Synthesizing avian dreams.

During sleep, sporadically, it is possible to find neural patterns of activity in areas of the avian brain that are activated during the generation of the song. It has recently been found that in the vocal muscles of a sleeping bird, it is possible to detect activity patterns during these silent replays. In this work, we employ a dynamical systems model for song production in suboscine birds in order to translate the vocal muscles activity during sleep into synthetic songs. Besides allowing us to translate muscle activity into behavior, we argue that this approach poses the biomechanics as a unique window into the avian brain, with biophysical models as its probe.

The dynamics behind diversity in suboscine songs.

Vocal behavior plays a crucial evolutionary role. In the case of birds, song is critically important in courtship, male-male competition and other key behaviors linked to reproduction. However, under natural conditions, a variety of avian species live in close proximity and share an 'acoustic landscape'. Therefore, they need to be able to differentiate their calls or songs from those of other species and also from those of other individuals of the same species. To do this efficiently, birds display a remarkable diversity of sounds. For example, in the case of vocal learners, such as oscine passerines (i.e. songbirds), complex sequences and subtle acoustic effects are produced through the generation of complex neuromuscular instructions driving the vocal organ, which is remarkably conserved across approximately 4000 oscine species. By contrast, the majority of the sister clade of oscines, the suboscine passerines, are thought not to be vocal learners. Despite this, different suboscine species can generate a rich variety of songs and quite subtle acoustic effects. In the last few years, different suboscine species have been shown to possess morphological adaptations that allow them to produce a diversity of acoustic characteristics. Here, we briefly review the mechanisms of sound production in birds, before considering three suboscine species in more detail. The examples discussed in this Review, integrating biological experiments and biomechanical modeling using non-linear dynamical systems, illustrate how a morphological adaptation can produce complex acoustic properties without the need for complex neuromuscular control.

Neural oscillations are locked to birdsong rhythms in canaries.

How vocal communication signals are represented in the cortex is a major challenge for behavioural neuroscience. Beyond a descriptive code, it is relevant to unveil the dynamical mechanism responsible for the neural representation of auditory stimuli. In this work, we report evidence of synchronous neural activity in nucleus HVC, a telencephalic area of canaries (Serinus canaria), in response to auditory playback of the bird's own song. The rhythmic features of canary song allowed us to show that this large-scale synchronization was locked to defined features of the behaviour. We recorded neural activity in a brain region where sensorimotor integration occurs, showing the presence of well-defined oscillations in the local field potentials, which are locked to song rhythm. We also show a correspondence between local field potentials, multiunit activity and single unit activity within the same brain region. Overall, our results show that the rhythmic features of the vocal behaviour are represented in a telencephalic region of canaries.

Replay of innate vocal patterns during night sleep in suboscines.

Activation of forebrain circuitry during sleep has been variably characterized as 'pre- or replay' and has been linked to memory consolidation. The evolutionary origins of this mechanism, however, are unknown. Sleep activation of the sensorimotor pathways of learned birdsong is a particularly useful model system because the muscles controlling the vocal organ are activated, revealing syringeal activity patterns for direct comparison with those of daytime vocal activity. Here, we show that suboscine birds, which develop their species-typical songs innately without the elaborate forebrain-thalamic circuitry of the vocal learning taxa, also engage in replay during sleep. In two tyrannid species, the characteristic syringeal activation patterns of the song could also be identified during sleep. Similar to song-learning oscines, the burst structure was more variable during sleep than daytime song production. In kiskadees (Pitangus sulphuratus), a second vocalization, which is part of a multi-modal display, was also replayed during sleep along with one component of the visual display. These data show unambiguously that variable 'replay' of stereotyped vocal motor programmes is not restricted to programmes confined within forebrain circuitry. The proposed effects on vocal motor programme maintenance are, therefore, building on a pre-existing neural mechanism that predates the evolution of learned vocal motor behaviour.

Towards an integrated view of vocal development.

Vocal development is usually studied from the perspective of neuroscience. In this issue, Zhang and Ghazanfar propose a way in which body growth might condition the process. They study the vocalizations of marmoset infants with a wide range of techniques that include computational models and experiments that mimic growth reversal. Their results suggest that the qualitative changes that occur during development are rooted in the nonlinear interaction between the nervous system and the biomechanics involved in respiration. This work illustrates how an integrative approach enriches our understanding of behavior.

Syringeal EMGs and synthetic stimuli reveal a switch-like activation of the songbird's vocal motor program.

