A revised model of the inner-hair cell and auditory-nerve complex

An Adaptive Leaky-Integrate and Firing Probability Model of an Electrically Stimulated Auditory Nerve Fiber

Most neural models produce a spiking output and often represent the stochastic nature of the spike generation process via a stochastic output. Nonspiking neural models, on the other hand, predict the probability of a spike occurring in response to a stimulus. We propose a nonspiking model for an electrically stimulated auditory nerve fiber, which not only predicts the total probability of a spike occurring in response to a biphasic pulse but also the distribution of the spike time. Our adaptive leaky-integrate and firing probability (aLIFP) model can account for refractoriness, facilitation, accommodation, and long-term adaptation. All model parameters have been fitted to single cell recordings from electrically stimulated cat auditory nerve fibers. Afterward, the model was validated on recordings from auditory nerve fibers from cats and guinea pigs. The nonspiking nature of the model makes it fast and deterministic while still accounting for the stochastic nature of the spike generation process. Therefore, the relationship between the input to the model or model parameters and the model's output can be observed more directly than with stochastically spiking models.

From the outer ear to the nerve: A complete computer model of the peripheral auditory system

Computer models of the individual components of the peripheral auditory system - the outer, middle, and inner ears and the auditory nerve - have been developed in the past, with varying level of detail, breadth, and faithfulness of the underlying parameters. Building on previous work, we advance the modeling of the ear by presenting a complete, physiologically justified, bottom-up computer model based on up-to-date experimental data that integrates all of these parts together seamlessly. The detailed bottom-up design of the present model allows for the investigation of partial hearing mechanisms and their defects, including genetic, molecular, and microscopic factors. Also, thanks to the completeness of the model, one can study microscopic effects in the context of their implications on hearing as a whole, enabling the correlation with neural recordings and non-invasive psychoacoustic methods. Such a model is instrumental for advancing quantitative understanding of the mechanism of hearing, for investigating various forms of hearing impairment, as well as for devising next generation hearing aids and cochlear implants.

A Fundamental Inequality Governing the Rate Coding Response of Sensory Neurons

A fundamental inequality governing the spike activity of peripheral neurons is derived and tested against auditory data. This inequality states that the steady-state firing rate must lie between the arithmetic and geometric means of the spontaneous and peak activities during adaptation. Implications towards the development of auditory mechanistic models are explored.

Modeling temporal information encoding by the population of fibers in the healthy and synaptopathic auditory nerve

We report a theoretical study aimed at investigating the impact of cochlear synapse loss (synaptopathy) on the encoding of the envelope (ENV) and temporal fine structure (TFS) of sounds by the population of auditory nerve fibers. A computational model was used to simulate auditory-nerve spike trains evoked by sinusoidally amplitude-modulated (AM) tones at 10 Hz with various carrier frequencies and levels. The model included 16 cochlear channels with characteristic frequencies (CFs) from 250 Hz to 8 kHz. Each channel was innervated by 3, 4 and 10 fibers with low (LSR), medium (MSR), and high spontaneous rates (HSR), respectively. For each channel, spike trains were collapsed into three separate 'population' post-stimulus time histograms (PSTHs), one per fiber type. Information theory was applied to reconstruct the stimulus waveform, ENV, and TFS from one or more PSTHs in a mathematically optimal way. The quality of the reconstruction was regarded as an estimate of the information present in the used PSTHs. Various synaptopathy scenarios were simulated by removing fibers of specific types and/or cochlear regions before stimulus reconstruction. We found that the TFS was predominantly encoded by HSR fibers at all stimulus carrier frequencies and levels. The encoding of the ENV was more complex. At lower levels, the ENV was predominantly encoded by HSR fibers with CFs near the stimulus carrier frequency. At higher levels, the ENV was equally well or better encoded by HSR fibers with CFs different from the AM carrier frequency as by LSR fibers with CFs at the carrier frequency. Altogether, findings suggest that a healthy population of HSR fibers (i.e., including fibers with CFs around and remote from the AM carrier frequency) might be sufficient to encode the ENV and TFS over a wide range of stimulus levels. Findings are discussed regarding their relevance for diagnosing synaptopathy using non-invasive ENV- and TFS-based measures.

