Budgerigars have complex sleep structure similar to that of mammals.

Birds and mammals share specialized forms of sleep including slow wave sleep (SWS) and rapid eye movement sleep (REM), raising the question of why and how specialized sleep evolved. Extensive prior studies concluded that avian sleep lacked many features characteristic of mammalian sleep, and therefore that specialized sleep must have evolved independently in birds and mammals. This has been challenged by evidence of more complex sleep in multiple songbird species. To extend this analysis beyond songbirds, we examined a species of parrot, the sister taxon to songbirds. We implanted adult budgerigars (Melopsittacus undulatus) with electroencephalogram (EEG) and electrooculogram (EOG) electrodes to evaluate sleep architecture, and video monitored birds during sleep. Sleep was scored with manual and automated techniques, including automated detection of slow waves and eye movements. This can help define a new standard for how to score sleep in birds. Budgerigars exhibited consolidated sleep, a pattern also observed in songbirds, and many mammalian species, including humans. We found that REM constituted 26.5% of total sleep, comparable to humans and an order of magnitude greater than previously reported. Although we observed no spindles, we found a clear state of intermediate sleep (IS) similar to non-REM (NREM) stage 2. Across the night, SWS decreased and REM increased, as observed in mammals and songbirds. Slow wave activity (SWA) fluctuated with a 29-min ultradian rhythm, indicating a tendency to move systematically through sleep states as observed in other species with consolidated sleep. These results are at variance with numerous older sleep studies, including for budgerigars. Here, we demonstrated that lighting conditions used in the prior budgerigar study-and commonly used in older bird studies-dramatically disrupted budgerigar sleep structure, explaining the prior results. Thus, it is likely that more complex sleep has been overlooked in a broad range of bird species. The similarities in sleep architecture observed in mammals, songbirds, and now budgerigars, alongside recent work in reptiles and basal birds, provide support for the hypothesis that a common amniote ancestor possessed the precursors that gave rise to REM and SWS at one or more loci in the parallel evolution of sleep in higher vertebrates. We discuss this hypothesis in terms of the common plan of forebrain organization shared by reptiles, birds, and mammals.

Intrinsic plasticity and birdsong learning.

Although information processing and storage in the brain is thought to be primarily orchestrated by synaptic plasticity, other neural mechanisms such as intrinsic plasticity are available. While a number of recent studies have described the plasticity of intrinsic excitability in several types of neurons, the significance of non-synaptic mechanisms in memory and learning remains elusive. After reviewing plasticity of intrinsic excitation in relation to learning and homeostatic mechanisms, we focus on the intrinsic properties of a class of basal-ganglia projecting song system neurons in zebra finch, how these related to each bird's unique learned song, how these properties change over development, and how they are maintained dynamically to rapidly change in response to auditory feedback perturbations. We place these results in the broader theme of learning and changes in intrinsic properties, emphasizing the computational implications of this form of plasticity, which are distinct from synaptic plasticity. The results suggest that exploring reciprocal interactions between intrinsic and network properties will be a fruitful avenue for understanding mechanisms of birdsong learning.

Elemental gesture dynamics are encoded by song premotor cortical neurons.

Quantitative biomechanical models can identify control parameters that are used during movements, and movement parameters that are encoded by premotor neurons. We fit a mathematical dynamical systems model including subsyringeal pressure, syringeal biomechanics and upper-vocal-tract filtering to the songs of zebra finches. This reduces the dimensionality of singing dynamics, described as trajectories (motor 'gestures') in a space of syringeal pressure and tension. Here we assess model performance by characterizing the auditory response 'replay' of song premotor HVC neurons to the presentation of song variants in sleeping birds, and by examining HVC activity in singing birds. HVC projection neurons were excited and interneurons were suppressed within a few milliseconds of the extreme time points of the gesture trajectories. Thus, the HVC precisely encodes vocal motor output through activity at the times of extreme points of movement trajectories. We propose that the sequential activity of HVC neurons is used as a 'forward' model, representing the sequence of gestures in song to make predictions on expected behaviour and evaluate feedback.

Differential development of retroactive and proactive interference during post-learning wakefulness.

Newly encoded, labile memories are prone to disruption during post-learning wakefulness. Here we examine the contributions of retroactive and proactive interference to daytime forgetting on an auditory classification task in a songbird. While both types of interference impair performance, they do not develop concurrently. The retroactive interference of task-B on task-A developed during the learning of task-B, whereas the proactive interference of task-A on task-B emerged during subsequent waking retention. These different time courses indicate an asymmetry in the emergence of retroactive and proactive interference and suggest a mechanistic framework for how different types of interference between new memories develop.