The coordination of complex vocal behaviors like human speech and oscine birdsong requires fine interactions between sensory and motor programs, the details of which are not completely understood. Here, we show that in sleeping male zebra finches (Taeniopygia guttata), the activity of the song system selectively evoked by playbacks of their own song can be detected in the syrinx. Electromyograms (EMGs) of a syringeal muscle show playback-evoked patterns strikingly similar to those recorded during song execution, with preferred activation instants within the song. Using this global and continuous readout, we studied the activation dynamics of the song system elicited by different auditory stimuli. We found that synthetic versions of the bird's song, rendered by a physical model of the avian phonation apparatus, evoked very similar responses, albeit with lower efficiency. Modifications of autogenous or synthetic songs reduce the response probability, but when present, the elicited activity patterns match execution patterns in shape and timing, indicating an all-or-nothing activation of the vocal motor program.

Birds breathe at an aerodynamic resonance.

We present a dynamical model for the avian respiratory system and report the measurement of its variables in normal breathing canaries (Serinus canaria). Fitting the parameters of the model, we are able to show that the birds in our study breathe at an aerodynamic resonance of their respiratory system. For different respiratory regimes, such as singing, where rapid respiratory gestures are used, the nonlinearities of the model lead to a shift in its resonances toward higher frequency values.

Unusual Avian Vocal Mechanism Facilitates Encoding of Body Size.

In this work we study the sound production mechanism of the raspy sounding song of the white-tipped plantcutter (Phytotoma rutila), a species with a most unusual vocalization. The biomechanics involved in the production of this song, and scaling arguments, allowed us to predict the precise way in which body size is encoded in its vocalizations. We tested this prediction through acoustic analysis of recorded songs, computational modeling of its unusual vocal strategy, and inspection of museum specimens captured across southeastern and south-central South America.

The structure of reconstructed flows in latent spaces.

Reconstructing the flow of a dynamical system from experimental data has been a key tool in the study of nonlinear problems. It allows one to discover the equations ruling the dynamics of a system as well as to quantify its complexity. In this work, we study the topology of the flow reconstructed by autoencoders, a dimensionality reduction method based on deep neural networks that has recently proved to be a very powerful tool for this task. We show that, although in many cases proper embeddings can be obtained with this method, it is not always the case that the topological structure of the flow is preserved.

Dynamical model for the neural activity of singing Serinus canaria.

Vocal production in songbirds is a key topic regarding the motor control of a complex, learned behavior. Birdsong is the result of the interaction between the activity of an intricate set of neural nuclei specifically dedicated to song production and learning (known as the "song system"), the respiratory system and the vocal organ. These systems interact and give rise to precise biomechanical motor gestures which result in song production. Telencephalic neural nuclei play a key role in the production of motor commands that drive the periphery, and while several attempts have been made to understand their coding strategy, difficulties arise when trying to understand neural activity in the frame of the song system as a whole. In this work, we report neural additive models embedded in an architecture compatible with the song system to provide a tool to reduce the dimensionality of the problem by considering the global activity of the units in each neural nucleus. This model is capable of generating outputs compatible with measurements of air sac pressure during song production in canaries (Serinus canaria). In this work, we show that the activity in a telencephalic nucleus required by the model to reproduce the observed respiratory gestures is compatible with electrophysiological recordings of single neuron activity in freely behaving animals.

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.

From electromyographic activity to frequency modulation in zebra finch song.

Behavior emerges from the interaction between the nervous system and peripheral devices. In the case of birdsong production, a delicate and fast control of several muscles is required to control the configuration of the syrinx (the avian vocal organ) and the respiratory system. In particular, the syringealis ventralis muscle is involved in the control of the tension of the vibrating labia and thus affects the frequency modulation of the sound. Nevertheless, the translation of the instructions (which are electrical in nature) into acoustical features is complex and involves nonlinear, dynamical processes. In this work, we present a model of the dynamics of the syringealis ventralis muscle and the labia, which allows calculating the frequency of the generated sound, using as input the electrical activity recorded in the muscle. In addition, the model provides a framework to interpret inter-syllabic activity and hints at the importance of the biomechanical dynamics in determining behavior.

Temperature manipulation of neuronal dynamics in a forebrain motor control nucleus.