Intrinsic Noise Improves Speech Recognition in a Computational Model of the Auditory Pathway

Noise is generally considered to harm information processing performance. However, in the context of stochastic resonance, noise has been shown to improve signal detection of weak sub- threshold signals, and it has been proposed that the brain might actively exploit this phenomenon. Especially within the auditory system, recent studies suggest that intrinsic noise plays a key role in signal processing and might even correspond to increased spontaneous neuronal firing rates observed in early processing stages of the auditory brain stem and cortex after hearing loss. Here we present a computational model of the auditory pathway based on a deep neural network, trained on speech recognition. We simulate different levels of hearing loss and investigate the effect of intrinsic noise. Remarkably, speech recognition after hearing loss actually improves with additional intrinsic noise. This surprising result indicates that intrinsic noise might not only play a crucial role in human auditory processing, but might even be beneficial for contemporary machine learning approaches.

Binaural Processing Deficits Due to Synaptopathy and Myelin Defects

Hidden hearing loss (HHL) is a deficit in auditory perception and speech intelligibility that occurs despite normal audiometric thresholds and results from noise exposure, aging, or myelin defects. While mechanisms causing perceptual deficits in HHL patients are still unknown, results from animal models indicate a role for peripheral auditory neuropathies in HHL. In humans, sound localization is particularly important for comprehending speech, especially in noisy environments, and its disruption may contribute to HHL. In this study, we hypothesized that neuropathies of cochlear spiral ganglion neurons (SGNs) that are observed in animal models of HHL disrupt the activity of neurons in the medial superior olive (MSO), a nucleus in the brainstem responsible for locating low-frequency sound in the horizontal plane using binaural temporal cues, leading to sound localization deficits. To test our hypothesis, we constructed a network model of the auditory processing system that simulates peripheral responses to sound stimuli and propagation of responses via SGNs to cochlear nuclei and MSO populations. To simulate peripheral auditory neuropathies, we used a previously developed biophysical SGN model with myelin defects at SGN heminodes (myelinopathy) and with loss of inner hair cell-SGN synapses (synaptopathy). Model results indicate that myelinopathy and synaptopathy in SGNs give rise to decreased interaural time difference (ITD) sensitivity of MSO cells, suggesting a possible mechanism for perceptual deficits in HHL patients. This model may be useful to understand downstream impacts of SGN-mediated disruptions on auditory processing and to eventually discover possible treatments for various mechanisms of HHL.

Region-dependent Millisecond Time-scale Sensitivity in Spectrotemporal Integrations in Guinea Pig Primary Auditory Cortex

Spectrotemporal integration is a key function of our auditory system for discriminating spectrotemporally complex sounds, such as words. Response latency in the auditory cortex is known to change with the millisecond time-scale depending on acoustic parameters, such as sound frequency and intensity. The functional significance of the millisecond-range latency difference in the integration remains unclear. Actually, whether the auditory cortex has a sensitivity to the millisecond-range difference has not been systematically examined. Herein, we examined the sensitivity in the primary auditory cortex (A1) using voltage-sensitive dye imaging techniques in guinea pigs. Bandpass noise bursts in two different bands (band-noises), centered at 1 and 16 kHz, respectively, were used for the examination. Onset times of individual band-noises (spectral onset-times) were varied to virtually cancel or magnify the latency difference observed with the band-noises. Conventionally defined nonlinear effects in integration were analyzed at A1 with varying sound intensities (or response latencies) and/or spectral onset-times of the two band-noises. The nonlinear effect measured in the high-frequency region of the A1 linearly changed depending on the millisecond difference of the response onset-times, which were estimated from the spatially-local response latencies and spectral onset-times. In contrast, the low-frequency region of the A1 had no significant sensitivity to the millisecond difference. The millisecond-range latency difference may have functional significance in the spectrotemporal integration with the millisecond time-scale sensitivity at the high-frequency region of A1 but not at the low-frequency region.