A unifying view of synchronization for data assimilation in complex nonlinear networks.

Networks of nonlinear systems contain unknown parameters and dynamical degrees of freedom that may not be observable with existing instruments. From observable state variables, we want to estimate the connectivity of a model of such a network and determine the full state of the model at the termination of a temporal observation window during which measurements transfer information to a model of the network. The model state at the termination of a measurement window acts as an initial condition for predicting the future behavior of the network. This allows the validation (or invalidation) of the model as a representation of the dynamical processes producing the observations. Once the model has been tested against new data, it may be utilized as a predictor of responses to innovative stimuli or forcing. We describe a general framework for the tasks involved in the "inverse" problem of determining properties of a model built to represent measured output from physical, biological, or other processes when the measurements are noisy, the model has errors, and the state of the model is unknown when measurements begin. This framework is called statistical data assimilation and is the best one can do in estimating model properties through the use of the conditional probability distributions of the model state variables, conditioned on observations. There is a very broad arena of applications of the methods described. These include numerical weather prediction, properties of nonlinear electrical circuitry, and determining the biophysical properties of functional networks of neurons. Illustrative examples will be given of (1) estimating the connectivity among neurons with known dynamics in a network of unknown connectivity, and (2) estimating the biophysical properties of individual neurons in vitro taken from a functional network underlying vocalization in songbirds.

Nonlinear statistical data assimilation for HVC[Formula: see text] neurons in the avian song system.

With the goal of building a model of the HVC nucleus in the avian song system, we discuss in detail a model of HVC[Formula: see text] projection neurons comprised of a somatic compartment with fast Na[Formula: see text] and K[Formula: see text] currents and a dendritic compartment with slower Ca[Formula: see text] dynamics. We show this model qualitatively exhibits many observed electrophysiological behaviors. We then show in numerical procedures how one can design and analyze feasible laboratory experiments that allow the estimation of all of the many parameters and unmeasured dynamical variables, given observations of the somatic voltage [Formula: see text] alone. A key to this procedure is to initially estimate the slow dynamics associated with Ca, blocking the fast Na and K variations, and then with the Ca parameters fixed estimate the fast Na and K dynamics. This separation of time scales provides a numerically robust method for completing the full neuron model, and the efficacy of the method is tested by prediction when observations are complete. The simulation provides a framework for the slice preparation experiments and illustrates the use of data assimilation methods for the design of those experiments.

Automatic reconstruction of physiological gestures used in a model of birdsong production.

Highly coordinated learned behaviors are key to understanding neural processes integrating the body and the environment. Birdsong production is a widely studied example of such behavior in which numerous thoracic muscles control respiratory inspiration and expiration: the muscles of the syrinx control syringeal membrane tension, while upper vocal tract morphology controls resonances that modulate the vocal system output. All these muscles have to be coordinated in precise sequences to generate the elaborate vocalizations that characterize an individual's song. Previously we used a low-dimensional description of the biomechanics of birdsong production to investigate the associated neural codes, an approach that complements traditional spectrographic analysis. The prior study used algorithmic yet manual procedures to model singing behavior. In the present work, we present an automatic procedure to extract low-dimensional motor gestures that could predict vocal behavior. We recorded zebra finch songs and generated synthetic copies automatically, using a biomechanical model for the vocal apparatus and vocal tract. This dynamical model described song as a sequence of physiological parameters the birds control during singing. To validate this procedure, we recorded electrophysiological activity of the telencephalic nucleus HVC. HVC neurons were highly selective to the auditory presentation of the bird's own song (BOS) and gave similar selective responses to the automatically generated synthetic model of song (AUTO). Our results demonstrate meaningful dimensionality reduction in terms of physiological parameters that individual birds could actually control. Furthermore, this methodology can be extended to other vocal systems to study fine motor control.

Temporal and rate code analysis of responses to low-frequency components in the bird's own song by song system neurons.

Auditory feedback (AF) plays a critical role in vocal learning. Previous studies in songbirds suggest that low-frequency (<~1 kHz) components may be salient cues in AF. We explored this with auditory stimuli including the bird's own song (BOS) and BOS variants with increased relative power at low frequencies (LBOS). We recorded single units from BOS-selective neurons in two forebrain nuclei (HVC and Area X) in anesthetized zebra finches. Song-evoked responses were analyzed based on both rate (spike counts) and temporal coding of spike trains. The BOS and LBOS tended to evoke similar spike-count responses in substantially overlapping populations of neurons in both HVC and Area X. Analysis of spike patterns demonstrated temporal coding information that discriminated among the BOS and LBOS stimuli significantly better than spike counts in the majority of HVC (94 %) and Area X (85 %) neurons. HVC neurons contained more and a broader range of temporal coding information to discriminate among the stimuli than Area X neurons. These results are consistent with a role of spike timing in coding differences in the spectral components of BOS in HVC and Area X neurons.

