, kept on a 14:10-h light-dark photoperiod, and provided finch seed, grit, and water ad libitum. All procedures were approved by the Institutional Animal Care and Use Committees at the University of California, San Francisco and McGill University in accordance with Canadian Council on Animal Care and NIH guidelines for the care and use of animals.

Surgery for Electrophysiology

Males were selected on the basis of spontaneous singing frequency and body size and were housed in isolation in a small cage inside a sound-attenuating chamber (soundbox; Acoustic Systems, Austin, TX) a few days before surgery. Surgery for the implantation of a lightweight microdrive carrying electrodes was performed as previously described (Hessler and Doupe 1999; Kao et al. 2008; Woolley et al. 2014). Briefly, birds were anesthetized with an intramuscular injection of equithesin (0.85 g of chloral hydrate, 0.21 g of pentobarbital sodium, 0.42 g of MgSO4, 2.2 ml of 100% ethanol, and 8.6 ml of propylene glycol to a total volume of 20 ml with water) or an intramuscular injection of ketamine and midazolam followed by gaseous isoflurane (0.2–1.0%). Males were positioned into a stereotaxic device, stabilizing the head at the beak and ear canals with the beak at an angle of 50° from vertical. Two to three tungsten electrodes with impedances of 3–5 MΩ (MicroProbes, Gaithersburg, MD), carried by a lightweight microdrive, were stereotaxically targeted to area X on the right side. An uninsulated ground electrode was implanted on the contralateral side such that the tip was within a few millimeters of the target nucleus. The microdrive, ground electrode, and connector plug were secured to the skull with epoxy and dental cement.

Electrophysiological Recording

After surgery, each male was housed individually in a cage inside of a soundbox. On the day of recording, the male was attached to a rotating commutator (Dragonfly, Ridgeley, WV) via a flexible lead containing a unity gain operational amplifier (Texas Instruments, Dallas, TX). The commutator was connected to a microelectrode amplifier (A-M Systems, Colorado Springs, CO) that amplified (×1,000) and filtered (0.3–10 kHz) neurophysiological data. Sounds were recorded by a small microphone (Countryman Associates, Menlo Park, CA) inside the soundbox and were bandpass filtered from 0.3–10 kHz (Krohn-Hite, Brockton, MA). Behavior was monitored through a video camera inside the soundbox.

On each recording day, neural activity was monitored on an oscilloscope and audio monitor while the electrodes were lowered until large, spontaneous action potentials were encountered. Electrode position was adjusted so that the action potentials could be clearly differentiated from background (e.g., Fig. 2, A, B, F, G). Spontaneous activity, while the bird was sitting quietly, as well as neural activity during singing, were recorded using a custom-written computer program [A. Leonardo (California Institute of Technology, Pasadena, CA) and C. Roddey (University of California, San Francisco, CA)]. For recording female-directed song, a small cage containing a muted female (see Woolley and Doupe 2008 for muting details) was placed next to the male in the soundbox, and the male was viewed on the video monitor to determine whether he performed female-directed song. During courtship, males orient toward the female, fluff the body feathers while flattening feathers on top of the head, and hop, dance, and beak wipe while singing the female-directed song (Jarvis et al. 1998; Kao and Brainard 2006; Woolley and Doupe 2008; Zann 1996). Only songs during which males performed at least two of the above courtship components while singing were considered female-directed songs. For each presentation, the female remained in the soundbox for up to 10 min during which time males could perform multiple bouts of female-directed song. After the female was removed, I waited at least 10 min or until the male sang an equivalent amount of undirected song before presenting the female again. For each neuron that I recorded, female-directed and undirected songs were collected in this interleaved fashion for 15 min to several hours, depending on the stability of the recording site.

Fig. 2.

Fig. 2.

Open in a new tab

Social modulation of activity in 2 interneuron types. A and B: raw traces of singing-related activity during a bout of female-directed (A) and undirected singing (B) in an FSI neuron. Top panel of each is a spectrogram of the song. C: spectrogram of a single female-directed motif (FD; top) and undirected motif (UD; bottom). D: raster plots of activity during the motif for FD (top) and UD (middle) singing and during an equivalent duration of silence (SP; bottom). E: peristimulus time histogram (PSTH) of the motif-related activity shown in D. F–J: examples of the same singing-related and spontaneous activity in a GPe neuron.

After all recordings for each bird were completed, small electrolytic lesions were made at different depths along the recording track. Then animals were deeply anesthetized using isoflurane and transcardially perfused with 0.9% saline followed by 3.7% formaldehyde. Electrode tracks and lesions were localized on 40-μm Nissl-stained sections, and all recorded sites were verified to lie within area X.

Data Analysis

Spike sorting.

Neural activity was analyzed offline using custom software written in the MATLAB programming language (The MathWorks, Natick, MA). Large single units were discriminated using a Bayesian spike-sorting program (Kao et al. 2008; Lewicki 1994; Woolley et al. 2014). Isolation of single units was verified using visual inspection of the waveforms and the presence of a refractory period in the interspike interval (ISI) histogram. For all area X neurons, fewer than 0.1% of ISIs were <0.7 ms.

Cell-type identification.

I recorded from multiple cell types within area X. As previously described (Goldberg and Fee 2010; Woolley et al. 2014), pallidal neurons could be distinguished from striatal neurons on the basis of firing rate. Moreover, both striatal and pallidal neurons could then be further subdivided by using a combination of the spontaneous firing rate and the Fano factor during undirected singing. With the use of this approach, there were two classes of pallidal neurons that corresponded to those described previously (Goldberg et al. 2010; Tanaka et al. 2016; Woolley et al. 2014). Specifically, I identified one population that, like mammalian GPi neurons, showed substantial increases in firing and patterned pauses during both female-directed and undirected singing. A second population of pallidal neurons had firing characteristics similar to those hypothesized to be GPe neurons, including slight increases in firing and long bursts and long pauses during undirected singing, which were reflected in a Fano factor greater than 1. Because a previous publication characterized GPi-like neurons (Woolley et al. 2014), I focus on GPe-like neurons in this article.