Understanding degraded speech leads to perceptual gating of a brainstem reflex in human listeners

The ability to navigate "cocktail party" situations by focusing on sounds of interest over irrelevant, background sounds is often considered in terms of cortical mechanisms. However, subcortical circuits such as the pathway underlying the medial olivocochlear (MOC) reflex modulate the activity of the inner ear itself, supporting the extraction of salient features from auditory scene prior to any cortical processing. To understand the contribution of auditory subcortical nuclei and the cochlea in complex listening tasks, we made physiological recordings along the auditory pathway while listeners engaged in detecting non(sense) words in lists of words. Both naturally spoken and intrinsically noisy, vocoded speech-filtering that mimics processing by a cochlear implant (CI)-significantly activated the MOC reflex, but this was not the case for speech in background noise, which more engaged midbrain and cortical resources. A model of the initial stages of auditory processing reproduced specific effects of each form of speech degradation, providing a rationale for goal-directed gating of the MOC reflex based on enhancing the representation of the energy envelope of the acoustic waveform. Our data reveal the coexistence of 2 strategies in the auditory system that may facilitate speech understanding in situations where the signal is either intrinsically degraded or masked by extrinsic acoustic energy. Whereas intrinsically degraded streams recruit the MOC reflex to improve representation of speech cues peripherally, extrinsically masked streams rely more on higher auditory centres to denoise signals.

A convolutional neural-network framework for modelling auditory sensory cells and synapses

In classical computational neuroscience, analytical model descriptions are derived from neuronal recordings to mimic the underlying biological system. These neuronal models are typically slow to compute and cannot be integrated within large-scale neuronal simulation frameworks. We present a hybrid, machine-learning and computational-neuroscience approach that transforms analytical models of sensory neurons and synapses into deep-neural-network (DNN) neuronal units with the same biophysical properties. Our DNN-model architecture comprises parallel and differentiable equations that can be used for backpropagation in neuro-engineering applications, and offers a simulation run-time improvement factor of 70 and 280 on CPU or GPU systems respectively. We focussed our development on auditory neurons and synapses, and show that our DNN-model architecture can be extended to a variety of existing analytical models. We describe how our approach for auditory models can be applied to other neuron and synapse types to help accelerate the development of large-scale brain networks and DNN-based treatments of the pathological system.

Drakopoulos et al developed a machine-learning and computational-neuroscience approach that transforms analytical models of sensory neurons and synapses into deep-neural-network (DNN) neuronal units with the same biophysical properties. Focusing on auditory neurons and synapses, they showed that their DNN-model architecture could be extended to a variety of existing analytical models and to other neuron and synapse types, thus potentially assisting the development of large-scale brain networks and DNN-based treatments.

A simplified physiological model of rate-level functions of auditory-nerve fibers

Several approaches have been used to describe the rate-level functions of auditory-nerve fibers (ANFs). One approach uses descriptive models that can be fitted easily to data. Another derives rate-level functions from comprehensive physiological models of auditory peripheral processing. Here, we seek to identify the minimal set of components needed to provide a physiologically plausible account of rate-level functions. Our model consists of a first-order Boltzmann mechanoelectrical transducer function relating the instantaneous stimulus pressure to an instantaneous output, followed by a lowpass filter that eliminates the AC component, followed by an exponential synaptic transfer function relating the DC component to the mean spike rate. This is perhaps the simplest physiologically plausible model capable of accounting for rate-level functions under the assumption that the model parameters for a given ANF and stimulus frequency are level-independent. We find that the model typically accounts well for rate-level functions from cat ANFs for all stimulus frequencies. More complicated model variants having saturating synaptic transfer functions do not perform significantly better, implying the system operates far away from synaptic saturation. Rate saturation in the model is caused by saturation of the DC component of the filter output (e.g., the receptor potential), which in turn is due to the saturation of the transducer function. The maximum mean spike rate is approximately constant across ANFs, such that the slope parameter of the exponential synaptic transfer function decreases with increasing spontaneous rate. If the synaptic parameters for a given ANF are assumed to be constant across stimulus frequencies, then frequency- and level-dependent input nonlinearities are derived that are qualitatively similar to those reported in the literature. Contrary to assumptions in the literature, such nonlinearities are obtained even for ANFs having high spontaneous rates. Finally, spike-rate adaptation is examined and found to be accounted for by a decrease in the slope parameter of the synaptic transfer function over time following stimulus onset.