Sleep and sensorimotor integration during early vocal learning in a songbird.

Behavioural studies widely implicate sleep in memory consolidation in the learning of a broad range of behaviours. During sleep, brain regions are reactivated, and specific patterns of neural activity are replayed, consistent with patterns observed in previous waking behaviour. Birdsong learning is a paradigmatic model system for skill learning. Song development in juvenile zebra finches (Taeniopygia guttata) is characterized by sleep-dependent circadian fluctuations in singing behaviour, with immediate post-sleep deterioration in song structure followed by recovery later in the day. In sleeping adult birds, spontaneous bursting activity of forebrain premotor neurons in the robust nucleus of the arcopallium (RA) carries information about daytime singing. Here we show that, in juvenile zebra finches, playback during the day of an adult 'tutor' song induced profound and tutor-song-specific changes in bursting activity of RA neurons during the following night of sleep. The night-time neuronal changes preceded tutor-song-induced changes in singing, first observed the following day. Interruption of auditory feedback greatly reduced sleep bursting and prevented the tutor-song-specific neuronal remodelling. Thus, night-time neuronal activity is shaped by the interaction of the song model (sensory template) and auditory feedback, with changes in night-time activity preceding the onset of practice associated with vocal learning. We hypothesize that night-time bursting induces adaptive changes in premotor networks during sleep as part of vocal learning. By this hypothesis, adaptive changes driven by replay of sensory information at night and by evaluation of sensory feedback during the day interact to produce the complex circadian patterns seen early in vocal development.

A mechanism for frequency modulation in songbirds shared with humans.

In most animals that vocalize, control of fundamental frequency is a key element for effective communication. In humans, subglottal pressure controls vocal intensity but also influences fundamental frequency during phonation. Given the underlying similarities in the biomechanical mechanisms of vocalization in humans and songbirds, songbirds offer an attractive opportunity to study frequency modulation by pressure. Here, we present a novel technique for dynamic control of subsyringeal pressure in zebra finches. By regulating the opening of a custom-built fast valve connected to the air sac system, we achieved partial or total silencing of specific syllables, and could modify syllabic acoustics through more complex manipulations of air sac pressure. We also observed that more nuanced pressure variations over a limited interval during production of a syllable concomitantly affected the frequency of that syllable segment. These results can be explained in terms of a mathematical model for phonation that incorporates a nonlinear description for the vocal source capable of generating the observed frequency modulations induced by pressure variations. We conclude that the observed interaction between pressure and frequency was a feature of the source, not a result of feedback control. Our results indicate that, beyond regulating phonation or its absence, regulation of pressure is important for control of fundamental frequencies of vocalizations. Thus, although there are separate brainstem pathways for syringeal and respiratory control of song production, both can affect airflow and frequency. We hypothesize that the control of pressure and frequency is combined holistically at higher levels of the vocalization pathways.

Dynamical estimation of neuron and network properties III: network analysis using neuron spike times.

Estimating the behavior of a network of neurons requires accurate models of the individual neurons along with accurate characterizations of the connections among them. Whereas for a single cell, measurements of the intracellular voltage are technically feasible and sufficient to characterize a useful model of its behavior, making sufficient numbers of simultaneous intracellular measurements to characterize even small networks is infeasible. This paper builds on prior work on single neurons to explore whether knowledge of the time of spiking of neurons in a network, once the nodes (neurons) have been characterized biophysically, can provide enough information to usefully constrain the functional architecture of the network: the existence of synaptic links among neurons and their strength. Using standardized voltage and synaptic gating variable waveforms associated with a spike, we demonstrate that the functional architecture of a small network of model neurons can be established.

Estimating parameters and predicting membrane voltages with conductance-based neuron models.

