The opponent channel population code of sound location is an efficient representation of natural binaural sounds

Barn owls specialized sound-driven behavior: Lessons in optimal processing and coding by the auditory system

The barn owl, a nocturnal raptor with remarkably efficient prey-capturing abilities, has been one of the initial animal models used for research of brain mechanisms underlying sound localization. Some seminal findings made from their specialized sound localizing auditory system include discoveries of a midbrain map of auditory space, mechanisms towards spatial cue detection underlying sound-driven orienting behavior, and circuit level changes supporting development and experience-dependent plasticity. These findings have explained properties of vital hearing functions and inspired theories in spatial hearing that extend across diverse animal species, thereby cementing the barn owl's legacy as a powerful experimental system for elucidating fundamental brain mechanisms. This concise review will provide an overview of the insights from which the barn owl model system has exemplified the strength of investigating diversity and similarity of brain mechanisms across species. First, we discuss some of the key findings in the specialized system of the barn owl that elucidated brain mechanisms toward detection of auditory cues for spatial hearing. Then we examine how the barn owl has validated mathematical computations and theories underlying optimal hearing across species. And lastly, we conclude with how the barn owl has advanced investigations toward developmental and experience dependent plasticity in sound localization, as well as avenues for future research investigations towards bridging commonalities across species. Analogous to the informative power of Astrophysics for understanding nature through diverse exploration of planets, stars, and galaxies across the universe, miscellaneous research across different animal species pursues broad understanding of natural brain mechanisms and behavior.

Drug resistance prediction and resistance genes identification in Mycobacterium tuberculosis based on a hierarchical attentive neural network utilizing genome-wide variants

Prediction of antimicrobial resistance based on whole-genome sequencing data has attracted greater attention due to its rapidity and convenience. Numerous machine learning-based studies have used genetic variants to predict drug resistance in Mycobacterium tuberculosis (MTB), assuming that variants are homogeneous, and most of these studies, however, have ignored the essential correlation between variants and corresponding genes when encoding variants, and used a limited number of variants as prediction input. In this study, taking advantage of genome-wide variants for drug-resistance prediction and inspired by natural language processing, we summarize drug resistance prediction into document classification, in which variants are considered as words, mutated genes in an isolate as sentences, and an isolate as a document. We propose a novel hierarchical attentive neural network model (HANN) that helps discover drug resistance-related genes and variants and acquire more interpretable biological results. It captures the interaction among variants in a mutated gene as well as among mutated genes in an isolate. Our results show that for the four first-line drugs of isoniazid (INH), rifampicin (RIF), ethambutol (EMB) and pyrazinamide (PZA), the HANN achieves the optimal area under the ROC curve of 97.90, 99.05, 96.44 and 95.14% and the optimal sensitivity of 94.63, 96.31, 92.56 and 87.05%, respectively. In addition, without any domain knowledge, the model identifies drug resistance-related genes and variants consistent with those confirmed by previous studies, and more importantly, it discovers one more potential drug-resistance-related gene.

Interaural time difference tuning in the rat inferior colliculus is predictive of behavioral sensitivity

While a large body of literature has examined the encoding of binaural spatial cues in the auditory midbrain, studies that ask how quantitative measures of spatial tuning in midbrain neurons compare with an animal's psychoacoustic performance remain rare. Researchers have tried to explain deficits in spatial hearing in certain patient groups, such as binaural cochlear implant users, in terms of declines in apparent reductions in spatial tuning of midbrain neurons of animal models. However, the quality of spatial tuning can be quantified in many different ways, and in the absence of evidence that a given neural tuning measure correlates with psychoacoustic performance, the interpretation of such finding remains very tentative. Here, we characterize ITD tuning in the rat inferior colliculus (IC) to acoustic pulse train stimuli with varying envelopes and at varying rates, and explore whether quality of tuning correlates behavioral performance. We quantified both mutual information (MI) and neural d' as measures of ITD sensitivity. Neural d' values paralleled behavioral ones, declining with increasing click rates or when envelopes changed from rectangular to Hanning windows, and they correlated much better with behavioral performance than MI. Meanwhile, MI values were larger in an older, more experienced cohort of animals than in naive animals, but neural d' did not differ between cohorts. However, the results obtained with neural d' and MI were highly correlated when ITD values were coded simply as left or right ear leading, rather than specific ITD values. Thus, neural measures of lateralization ability (e.g. d' or left/right MI) appear to be highly predictive of psychoacoustic performance in a two-alternative forced choice task.