Among the striatal neurons that I recorded, two classes of cells could be distinguished by using a combination of the spontaneous firing rate and the Fano factor during the production of undirected song. The first class consisted of cells with almost no spontaneous activity and a Fano factor of around 1. These cells exhibited sparse bursts or single spikes during singing and showed substantial similarity to the neurons presumed to be medium spiny neurons (MSNs; Goldberg and Fee 2010; Woolley et al. 2014). A second class of neurons had higher spontaneous firing rates and a Fano factor greater than 1, which reflects the prevalence of bursting during singing. The class of higher firing bursting neurons could be further separated on the basis of their waveforms into narrow (spike widths <0.06 ms) and wide (>0.06 ms) spike populations. As previously described (Goldberg and Fee 2010; Pidoux et al. 2015), the neurons with narrow spike widths putatively correspond to FSI neurons and are analyzed in this report.

Time warping of spike trains.

To compensate for variation in syllable and interval timing across individual renditions of a bird's song, I performed a piecewise linear time warp of each motif to linearly stretch or compress each syllable and interval to match the corresponding duration of a reference motif (see Kao et al. 2008; Woolley et al. 2014). All analyses of singing-related activity were conducted on time-warped spike trains.

Measures of Within-Trial Spiking Statistics

I investigated the regularity of spiking using two measures of ISIs. First, as an overall measure of spike-train variability, I calculated the coefficient of variation (CV) as the standard deviation of the ISI distribution divided by its mean. However, because the CV can overestimate the irregularity of bursty neurons, I also investigated the regularity of spiking independent of the influence of bursts by using the CV2, which is the CV of two consecutive ISIs (Holt et al. 1996; Kumbhare and Baron 2015). Assuming that spikes in a train occur at times ti (0 ≤ i ≤ N), where N is the number of spikes in the train, then the ISI is defined as

Δti=ti−ti−1,

The CV2 for spike i can then be calculated as

CV2=2|Δti+1−Δti|Δti+1+Δti.

I then calculated the mean CV2 for all spikes in a motif rendition and the mean CV2 across all renditions in each context.

To gain further insight into the degree to which social context and cell type affected spiking statistics, I calculated autocorrelograms (1-ms bin width, 200-ms duration) for each neuron in each context (spontaneous, female-directed singing, and undirected singing). For comparison, I also calculated Poisson autocorrelograms based on the statistics for each neuron in each condition and determined a 99.9% confidence limit for each cell for the expected bin contents (Abeles 1982; Bastian and Nguyenkim 2001). Autocorrelograms from Poisson spike trains are expected to be flat. Thus peaks or depressions in the autocorrelograms of firing, determined to be unlikely assuming a Poisson spike train, can be taken as the indication of nonrenewal firing. There appeared to be four categories of correlograms: positive correlation functions, negative correlation functions, correlation functions that alternated in sign, and flat correlation functions (Avila-Akerberg and Chacron 2011). For autocorrelation functions characterized as positive or negative, I measured the peak and width of the function. The peak was the point (positive or negative) that was farthest from zero. The width was the lag at which either the falling phase (for positive correlations) or the rising phase (for negative correlations) of the initial peak of the autocorrelogram crossed the 99.9% Poisson confidence limit for both cell types in each context (Khosravi-Hashemi et al. 2011). In the case of oscillatory functions, where there were alternating positive and negative peaks in the autocorrelogram, I counted the number of positive and negative peaks that were outside of the Poisson confidence limit.

Measures of Across-Trial Spike Statistics

To look at the degree of coherence in the average pattern of firing between female-directed and undirected singing, I computed the correlation coefficient between the mean peristimulus time histograms (PSTHs; generated using a 10-ms Hanning window) for activity during female-directed and undirected singing. Correlation coefficients were also calculated for spike trains with randomly introduced time shifts (Ölveczky et al. 2005). For this analysis, I performed random circular shifts for each trial such that spikes wrapped around to the beginning of the motif, and then I computed the PSTH for the shifted data, as well as the correlation between PSTHs for shifted female-directed and shifted undirected shifted PSTHs. I repeated this process 100 times and calculated the average and standard deviation of the correlation coefficients across iterations. I then tested whether the correlations for recorded activity were different from the correlations for shifted spike trains (mean + 2SD) and, additionally, whether this was influenced by cell type. I performed a mixed-effects ANOVA with data type (recorded vs. simulated data) and cell type (GPe, FSI) as independent variables and the correlation coefficients of the PSTHs as the dependent variable. I included male identification (ID) and recording date nested in male ID as random variables in the model to control for multiple recordings from the same bird.

To quantify trial-by-trial variability in activity across the entire motif, I computed the cross-correlation (CC) between the instantaneous firing rates for all pairs of motifs (Kao et al. 2008; Woolley et al. 2014). Briefly, for each trial, I smoothed the spike train with a Gaussian filter (SD = 10 ms) and then subtracted the time-averaged firing rate for that trial to estimate the instantaneous firing rate for that trial. I then computed the average CC between the instantaneous firing rates for all pairs of trials within a social context.

I used a similar approach to investigate the influence of bursting on the precision of firing. Because the autocorrelograms indicated there was little or no bursting during female-directed singing, these analyses focus exclusively on undirected singing. Specifically, I analyzed the trial-by-trial CCs for undirected spike trains that were split into isolated spikes and into burst onsets (Khosravi-Hashemi et al. 2011). Bursts were designated using all possible burst thresholds, at 1-Hz increments, between 10 and 700 Hz. At each threshold, I generated separate arrays of burst onsets and isolated spikes, and then calculated the trial-by-trial CC independently for each array and measured the peak in the CC function. To determine a lower limit to the trial-by-trial CC, 100 iterations of random circular shifts, as described above, were generated from the undirected activity for each cell, and the trial-by-trial CC was calculated for each iteration. A mixed-effects ANOVA was used with spike-train type (undirected isolated spikes, undirected burst onsets, and circularly shifted data) and cell type (GPe and FSI) as independent variables and the peak trial-by-trial CC as the dependent variable.