Recent results demonstrate techniques for fully quantitative, statistical inference of the dynamics of individual neurons under the Hodgkin-Huxley framework of voltage-gated conductances. Using a variational approximation, this approach has been successfully applied to simulated data from model neurons. Here, we use this method to analyze a population of real neurons recorded in a slice preparation of the zebra finch forebrain nucleus HVC. Our results demonstrate that using only 1,500 ms of voltage recorded while injecting a complex current waveform, we can estimate the values of 12 state variables and 72 parameters in a dynamical model, such that the model accurately predicts the responses of the neuron to novel injected currents. A less complex model produced consistently worse predictions, indicating that the additional currents contribute significantly to the dynamics of these neurons. Preliminary results indicate some differences in the channel complement of the models for different classes of HVC neurons, which accords with expectations from the biology. Whereas the model for each cell is incomplete (representing only the somatic compartment, and likely to be missing classes of channels that the real neurons possess), our approach opens the possibility to investigate in modeling the plausibility of additional classes of channels the cell might possess, thus improving the models over time. These results provide an important foundational basis for building biologically realistic network models, such as the one in HVC that contributes to the process of song production and developmental vocal learning in songbirds.

Bursts from the past: Intrinsic properties link a network model to zebra finch song.

Neuronal intrinsic excitability is a mechanism implicated in learning and memory that is distinct from synaptic plasticity. Prior work in songbirds established that intrinsic properties (IPs) of premotor basal-ganglia-projecting neurons (HVC X ) relate to learned song. Here we find that temporal song structure is related to specific HVC X IPs: HVC X from birds who sang longer songs including longer invariant vocalizations (harmonic stacks) had IPs that reflected increased post-inhibitory rebound. This suggests a rebound excitation mechanism underlying the ability of HVC X neurons to integrate over long periods of time and represent sequence information. To explore this, we constructed a network model of realistic neurons showing how in-vivo HVC bursting properties link rebound excitation to network structure and behavior. These results demonstrate an explicit link between neuronal IPs and learned behavior. We propose that sequential behaviors exhibiting temporal regularity require IPs to be included in realistic network-level descriptions.

Emergence of selectivity and tolerance in the avian auditory cortex.

The ability to recognize auditory objects like words and bird songs is thought to depend on neural responses that are selective between categories of the objects and tolerant of variation within those categories. To determine whether a hierarchy of increasing selectivity and tolerance exists in the avian auditory system, we trained European starlings (Sturnus vulgaris) to differentially recognize sets of songs, then measured extracellular single unit responses under urethane anesthesia in six areas of the auditory cortex. Responses were analyzed with a novel, generalized linear mixed model that provides robust estimates of the variance in responses to different stimuli. There were significant differences between areas in selectivity, tolerance, and the effects of training. The L2b and L1 subdivisions of field L had the least selectivity and tolerance. The caudal nidopallium (NCM) and subdivision L3 of field L were more selective than other areas, whereas the medial and lateral caudal mesopallium were more tolerant than NCM or L2b. L3 had a multimodal distribution of tolerance. Sensitivity to songs that were familiar and those that were not also distinguished the responses of caudomedial mesopallium and NCM. There were significant differences across areas between neurons with wide and narrow spikes. Collectively these results do not fit the traditional hierarchical view of the avian auditory forebrain, but are consistent with emerging concepts homologizing avian cortical and neocortical circuitry. The results suggest a functional divergence within the cortex into processing streams that respond to complementary aspects of the variability in communicative sounds.

Sleep consolidation of interfering auditory memories in starlings.

Memory consolidation has been described as a process to strengthen newly formed memories and to stabilize them against interference from similar learning experiences. Sleep facilitates memory consolidation in humans, improving memory performance and protecting against interference encountered after sleep. The European starling, a songbird, has also manifested sleep-dependent memory consolidation when trained on an auditory-classification task. Here, we examined how memory for two similar classification tasks is consolidated across waking and sleep in starlings. We demonstrated for the first time that the learning of each classification reliably interferes with the retention of the other classification across waking retention but that sleep enhances and stabilizes the memory of both classifications even after performance is impaired by interference. These observations demonstrate that sleep consolidation enhances retention of interfering experiences, facilitating opportunistic daytime learning and the subsequent formation of stable long-term memories.

Rhythmically bursting songbird vocomotor neurons are organized into multiple sequences, suggesting a network/intrinsic properties model encoding song and error, not time.