Ecological origins of perceptual grouping principles in the auditory system

Events and objects in the world must be inferred from sensory signals to support behavior. Because sensory measurements are temporally and spatially local, the estimation of an object or event can be viewed as the grouping of these measurements into representations of their common causes. Perceptual grouping is believed to reflect internalized regularities of the natural environment, yet grouping cues have traditionally been identified using informal observation and investigated using artificial stimuli. The relationship of grouping to natural signal statistics has thus remained unclear, and additional or alternative cues remain possible. Here, we develop a general methodology for relating grouping to natural sensory signals and apply it to derive auditory grouping cues from natural sounds. We first learned local spectrotemporal features from natural sounds and measured their co-occurrence statistics. We then learned a small set of stimulus properties that could predict the measured feature co-occurrences. The resulting cues included established grouping cues, such as harmonic frequency relationships and temporal coincidence, but also revealed previously unappreciated grouping principles. Human perceptual grouping was predicted by natural feature co-occurrence, with humans relying on the derived grouping cues in proportion to their informativity about co-occurrence in natural sounds. The results suggest that auditory grouping is adapted to natural stimulus statistics, show how these statistics can reveal previously unappreciated grouping phenomena, and provide a framework for studying grouping in natural signals.

Synthesis of Hemispheric ITD Tuning from the Readout of a Neural Map: Commonalities of Proposed Coding Schemes in Birds and Mammals

A major cue to infer sound direction is the difference in arrival time of the sound at the left and right ears, called interaural time difference (ITD). The neural coding of ITD and its similarity across species have been strongly debated. In the barn owl, an auditory specialist relying on sound localization to capture prey, ITDs within the physiological range determined by the head width are topographically represented at each frequency. The topographic representation suggests that sound direction may be inferred from the location of maximal neural activity within the map. Such topographical representation of ITD, however, is not evident in mammals. Instead, the preferred ITD of neurons in the mammalian brainstem often lies outside the physiological range and depends on the neuron's best frequency. Because of these disparities, it has been assumed that how spatial hearing is achieved in birds and mammals is fundamentally different. However, recent studies reveal ITD responses in the owl's forebrain and midbrain premotor area that are consistent with coding schemes proposed in mammals. Particularly, sound location in owls could be decoded from the relative firing rates of two broadly and inversely ITD-tuned channels. This evidence suggests that, at downstream stages, the code for ITD may not be qualitatively different across species. Thus, while experimental evidence continues to support the notion of differences in ITD representation across species and brain regions, the latest results indicate notable commonalities, suggesting that codes driving orienting behavior in mammals and birds may be comparable.

Cortical mechanisms of spatial hearing

Humans and other animals use spatial hearing to rapidly localize events in the environment. However, neural encoding of sound location is a complex process involving the computation and integration of multiple spatial cues that are not represented directly in the sensory organ (the cochlea). Our understanding of these mechanisms has increased enormously in the past few years. Current research is focused on the contribution of animal models for understanding human spatial audition, the effects of behavioural demands on neural sound location encoding, the emergence of a cue-independent location representation in the auditory cortex, and the relationship between single-source and concurrent location encoding in complex auditory scenes. Furthermore, computational modelling seeks to unravel how neural representations of sound source locations are derived from the complex binaural waveforms of real-life sounds. In this article, we review and integrate the latest insights from neurophysiological, neuroimaging and computational modelling studies of mammalian spatial hearing. We propose that the cortical representation of sound location emerges from recurrent processing taking place in a dynamic, adaptive network of early (primary) and higher-order (posterior-dorsal and dorsolateral prefrontal) auditory regions. This cortical network accommodates changing behavioural requirements and is especially relevant for processing the location of real-life, complex sounds and complex auditory scenes.