Statistical Analyses

For all analyses I used mixed-effects ANOVA models. Unless otherwise noted, I used cell type (FSI or GPe) as the independent variable for analyses involving a single context (e.g., autocorrelation analyses). To look additionally at the effect of context for a variety of different neurophysiological measures, I included cell type, context (directed singing, undirected singing, and spontaneous activity), and the interaction between cell type and context as independent variables. I included bird ID and recording date nested within bird ID to control for multiple recordings within the same individual. Tukey's honestly significant difference tests were used for all post hoc contrasts, and I set α = 0.05.

RESULTS

To date, the singing-related activity of FSI and GPe neurons has only been described in juvenile zebra finches performing undirected song (Goldberg and Fee 2010; Goldberg et al. 2010; Pidoux et al. 2015). In general, I found that the activity patterns of these neuron types in adult songbirds performing undirected song were similar to those observed in juveniles. In addition, I found that both neuron types were modulated by social context and that the nature of modulation varied between FSI and GPe neurons.

Singing and Social Context Differentially Modulate the Firing Rates of FSI and GPe Neurons

Both FSI and GPe neurons showed singing-related activity, evident in the raster plots and the peaks and troughs in the PSTH for each neuron (Fig. 2). The song-locked patterns of activity were accompanied by singing-related changes in firing rates in both cell types (Goldberg and Fee 2010). Moreover, there were significant differences between the cell types in their firing rates [Fig. 3, A–D; F(1,25.3) = 66.9, P < 0.0001], with higher firing rates overall for GPe neurons than for FSI neurons. In addition, the effects of behavioral context on firing rate significantly varied across cell types [Fig. 3, A–D; F(2,32.6) = 3.7, P = 0.0351]. FSI and GPe neurons differed in their spontaneous firing rates, in the direction and magnitude that firing changed during singing, and in the degree of social modulation (Figs. 2 and 3).

Fig. 3.

Fig. 3.

Open in a new tab

Social context differentially affects the rate and regularity of firing in FSI and GPe neurons. A and B: interspike interval (ISI) plots for FSI and GPe neurons, respectively, during spontaneous (SP) activity and during female-directed (FD) and undirected singing (UD). C: in FSI neurons, firing rates are increased during UD singing. D: firing rates in GPe neurons are increased during UD singing and either unchanged or decreased during FD singing relative to SP activity. The coefficient of variation (CV) of the ISI was highest during UD singing in both FSI (E) and GPe (F) neurons. In GPe neurons (F), the CV of the ISI was also significantly higher during FD singing than during SP activity. G: the CV2 of spiking in FSI neurons was not significantly different (NS) between SP, FD, and UD birds. H: in contrast, the CV2 of GPe neurons increased during both FD and UD singing compared with the tonic and regular firing during SP activity. *P < 0.05.

For FSI neurons, firing rates increased during singing, especially undirected singing. For example, the neuron in Fig. 2 had mean firing rates of 14 ± 1 Hz when the bird was silent (spontaneous activity), 27 ± 1 Hz when the bird produced female-directed song, and 37 ± 2 Hz when the bird produced undirected song. Across all FSI neurons, firing rates during female-directed singing were slightly but not significantly higher than spontaneous activity (P = 0.3279), whereas firing rates during undirected singing were significantly higher than spontaneous activity (P = 0.0429). However, singing-related firing rates during undirected and female-directed singing were not significantly different (Fig. 3C; P = 0.8367).

The firing rates of putative GPe neurons were also modulated by singing and behavioral state (Figs. 2, F–J, and 3D). However, unlike the FSI neurons, GPe firing showed both increases and decreases in firing depending on the social context. GPe neurons exhibited high, regular firing when birds were silent (112 ± 10 Hz; mean ± SE), and firing rates increased when birds produced undirected song (150 ± 24 Hz; Fig. 2, I and J). In contrast, GPe neurons displayed either no change or, in many cases, a decrease in firing during female-directed song relative to spontaneous activity (81 ± 10 Hz). For example, the neuron depicted in Fig. 2 fired at 122 ± 4 Hz when the bird was silent and increased to 262 ± 9 Hz when the male produced undirected song but decreased to 60 ± 5 Hz when the male produced female-directed song. Across GPe neurons, firing rates during female-directed song were not significantly different from spontaneous firing rates; however, firing rates during undirected song were significantly higher than both spontaneous activity (Fig. 3D; P = 0.0064) and activity during female-directed song (Fig. 3D; P < 0.0001).

Social Context Affects Firing Regularity in FSI and GPe Neurons

In addition to differences in firing rate, social context could modulate the temporal dynamics of firing in FSI and GPe neurons. For example, the shape of the ISI distributions differed between cell types and contexts, and I quantified these differences using the CV of the ISI (Fig. 3, E and F). There was a significant interaction between cell type and context on the CV of the ISI [F(2,29.5) = 3.4, P = 0.0480]. In particular, the CV of the ISI was higher during undirected than female-directed singing in both FSI (Fig. 3E; P < 0.0001) and GPe neurons (Fig. 3F; P = 0.0148). Whereas both neuron types displayed a significant social modulation, they differed in the variability of their spontaneous activity. The CV of FSI neurons during silence was significantly lower (more regular) than during undirected singing (P < 0.0001) but was not different from the regularity during female-directed singing (Fig. 3E; P = 0.9951). For GPe neurons, the spontaneous firing was highly regular during spontaneous activity, and the CV of the ISI was significantly lower during silence than during either female-directed (P = 0.0108) or undirected singing (Fig. 3F; P = 0.0001). The CV of spontaneous firing of GPe neurons was also lower than the CV of spontaneous firing of FSI neurons (P = 0.0538).