In zebra finch, basal ganglia projecting "HVC X " neurons emit one or more spike bursts during each song motif (canonical sequence of syllables), which are thought to be driven in part by a process of spike rebound excitation. Zebra finch songs are highly stereotyped and recent results indicate that the intrinsic properties of HVC X neurons are similar within each bird, vary among birds depending on similarity of the songs, and vary with song errors. We tested the hypothesis that the timing of spike bursts during singing also evince individual-specific distributions. Examining previously published data, we demonstrated that the intervals between bursts of multibursting HVC X are similar for neurons within each bird, in many cases highly clustered at distinct peaks, with the patterns varying among birds. The fixed delay between bursts and different times when neurons are first recruited in the song yields precisely timed multiple sequences of bursts throughout the song, not the previously envisioned single sequence of bursts treated as events having statistically independent timing. A given moment in time engages multiple sequences and both single bursting and multibursting HVC X simultaneously. This suggests a model where a population of HVC X sharing common intrinsic properties driving spike rebound excitation influence the timing of a given HVC X burst through lateral inhibitory interactions. Perturbations in burst timing, representing error, could propagate in time. Our results extend the concept of central pattern generators to complex vertebrate vocal learning and suggest that network activity (timing of inhibition) and HVC X intrinsic properties become coordinated during developmental birdsong learning.

Song tutoring in presinging zebra finch juveniles biases a small population of higher-order song-selective neurons toward the tutor song.

We explored physiological changes correlated with song tutoring by recording the responses of caudal nidopallium neurons of zebra finches aged P21-P24 (days post hatching) to a broad spectrum of natural and synthetic stimuli. Those birds raised with their fathers tended to show behavioral evidence of song memorization but not of singing; thus auditory responses were not confounded by the birds' own vocalizations. In study 1, 37 of 158 neurons (23%) in 17 of 22 tutored and untutored birds were selective for only 1 of 10 stimuli comprising broadband signals, early juvenile songs and calls, female calls, and adult songs. Approximately 30% of the selective neurons (12/37 neurons in 9 birds) were selective for adult conspecific songs. All these were found in the song system nuclei HVC and paraHVC. Of 122 neurons (17 birds) in tutored birds, all of the conspecific song-selective neurons (8 neurons in 6 birds) were selective for the adult tutor song; none was selective for unfamiliar song. In study 2 with a different sampling strategy, we found that 11 of 12 song-selective neurons in 6 of 7 birds preferred the tutor song; none preferred unfamiliar or familiar conspecific songs. Most of these neurons were found in caudal lateral nidopallium (NCL) below HVC. Thus by the time a bird begins to sing, there are small numbers of tutor song-selective neurons distributed in several forebrain regions. We hypothesize that a small population of higher-order auditory neurons is innately selective for complex features of behaviorally relevant stimuli and these responses are modified by specific perceptual/social experience during development.

Dynamical estimation of neuron and network properties II: Path integral Monte Carlo methods.

Hodgkin-Huxley (HH) models of neuronal membrane dynamics consist of a set of nonlinear differential equations that describe the time-varying conductance of various ion channels. Using observations of voltage alone we show how to estimate the unknown parameters and unobserved state variables of an HH model in the expected circumstance that the measurements are noisy, the model has errors, and the state of the neuron is not known when observations commence. The joint probability distribution of the observed membrane voltage and the unobserved state variables and parameters of these models is a path integral through the model state space. The solution to this integral allows estimation of the parameters and thus a characterization of many biological properties of interest, including channel complement and density, that give rise to a neuron's electrophysiological behavior. This paper describes a method for directly evaluating the path integral using a Monte Carlo numerical approach. This provides estimates not only of the expected values of model parameters but also of their posterior uncertainty. Using test data simulated from neuronal models comprising several common channels, we show that short (<50 ms) intracellular recordings from neurons stimulated with a complex time-varying current yield accurate and precise estimates of the model parameters as well as accurate predictions of the future behavior of the neuron. We also show that this method is robust to errors in model specification, supporting model development for biological preparations in which the channel expression and other biophysical properties of the neurons are not fully known.

Recursive syntactic pattern learning by songbirds.

Humans regularly produce new utterances that are understood by other members of the same language community. Linguistic theories account for this ability through the use of syntactic rules (or generative grammars) that describe the acceptable structure of utterances. The recursive, hierarchical embedding of language units (for example, words or phrases within shorter sentences) that is part of the ability to construct new utterances minimally requires a 'context-free' grammar that is more complex than the 'finite-state' grammars thought sufficient to specify the structure of all non-human communication signals. Recent hypotheses make the central claim that the capacity for syntactic recursion forms the computational core of a uniquely human language faculty. Here we show that European starlings (Sturnus vulgaris) accurately recognize acoustic patterns defined by a recursive, self-embedding, context-free grammar. They are also able to classify new patterns defined by the grammar and reliably exclude agrammatical patterns. Thus, the capacity to classify sequences from recursive, centre-embedded grammars is not uniquely human. This finding opens a new range of complex syntactic processing mechanisms to physiological investigation.