Cooperative population coding facilitates efficient sound-source separability by adaptation to input statistics

Our sensory environment changes constantly. Accordingly, neural systems continually adapt to the concurrent stimulus statistics to remain sensitive over a wide range of conditions. Such dynamic range adaptation (DRA) is assumed to increase both the effectiveness of the neuronal code and perceptual sensitivity. However, direct demonstrations of DRA-based efficient neuronal processing that also produces perceptual benefits are lacking. Here, we investigated the impact of DRA on spatial coding in the rodent brain and the perception of human listeners. Complex spatial stimulation with dynamically changing source locations elicited prominent DRA already on the initial spatial processing stage, the Lateral Superior Olive (LSO) of gerbils. Surprisingly, on the level of individual neurons, DRA diminished spatial tuning because of large response variability across trials. However, when considering single-trial population averages of multiple neurons, DRA enhanced the coding efficiency specifically for the concurrently most probable source locations. Intrinsic LSO population imaging of energy consumption combined with pharmacology revealed that a slow-acting LSO gain-control mechanism distributes activity across a group of neurons during DRA, thereby enhancing population coding efficiency. Strikingly, such “efficient cooperative coding” also improved neuronal source separability specifically for the locations that were most likely to occur. These location-specific enhancements in neuronal coding were paralleled by human listeners exhibiting a selective improvement in spatial resolution. We conclude that, contrary to canonical models of sensory encoding, the primary motive of early spatial processing is efficiency optimization of neural populations for enhanced source separability in the concurrent environment.

The efficient coding hypothesis suggests that sensory processing adapts to the stimulus statistics to maximize information while minimizing energetic costs. This study finds that an auditory spatial processing circuit distributes activity across neurons to enhance processing efficiency, focally improving spatial resolution both in neurons and in human listeners.

On the Sparse Structure of Natural Sounds and Natural Images: Similarities, Differences, and Implications for Neural Coding

Sparse coding models of natural images and sounds have been able to predict several response properties of neurons in the visual and auditory systems. While the success of these models suggests that the structure they capture is universal across domains to some degree, it is not yet clear which aspects of this structure are universal and which vary across sensory modalities. To address this, we fit complete and highly overcomplete sparse coding models to natural images and spectrograms of speech and report on differences in the statistics learned by these models. We find several types of sparse features in natural images, which all appear in similar, approximately Laplace distributions, whereas the many types of sparse features in speech exhibit a broad range of sparse distributions, many of which are highly asymmetric. Moreover, individual sparse coding units tend to exhibit higher lifetime sparseness for overcomplete models trained on images compared to those trained on speech. Conversely, population sparseness tends to be greater for these networks trained on speech compared with sparse coding models of natural images. To illustrate the relevance of these findings to neural coding, we studied how they impact a biologically plausible sparse coding network's representations in each sensory modality. In particular, a sparse coding network with synaptically local plasticity rules learns different sparse features from speech data than are found by more conventional sparse coding algorithms, but the learned features are qualitatively the same for these models when trained on natural images.

Efficient Neural Coding in Auditory and Speech Perception

Speech has long been recognized as 'special'. Here, we suggest that one of the reasons for speech being special is that our auditory system has evolved to encode it in an efficient, optimal way. The theory of efficient neural coding argues that our perceptual systems have evolved to encode environmental stimuli in the most efficient way. Mathematically, this can be achieved if the optimally efficient codes match the statistics of the signals they represent. Experimental evidence suggests that the auditory code is optimal in this mathematical sense: statistical properties of speech closely match response properties of the cochlea, the auditory nerve, and the auditory cortex. Even more interestingly, these results may be linked to phenomena in auditory and speech perception.