Different neuronal types within brain motor areas contribute to the generation of complex motor behaviors. A widely studied songbird forebrain nucleus (HVC) has been recognized as fundamental in shaping the precise timing characteristics of birdsong. This is based, among other evidence, on the stretching and the "breaking" of song structure when HVC is cooled. However, little is known about the temperature effects that take place in its neurons. To address this, we investigated the dynamics of HVC both experimentally and computationally. We developed a technique where simultaneous electrophysiological recordings were performed during temperature manipulation of HVC. We recorded spontaneous activity and found three effects: widening of the spike shape, decrease of the firing rate and change in the interspike interval distribution. All these effects could be explained with a detailed conductance based model of all the neurons present in HVC. Temperature dependence of the ionic channel time constants explained the first effect, while the second was based in the changes of the maximal conductance using single synaptic excitatory inputs. The last phenomenon, only emerged after introducing a more realistic synaptic input to the inhibitory interneurons. Two timescales were present in the interspike distributions. The behavior of one timescale was reproduced with different input balances received form the excitatory neurons, whereas the other, which disappears with cooling, could not be found assuming poissonian synaptic inputs. Furthermore, the computational model shows that the bursting of the excitatory neurons arises naturally at normal brain temperature and that they have an intrinsic delay at low temperatures. The same effect occurs at single synapses, which may explain song stretching. These findings shed light on the temperature dependence of neuronal dynamics and present a comprehensive framework to study neuronal connectivity. This study, which is based on intrinsic neuronal characteristics, may help to understand emergent behavioral changes.

Gating related activity in a syringeal muscle allows the reconstruction of zebra finches songs.

Birdsong production involves the simultaneous and precise control of a set of muscles that change the configuration and dynamics of the vocal organ. Although it has been reported that each one of the different muscles is primarily involved in the control of one acoustic feature, recent advances have shown that they act synergistically to achieve the dynamical state necessary for phonation. In this work, we present a set of criteria that allow the extraction of gating-related information from the electromyographic activity of the syringealis ventralis muscle, a muscle that has been shown to be involved in frequency modulation. Using dynamical models of the muscle and syringeal dynamics, we obtain a full reconstruction of the zebra finch song using only the activity of this muscle.

Nonlinear dynamics in the study of birdsong.

Birdsong, a rich and complex behavior, is a stellar model to understand a variety of biological problems, from motor control to learning. It also enables us to study how behavior emerges when a nervous system, a biomechanical device and the environment interact. In this review, I will show that many questions in the field can benefit from the approach of nonlinear dynamics, and how birdsong can inspire new directions for research in dynamics.

Difference between the vocalizations of two sister species of pigeons explained in dynamical terms.

Vocal communication is an unique example, where the nonlinear nature of the periphery can give rise to complex sounds even when driven by simple neural instructions. In this work we studied the case of two close-related bird species, Patagioenas maculosa and Patagioenas picazuro, whose vocalizations differ only in the timbre. The temporal modulation of the fundamental frequency is similar in both cases, differing only in the existence of sidebands around the fundamental frequency in the P. maculosa. We tested the hypothesis that the qualitative difference between these vocalizations lies in the nonlinear nature of the syrinx. In particular, we propose that the roughness of maculosa's vocalizations is due to an asymmetry between the right and left vibratory membranes, whose nonlinear dynamics generate the sound. To test the hypothesis, we generated a biomechanical model for vocal production with an asymmetric parameter Q with which we can control the level of asymmetry between these membranes. Using this model we generated synthetic vocalizations with the principal acoustic features of both species. In addition, we confirmed the anatomical predictions by making post mortem inspection of the syrinxes, showing that the species with tonal song (picazuro) has a more symmetrical pair of membranes compared to maculosa.

Average activity of excitatory and inhibitory neural populations.

We develop an extension of the Ott-Antonsen method [E. Ott and T. M. Antonsen, Chaos 18(3), 037113 (2008)] that allows obtaining the mean activity (spiking rate) of a population of excitable units. By means of the Ott-Antonsen method, equations for the dynamics of the order parameters of coupled excitatory and inhibitory populations of excitable units are obtained, and their mean activities are computed. Two different excitable systems are studied: Adler units and theta neurons. The resulting bifurcation diagrams are compared with those obtained from studying the phenomenological Wilson-Cowan model in some regions of the parameter space. Compatible behaviors, as well as higher dimensional chaotic solutions, are observed. We study numerical simulations to further validate the equations.

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.

Average dynamics of a finite set of coupled phase oscillators.

We study the solutions of a dynamical system describing the average activity of an infinitely large set of driven coupled excitable units. We compared their topological organization with that reconstructed from the numerical integration of finite sets. In this way, we present a strategy to establish the pertinence of approximating the dynamics of finite sets of coupled nonlinear units by the dynamics of its infinitely large surrogate.

Response of wild songbirds to songs synthesized with a low-dimensional model.

In this work, we used a dynamical system derived from an avian vocal production model to generate synthetic songs that mimic the Zonotrichia capensis songs. We confirmed that these synthetic renditions elicited behavioral responses similar to those evoked by real songs in wild songbirds of the same species. Specifically, we observed an increase in the singing rate of individual birds when a playback device was introduced into their territories. The success of our approach instills confidence in the hypotheses underpinning the model and provides a valuable tool for investigating a wide range of biological questions.