Because the CV can overestimate irregularity in firing due to bursting, I also calculated the CV2 as a measure of local variation (see materials and methods). The CV2 was significantly affected by cell type [F(1,34) = 10.2, P = 0.0030], context [F(2,26) = 8.4, P = 0.0015], and the interaction between cell type and context [F(2,26) = 8.8, P = 0.0012]. Overall, the CV2 was significantly higher in FSI than GPe neurons (Fig. 3, G and H). Moreover, as was the case for the CV, the CV2 indicated that the spontaneous firing of GPe neurons was significantly more regular than either singing-related firing in GPe neurons (Fig. 3H; undirected singing vs. spontaneous activity, P = 0.0002; female-directed singing vs. spontaneous activity, P = 0.0003) or than spontaneous firing of FSI neurons (Fig. 3G; P = 0.0002). However, in contrast to the CV, there was not a social modulation of the CV2 in either FSI neurons or in GPe neurons (Fig. 3, G and H; female-directed vs. undirected singing, P > 0.20 for both cell types). Thus both cell types have higher CVs of the ISI but similar CV2s during undirected compared with female-directed song. Given that the CV is more influenced by bursting than the CV2, this increase in irregularity could relate to the increase in bursting during undirected singing. In addition, although the regularity of firing is similar between the cell types during female-directed singing, because the spontaneous activity of GPe neurons is more regular than that of FSI neurons, the cell types differ in the change in regularity between spontaneous and female-directed singing.

To further investigate social context effects on the temporal dynamics of firing, I characterized how the firing of both cell types compared with Poisson firing by using the autocorrelation functions for each context (Fig. 4, A and B; Bastian and Nguyenkim 2001; Goldberg and Fee 2010; Khosravi-Hashemi and Chacron 2012; Khosravi-Hashemi et al. 2011). In both cell types, there were differences in the shape of the autocorrelation functions depending on social context, and these were reflected in a social modulation of the height of the autocorrelation peak [Fig. 4, A–C; F(1,23) = 52.8, P < 0.0001]. In particular, the peak was significantly higher during undirected than female-directed singing in both FSI (P = 0.0025) and GPe neurons (Fig. 4C; P < 0.0001). However, although the social modulation was evident in both cell types, there were also differences in the shape of the autocorrelation function between the cell types (Fig. 4, A and B). For example, whereas FSI and GPe neurons had similar peak heights during undirected singing (P = 0.9999), they differed significantly in the width of the peak. FSI neurons took less time to converge to Poisson levels during undirected singing than did GPe neurons (Fig. 4, A and D; P = 0.0003). The cell types also differed in the shape of the autocorrelation functions during female-directed singing. Whereas the autocorrelation functions during female-directed singing were generally flat for FSI neurons, they had negative peaks in GPe neurons (Fig. 4C; P = 0.0494). Thus the ISI correlations in FSI and GPe neurons are differentially affected by social context.

Fig. 4.

Fig. 4.

Open in a new tab

Social context differentially modulates spike statistics in FSI and GPe neurons. A: examples of the autocorrelation functions for single FSI (top) and GPe neurons (bottom) during undirected singing (UD; dark blue line), female-directed singing (FD; light blue line), and spontaneous activity (SP; gray line) highlighting the differences between cell types and social context in the correlations. B: autocorrelation functions for all FSI (top) and GPe neurons (bottom) for UD (dark blue), FD (light blue), and SP (gray). Dark lines indicate the mean and shading represents the SD. C: differences in the peak of the autocorrelation functions distinguish between social contexts in each cell type and between cell type functions during FD singing. a,b,cP < 0.05, groups not indicated by the same letter are significantly different from each other. D: GPe and FSI neurons also differed in the width of the autocorrelation function during UD singing, measured as the latency to reach Poisson firing. *P < 0.05.

The autocorrelation functions during spontaneous activity also differed between GPe and FSI neurons. GPe neurons displayed oscillations in the autocorrelation functions, whereas autocorrelation functions were flat for FSI neurons (Fig. 4, A and B). To quantify this difference, I counted the number of positive and negative peaks that were above or below the Poisson confidence limit (see materials and methods). The autocorrelation function during spontaneous activity for every GPe neuron had at least one positive and one negative peak outside the Poisson confidence limit, with an average 1.3 positive peaks (range 1–2) and 1.7 negative peaks (range 1–3). In contrast, there were no positive or negative peaks in the autocorrelation functions that fell outside the Poisson confidence limit for FSI neurons during spontaneous activity. Overall, GPe neurons demonstrated more positive [F(1,14.8) = 22.2, P = 0.0003] and negative peaks [F(1,14.9) = 34.3, P < 0.0001] than FSI neurons.

Firing Precision Across Trials is Higher During Female-Directed Singing for Both Cell Types

Previous work has found differences between MSNs and GPi neurons in the consistency of firing patterns across trials and the modulation of this precision by the social context (Woolley et al. 2014). For example, the timing of MSN spikes across trials during both female-directed and undirected singing is highly precise and not modulated by context, whereas GPi neuron firing is more variable overall and more modulated by social context than MSN firing (Woolley et al. 2014).

To investigate the degree to which the cross-trial precision of FSI and GPe neurons was modulated by context, I analyzed the trial-by-trial CC of the instantaneous firing rate when birds produced undirected and female-directed song (Kao et al. 2008; Woolley et al. 2014). Generally speaking, the trial-by-trial CC of spike timing across trials of FSI and GPe neurons was lower than that observed for both MSNs and GPi neurons in area X (Woolley et al. 2014). For example, the trial-by-trial CCs of FSI and GPe neurons when birds produced female-directed song were, respectively, 0.36 ± 0.03 and 0.27 ± 0.03, which were lower than the trial-by-trial CC of 0.49 ± 0.03 observed for GPi neurons and 0.88 ± 0.04 seen for MSNs (Woolley et al. 2014).