Adaptive coding for dynamic sensory inference

Behavior relies on the ability of sensory systems to infer properties of the environment from incoming stimuli. The accuracy of inference depends on the fidelity with which behaviorally relevant properties of stimuli are encoded in neural responses. High-fidelity encodings can be metabolically costly, but low-fidelity encodings can cause errors in inference. Here, we discuss general principles that underlie the tradeoff between encoding cost and inference error. We then derive adaptive encoding schemes that dynamically navigate this tradeoff. These optimal encodings tend to increase the fidelity of the neural representation following a change in the stimulus distribution, and reduce fidelity for stimuli that originate from a known distribution. We predict dynamical signatures of such encoding schemes and demonstrate how known phenomena, such as burst coding and firing rate adaptation, can be understood as hallmarks of optimal coding for accurate inference.

Learning Midlevel Auditory Codes from Natural Sound Statistics

Interaction with the world requires an organism to transform sensory signals into representations in which behaviorally meaningful properties of the environment are made explicit. These representations are derived through cascades of neuronal processing stages in which neurons at each stage recode the output of preceding stages. Explanations of sensory coding may thus involve understanding how low-level patterns are combined into more complex structures. To gain insight into such midlevel representations for sound, we designed a hierarchical generative model of natural sounds that learns combinations of spectrotemporal features from natural stimulus statistics. In the first layer, the model forms a sparse convolutional code of spectrograms using a dictionary of learned spectrotemporal kernels. To generalize from specific kernel activation patterns, the second layer encodes patterns of time-varying magnitude of multiple first-layer coefficients. When trained on corpora of speech and environmental sounds, some second-layer units learned to group similar spectrotemporal features. Others instantiate opponency between distinct sets of features. Such groupings might be instantiated by neurons in the auditory cortex, providing a hypothesis for midlevel neuronal computation.

Widespread and Opponent fMRI Signals Represent Sound Location in Macaque Auditory Cortex

In primates, posterior auditory cortical areas are thought to be part of a dorsal auditory pathway that processes spatial information. But how posterior (and other) auditory areas represent acoustic space remains a matter of debate. Here we provide new evidence based on functional magnetic resonance imaging (fMRI) of the macaque indicating that space is predominantly represented by a distributed hemifield code rather than by a local spatial topography. Hemifield tuning in cortical and subcortical regions emerges from an opponent hemispheric pattern of activation and deactivation that depends on the availability of interaural delay cues. Importantly, these opponent signals allow responses in posterior regions to segregate space similarly to a hemifield code representation. Taken together, our results reconcile seemingly contradictory views by showing that the representation of space follows closely a hemifield code and suggest that enhanced posterior-dorsal spatial specificity in primates might emerge from this form of coding.

Opponent Coding of Sound Location (Azimuth) in Planum Temporale is Robust to Sound-Level Variations

Coding of sound location in auditory cortex (AC) is only partially understood. Recent electrophysiological research suggests that neurons in mammalian auditory cortex are characterized by broad spatial tuning and a preference for the contralateral hemifield, that is, a nonuniform sampling of sound azimuth. Additionally, spatial selectivity decreases with increasing sound intensity. To accommodate these findings, it has been proposed that sound location is encoded by the integrated activity of neuronal populations with opposite hemifield tuning (“opponent channel model”). In this study, we investigated the validity of such a model in human AC with functional magnetic resonance imaging (fMRI) and a phase-encoding paradigm employing binaural stimuli recorded individually for each participant. In all subjects, we observed preferential fMRI responses to contralateral azimuth positions. Additionally, in most AC locations, spatial tuning was broad and not level invariant. We derived an opponent channel model of the fMRI responses by subtracting the activity of contralaterally tuned regions in bilateral planum temporale. This resulted in accurate decoding of sound azimuth location, which was unaffected by changes in sound level. Our data thus support opponent channel coding as a neural mechanism for representing acoustic azimuth in human AC.

Sound localization in a changing world

In natural environments, neural systems must be continuously updated to reflect changes in sensory inputs and behavioral goals. Recent studies of sound localization have shown that adaptation and learning involve multiple mechanisms that operate at different timescales and stages of processing, with other sensory and motor-related inputs playing a key role. We are only just beginning to understand, however, how these processes interact with one another to produce adaptive changes at the level of neuronal populations and behavior. Because there is no explicit map of auditory space in the cortex, studies of sound localization may also provide much broader insight into the plasticity of complex neural representations that are not topographically organized.