Moreover, the trial-by-trial CC was modulated by social context for both FSI and GPe neurons [F(1,10.8) = 67.0, P < 0.0001] and not different between the cell types [F(1,21) = 2.9, P = 0.1450]. Overall, the trial-by-trial CC was significantly greater when birds produced female-directed song (Fig. 5, A and B; FSI: 0.36 + 0.03; GPe: 0.27 + 0.03) than when they produced undirected song (Fig. 5, A and B; FSI: 0.11 + 0.03; GPe: 0.10 + 0.03) for both FSI (Fig. 5A; P = 0.0001) and GPe neurons (Fig. 5B; P = 0.0045). The trial-by-trial CCs were all significantly different from trial-by-trial CCs obtained from shifted data (for all comparisons, P < 0.05), indicating that firing is significantly patterned during singing (see materials and methods).

Fig. 5.

Fig. 5.

Open in a new tab

Social context and cell type affect trial-by-trial variability. The stereotypy of firing across trials, as measured by the trial-by-trial cross correlation (CC), was significantly higher during female-directed (FD) than undirected (UD) singing in both FSI (A) and GPe (B) neurons. C and D: the greater variability during UD song resulted from increased variability of isolated spikes as well as bursts. The peak trial-by-trial CC was not different between undirected spike trains, isolated spikes, or burst onsets in FSI (C) and GPe neurons (D). In addition, values for all three were higher than for randomly circularly shifted data (shading near 0 represents mean + SD for shifted data) and lower than the trial-by-trial CC for FD singing (long-dashed line). E: correlation coefficients between the average PSTH during female-directed and undirected singing for all FSI and GPe neurons. All correlation coefficients were significantly higher than the mean (dashed line) + 2SD (gray shaded area) for randomly circularly shifted data. However, FSI and GPe neurons were not different. * P < 0.05.

Given the results of firing regularity seen with the CV and CV2 analyses, I also investigated how bursting contributed to the pattern and variability of firing across trials for undirected song. For these analyses, recorded spike trains were separated into spike trains of burst onsets and isolated spikes using burst thresholds at 1-Hz intervals between 10 and 700 Hz (see materials and methods). The trial-by-trial CCs of the instantaneous firing rate were calculated for isolated spike or burst onset spike trains at each threshold, and the peak trial-by-trial CCs for isolated spike or burst onset spike trains were measured and analyzed.

There was a significant effect of spike-train type [F(2,22.7) = 23.0, P < 0.0001] but not cell type [F(1,18.6) = 1.1, P = 0.3206] on the peak trial-by-trial CC. For both cell types, isolated spike trains had higher trial-by-trial CC values than circularly shifted data (Fig. 5, C and D; FSI: P = 0.0004; GPe: P = 0.0033). In addition, the CC values for burst onsets were also higher than for circularly shifted data; however, this difference was only significant for FSI neurons (Fig. 5, C and D; FSI: P = 0.0129; GPe P = 0.1893). Thus there appears to be greater variability in the timing of bursts than the timing of isolated spikes across trials. To address whether removing bursts was sufficient to make undirected spiking “female-directed-like” in terms of precision, we also compared isolated spike trains during undirected singing to spike trains during female-directed singing. For both FSI and GPe neurons, we found that the firing precision during female-directed singing was significantly greater than the firing precision of undirected isolated spike trains (P < 0.05 for both cell types).

Both GPe and FSI Neurons Display Similar Song-Locked Firing Patterns Between Female-Directed and Undirected Singing

Despite substantial context-dependent differences in firing regularity within trials and spike precision between trials, both cell types had similar song-locked patterns of activity between female-directed and undirected singing (Fig. 2, E and J). In the example FSI neuron depicted in Fig. 2, the correlation of the PSTHs during female-directed and undirected singing is 0.67. Across all cells, the average CC between undirected and female-directed song for FSI neurons was 0.58 ± 0.08 (mean ± SE). Similarly, for the GPe neuron depicted in Fig. 2, the CC of the time-averaged firing patterns was 0.62, and the average cross-correlation for all cells was 0.50 ± 0.07. The average coherence between female-directed and undirected firing patterns, for both cell types, was significantly greater than that for randomly shifted spike trains [Fig. 5, C and D; F(1,12.5) = 52.4, P < 0.0001]. However, coherence was not significantly different between FSI and GPe neurons [Fig. 5E; F(1,12.5)=0.7, P = 0.4227]. Thus, for both FSI and GPe neurons, the average pattern of neurophysiological activity was similar when males performed undirected and female-directed song.

DISCUSSION

Basal ganglia circuits integrate motor and contextual information that is critical for the learning and execution of movements. In songbirds, singing-related activity of MSNs and GPi neurons within the basal ganglia nucleus area X is modulated by social context, and the connection between these neurons may be a locus for the generation of context-dependent variability (Fig. 1B; Woolley et al. 2014; Woolley and Kao 2015). In this study, I discovered that the neurophysiological activity of two interneuron types within area X, namely, FSI and GPe neurons, is also markedly modulated by singing and social context. In both GPe and FSI neurons, activity during female-directed singing was characterized by lower firing rates, less bursting, and higher trial-by-trial cross-correlations than the activity during undirected singing. However, because of differences in the spontaneous firing and activity during undirected singing, the magnitude and direction of changes to firing rate and spiking statistics during the performance of female-directed song varied between FSI and GPe neurons. This indicates that FSI and GPe neurons may encode information about social context in distinct ways, highlighting their different roles in shaping information outflow from the basal ganglia.

Previous studies have hypothesized that courtship song may represent a “cued” state, whereas the self-initiated, undirected song may represent an “uncued” behavior (Rajan and Doupe 2013; Woolley et al. 2014). Major support for this idea came from comparisons of GPi activity between the performance of cued and uncued behaviors in rodents and of GPi activity between the performance of female-directed and undirected song in songbirds (Gdowski et al. 2001; Lee and Assad 2003; Woolley et al. 2014). However, little is known about the degree to which other neuron types in the basal ganglia encode information about cued vs. uncued motor performance. I discovered that the activity of FSI and GPe neurons in area X also varied when males sang to females compared with when they sang while alone. Given the similarities in basal ganglia organization across birds and mammals (e.g., Doupe et al. 2005), these data suggest that mammalian FSI and GPe neurons may change their firing patterns in distinct ways across the performance of cued and uncued behaviors.

GPi neurons, the output neurons of area X (Fig. 1B), have high firing rates interrupted by brief pauses in spiking that carry the pattern of activity. Across trials, the pauses are reliable and precise during female-directed singing, and substantially less reliable and precise during undirected singing. In contrast, the timing of upstream MSN spikes, which provide inhibitory input thought to be responsible for the pauses in GPi firing, are equally precise in both social conditions. This raises the question of how variability in GPi pauses is generated. The context-dependent differences in the activity of both FSI and GPe neurons, combined with the fact that the two cell types may connect to the MSN-GPi circuit at different points, suggest particular hypotheses for how FSI and GPe neurons might influence GPi output and the social modulation of variability.

In mammals, FSI neurons are thought to provide subtle, modulatory input to MSNs and may be pivotal in shaping the timing of MSN spikes, in particular, the timing of spikes within bursts (Berke 2011; Gittis et al. 2011). However, although the importance of FSI firing for behavior has been demonstrated through pharmacological manipulations (Gittis et al. 2011), the lack of overt effects of FSI firing on MSN activity has made attempts to define their role in network activity challenging. It is interesting to speculate that FSI neurons may play a similar subtle but significant modulatory role in area X. The FSI neurons that I recorded differed significantly between female-directed and undirected singing in their spiking statistics, including their ISI correlations, within-trial firing regularity, and between-trial spike precision. Such differences in FSI spiking could modulate firing patterns of downstream neurons, including MSNs (Fig. 1B). Whereas previous work indicates that the timing of the first spike in a burst is equally precise between female-directed and undirected singing (Woolley et al. 2014), one possibility is that FSI neurons could shape the timing of MSN spikes within a burst in songbirds, which would affect downstream activity.

GPe neurons in the zebra finch demonstrate context-dependent differences in firing rate, precision, and bursting, and it is likely that these changes affect the gain on downstream targets such as GPi neurons (Fig. 1B). For example, GPe neurons transition from regular, tonic spontaneous activity to irregular activity characterized by both long bursts of spikes as well as long pauses in firing during undirected singing. Such fluctuations in activity could generate variably timed windows of greater and lesser inhibition on GPi neurons during the production of undirected song. The long bursts could provide inhibition onto GPi neurons, perhaps making GPi neurons more susceptible to inhibitory input from MSNs. Meanwhile, the long pauses in GPe firing could render even highly correlated MSN input less effective at eliciting pauses in GPi neural activity. In contrast, the overall decrease in firing rate coupled with the precise firing of single inhibitory spikes during female-directed song could provide restricted access for MSN inputs, thereby enhancing pausing, but only within narrow and precise windows. Indeed, spike trains with negative ISI correlations, like those observed for GPe neurons at short lags during female-directed singing, have been found to display less variance than spike trains with independent ISIs (Chacron et al. 2001; Ratnam and Nelson 2000) and are hypothesized to increase information transfer, signal detection, and discriminability (Avila-Akerberg and Chacron 2011; Brandman and Nelson 2002; Chacron et al. 2001, 2004; Goense and Ratnam 2003; Ratnam and Nelson 2000). Whether negative ISI correlations in GPe neurons during female-directed singing have similar implications remains to be determined. However, taking these findings together, I hypothesize that the activity of GPe neurons during both female-directed and undirected singing serves to differentially modulate the influence of MSN inputs to GPi neurons to affect the variability and reliability of pauses and coordinate activity in a manner analogous to what is seen in mammalian systems (Gittis et al. 2014; Goldberg and Bergman 2011).

During undirected singing, there is both an increase in bursting, indicated by the CV, CV2, and autocorrelation analyses, and an increase in the trial-by-trial variability in spike precision in both FSI and GPe neurons. There are multiple ways that such changes in the variability could arise. For example, greater trial-by-trial variability could result from the addition of bursts with variable timing or duration onto the more precise firing of isolated spikes seen during female-directed singing. Alternatively, both the timing of isolated spikes and bursts could be more variable during undirected singing. By dividing spike trains into bursts and isolated spikes at a range of thresholds, I observed that for both FSI and GPe neurons, the timing of isolated spikes and of burst onsets were more variable during undirected than female-directed singing. Thus it appears that the increased spike timing variability is not simply due to the addition of bursts but reflects changes to the overall spiking pattern of GPe and FSI neurons.

Despite the increase in variability during undirected singing, both neuron types show remarkable coherence between the average firing pattern during female-directed and undirected song. Thus the variability appears to occur around a pattern of activity. However, how the greater variability in the spiking of GPe and FSI neurons arises within the circuit remains unclear. The changes in firing are rapid and transient, and are apparent during the first motifs of singing in each context, indicating that they are not a consequence of long-term changes to the network. However, whether the changes in variability result from an increase in the variability of inputs or as a result of the postsynaptic properties of the cell is unknown. Area X receives inputs from two cortical nuclei, LMAN and HVC (Fig. 1), but little is known about the specific inputs to either GPe or FSI neurons in songbirds. Previous work has shown that HVC neurons that project to area X are highly and equally precise during female-directed and undirected singing, which would make them an unlikely source of variability for FSI and GPe neurons. In contrast, neurons in the cortical nucleus LMAN, which also projects to area X, show greater bursting and trial-by-trial variability during undirected than female-directed singing. One possibility is that LMAN inputs may contribute to the activity seen in FSI and GPe neurons. In addition, area X is also a target of a number of neuromodulatory systems, including dopamine, norepinephrine, and serotonin, each of which has been shown to produce rapid changes in spike dynamics in other systems (Deemyad et al. 2013; Nicola et al. 2000). Ultimately, elucidating the detailed connectivity of area X circuitry will be critical to further model and investigate what drives the rapid alterations in spike statistics of GPe and FSI neurons as well as how these may modulate downstream targets.

GRANTS

This work was funded by National Sciences and Engineering Research Council Grant RGPIN402186 (to S. C. Woolley).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author.

AUTHOR CONTRIBUTIONS

S.C.W. conception and design of research; S.C.W. performed experiments; S.C.W. analyzed data; S.C.W. interpreted results of experiments; S.C.W. prepared figures; S.C.W. drafted manuscript; S.C.W. edited and revised manuscript; S.C.W. approved final version of manuscript.

ACKNOWLEDGMENTS

I thank Jon Sakata, Mimi Kao, Logan James, and two anonymous reviewers for thoughtful comments on the manuscript, Maurice Chacron for discussions on analysis, and Allison Doupe for her training and support.

REFERENCES

  1. Abeles M. Quantification, smoothing, and confidence limits for single-units' histograms. J Neurosci Methods 5: 317–325, 1982. [DOI] [PubMed] [Google Scholar]
  2. Ali F, Otchy TM, Pehlevan C, Fantana AL, Burak Y, Ölveczky BP. The basal ganglia is necessary for learning spectral, but not temporal, features of birdsong. Neuron 80: 494–506, 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Avila-Akerberg O, Chacron MJ. Nonrenewal spike train statistics: causes and functional consequences on neural coding. Exp Brain Res 210: 353–371, 2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bastian J, Nguyenkim J. Dendritic modulation of burst-like firing in sensory neurons. J Neurophysiol 85: 10–22, 2001. [DOI] [PubMed] [Google Scholar]
  5. Berke JD. Functional properties of striatal fast-spiking interneurons. Front Syst Neurosci 5: 45, 2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bottjer SW, Miesner EA, Arnold AP. Forebrain lesions disrupt development but not maintenance of song in passerine birds. Science 224: 901–903, 1984. [DOI] [PubMed] [Google Scholar]
  7. Brandman R, Nelson ME. A simple model of long-term spike train regularization. Neural Comput 14: 1575–1597, 2002. [DOI] [PubMed] [Google Scholar]
  8. Carrillo G, Doupe AJ. Is the songbird area X striatal, pallidal, or both? An anatomical study. J Comp Neurol 473: 415–437, 2004. [DOI] [PubMed] [Google Scholar]
  9. Chacron MJ, Longtin A, Maler L. Negative interspike interval correlations increase the neuronal capacity for encoding time-dependent stimuli. J Neurosci 21: 5328–5343, 2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Chacron MJ, Longtin A, Maler L. To burst or not to burst? J Comput Neurosci 17: 127–136, 2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Deemyad T, Metzen MG, Pan Y, Chacron MJ. Serotonin selectively enhances perception and sensory neural responses to stimuli generated by same-sex conspecifics. Proc Natl Acad Sci USA 110: 19609–19614, 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Doupe AJ, Perkel DJ, Reiner A, Stern EA. Birdbrains could teach basal ganglia research a new song. Trends Neurosci 28: 353–363, 2005. [DOI] [PubMed] [Google Scholar]
  13. Farries MA, Perkel DJ. A telencephalic nucleus essential for song learning contains neurons with physiological characteristics of both striatum and globus pallidus. J Neurosci 22: 3773–3787, 2002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gage GJ, Stoetzner CR, Wiltschko AB, Berke JD. Selective activation of striatal fast-spiking interneurons during choice execution. Neuron 67: 466–479, 2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Garenne A, Pasquereau B, Guthrie M, Bioulac B, Boraud T. Basal ganglia preferentially encode context dependent choice in a two-armed bandit task. Front Syst Neurosci 5: 23, 2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gdowski MJ, Miller LE, Parrish T, Nenonene EK, Houk JC. Context dependency in the globus pallidus internal segment during targeted arm movements. J Neurophysiol 85: 998–1004, 2001. [DOI] [PubMed] [Google Scholar]
  17. Gittis AH, Berke JD, Bevan MD, Chan CS, Mallet N, Morrow MM, Schmidt R. New roles for the external globus pallidus in basal ganglia circuits and behavior. J Neurosci 34: 15178–15183, 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Gittis AH, Leventhal DK, Fensterheim BA, Pettibone JR, Berke JD, Kreitzer AC. Selective inhibition of striatal fast-spiking interneurons causes dyskinesias. J Neurosci 31: 15727–15731, 2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Goense JB, Ratnam R. Continuous detection of weak sensory signals in afferent spike trains: the role of anti-correlated interspike intervals in detection performance. J Comp Physiol A Neuroethol Sens Neural Behav Physiol 189: 741–759, 2003. [DOI] [PubMed] [Google Scholar]
  20. Goldberg JH, Adler A, Bergman H, Fee MS. Singing-related neural activity distinguishes two putative pallidal cell types in the songbird basal ganglia: comparison to the primate internal and external pallidal segments. J Neurosci 30: 7088–7098, 2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Goldberg JA, Bergman H. Computational physiology of the neural networks of the primate globus pallidus: function and dysfunction. Neuroscience 198: 171–192, 2011. [DOI] [PubMed] [Google Scholar]
  22. Goldberg JH, Fee MS. Singing-related neural activity distinguishes four classes of putative striatal neurons in the songbird basal ganglia. J Neurophysiol 103: 2002–2014, 2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hessler NA, Doupe AJ. Social context modulates singing related neural activity in the songbird forebrain. Nat Neurosci 2: 209–211, 1999. [DOI] [PubMed] [Google Scholar]
  24. Holt GR, Softky WR, Koch C, Douglas RJ. Comparison of discharge variability in vitro and in vivo in cat visual cortex neurons. J Neurophysiol 75: 1806–1814, 1996. [DOI] [PubMed] [Google Scholar]
  25. Jarvis ED, Scharff C, Grossman MR, Ramos JA, Nottebohm F. For whom the bird sings: context-dependent gene expression. Neuron 21: 775–788, 1998. [DOI] [PubMed] [Google Scholar]
  26. Kao MH, Brainard MS. Lesions of an avian basal ganglia circuit prevent context-dependent changes to song variability. J Neurophysiol 96: 1441–1455, 2006. [DOI] [PubMed] [Google Scholar]
  27. Kao MH, Doupe AJ, Brainard MS. Contributions of an avian basal ganglia-forebrain circuit to real-time modulation of song. Nature 433: 638–643, 2005. [DOI] [PubMed] [Google Scholar]
  28. Kao MH, Wright BD, Doupe AJ. Neurons in a forebrain nucleus required for vocal plasticity rapidly switch between precise firing and variable bursting depending on social context. J Neurosci 28: 13232–13247, 2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kawagoe R, Takikawa Y, Hikosaka O. Reward-predicting activity of dopamine and caudate neurons–a possible mechanism of motivational control of saccadic eye movement. J Neurophysiol 91: 1013–1024, 2004. [DOI] [PubMed] [Google Scholar]
  30. Khosravi-Hashemi N, Chacron MJ. Bursts and isolated spikes code for opposite movement directions in midbrain electrosensory neurons. PLoS One 7: e40339, 2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Khosravi-Hashemi N, Fortune ES, Chacron MJ. Coding movement direction by burst firing in electrosensory neurons. J Neurophysiol 106: 1954–1968, 2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Kojima S, Kao MH, Doupe AJ. Task-related “cortical” bursting depends critically on basal ganglia input and is linked to vocal plasticity. Proc Natl Acad Sci USA 110: 4756–4761, 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Kumbhare D, Baron MS. A novel tri-component scheme for classifying neuronal discharge patterns. J Neurosci Methods 239: 148–161, 2015. [DOI] [PubMed] [Google Scholar]
  34. Lau T, Gage GJ, Berke JD, Zochowski M. Local dynamics of gap-junction-coupled interneuron networks. Phys Biol 7: 16015, 2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Lee IH, Assad JA. Putaminal activity for simple reactions or self-timed movements. J Neurophysiol 89: 2528–2537, 2003. [DOI] [PubMed] [Google Scholar]
  36. Lewicki MS. Bayesian modelling and classification of neural signals. Neural Comput 6: 1005–1030, 1994. [Google Scholar]
  37. Nicola SM, Surmeier J, Malenka RC. Dopaminergic modulation of neuronal excitability in the striatum and nucleus accumbens. Annu Rev Neurosci 23: 185–215, 2000. [DOI] [PubMed] [Google Scholar]
  38. Ölveczky BP, Andalman AS, Fee MS. Vocal experimentation in the juvenile songbird requires a basal ganglia circuit. PLoS Biol 3: e153, 2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Pfenning AR, Hara E, Whitney O, Rivas MV, Wang R, Roulhac PL, Howard JT, Wirthlin M, Lovell PV, Ganapathy G, Mouncastle J, Moseley MA, Thompson JW, Soderblom EJ, Iriki A, Kato M, Gilbert MT, Zhang G, Bakken T, Bongaarts A, Bernard A, Lein E, Mello CV, Hartemink AJ, Jarvis ED. Convergent transcriptional specializations in the brains of humans and song-learning birds. Science 346: 1256846, 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Pidoux M, Bollu T, Riccelli T, Goldberg JH. Origins of basal ganglia output signals in singing juvenile birds. J Neurophysiol 113: 843–855, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Rajan R, Doupe AJ. Behavioral and neural signatures of readiness to initiate a learned motor sequence. Curr Biol 23: 87–93, 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Ratnam R, Nelson ME. Nonrenewal statistics of electrosensory afferent spike trains: implications for the detection of weak sensory signals. J Neurosci 20: 6672–6683, 2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Reiner A, Laverghetta AV, Meade CA, Cuthbertson SL, Bottjer SW. An immunohistochemical and pathway tracing study of the striatopallidal organization of area X in the male zebra finch. J Comp Neurol 469: 239–261, 2004. [DOI] [PubMed] [Google Scholar]
  44. Scharff C, Nottebohm F. A comparative study of the behavioral deficits following lesions of various parts of the zebra finch song system: implications for vocal learning. J Neurosci 11: 2896–2913, 1991. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Tanaka M, Alvarado JS, Murugan M, Mooney R. Focal expression of mutant huntingtin in the songbird basal ganglia disrupts cortico-basal ganglia networks and vocal sequences. Proc Natl Acad Sci USA 113: E1720–E1727, 2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Warren TL, Tumer EC, Charlesworth JD, Brainard MS. Mechanisms and time course of vocal learning and consolidation in the adult songbird. J Neurophysiol 106: 1806–1821, 2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Woolley SC, Doupe AJ. Social context-induced song variation affects female behavior and gene expression. PLoS Biol 6: e62, 2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Woolley SC, Kao MH. Variability in action: Contributions of a songbird cortical-basal ganglia circuit to vocal motor learning and control. Neuroscience 296: 39–47, 2015. [DOI] [PubMed] [Google Scholar]
  49. Woolley SC, Rajan R, Joshua M, Doupe AJ. Emergence of context-dependent variability across a basal ganglia network. Neuron 82: 208–223, 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Zann R. The Zebra Finch: A Synthesis of Field and Laboratory Studies. New York: Oxford University Press, 1996. [Google Scholar]

Articles from Journal of Neurophysiology are provided here courtesy of American Physiological Society

RESOURCES