A numerical method for computing interval distributions for an inhomogeneous Poisson point process modified by random dead times

The inhomogeneous Poisson point process is a common model for time series of discrete, stochastic events. When an event from a point process is detected, it may trigger a random dead time in the detector, during which subsequent events will fail to be detected. It can be difficult or impossible to obtain a closed-form expression for the distribution of intervals between detections, even when the rate function (often referred to as the intensity function) and the dead-time distribution are given. Here, a method is presented to numerically compute the interval distribution expected for any arbitrary inhomogeneous Poisson point process modified by dead times drawn from any arbitrary distribution. In neuroscience, such a point process is used to model trains of neuronal spikes triggered by the detection of excitatory events while the neuron is not refractory. The assumptions of the method are that the process is observed over a finite observation window and that the detector is not in a dead state at the start of the observation window. Simulations are used to verify the method for several example point processes. The method should be useful for modeling and understanding the relationships between the rate functions and interval distributions of the event and detection processes, and how these relationships depend on the dead-time distribution.

Contrasting mechanisms for hidden hearing loss: Synaptopathy vs myelin defects

Hidden hearing loss (HHL) is an auditory neuropathy characterized by normal hearing thresholds but reduced amplitudes of the sound-evoked auditory nerve compound action potential (CAP). In animal models, HHL can be caused by moderate noise exposure or aging, which induces loss of inner hair cell (IHC) synapses. In contrast, recent evidence has shown that transient loss of cochlear Schwann cells also causes permanent auditory deficits in mice with similarities to HHL. Histological analysis of the cochlea after auditory nerve remyelination showed a permanent disruption of the myelination patterns at the heminode of type I spiral ganglion neuron (SGN) peripheral terminals, suggesting that this defect could be contributing to HHL. To shed light on the mechanisms of different HHL scenarios observed in animals and to test their impact on type I SGN activity, we constructed a reduced biophysical model for a population of SGN peripheral axons whose activity is driven by a well-accepted model of cochlear sound processing. We found that the amplitudes of simulated sound-evoked SGN CAPs are lower and have greater latencies when heminodes are disorganized, i.e. they occur at different distances from the hair cell rather than at the same distance as in the normal cochlea. These results confirm that disruption of heminode positions causes desynchronization of SGN spikes leading to a loss of temporal resolution and reduction of the sound-evoked SGN CAP. Another mechanism resulting in HHL is loss of IHC synapses, i.e., synaptopathy. For comparison, we simulated synaptopathy by removing high threshold IHC-SGN synapses and found that the amplitude of simulated sound-evoked SGN CAPs decreases while latencies remain unchanged, as has been observed in noise exposed animals. Thus, model results illuminate diverse disruptions caused by synaptopathy and demyelination on neural activity in auditory processing that contribute to HHL as observed in animal models and that can contribute to perceptual deficits induced by nerve damage in humans.

Simple transformations capture auditory input to cortex

Sounds are processed by the ear and central auditory pathway. These processing steps are biologically complex, and many aspects of the transformation from sound waveforms to cortical response remain unclear. To understand this transformation, we combined models of the auditory periphery with various encoding models to predict auditory cortical responses to natural sounds. The cochlear models ranged from detailed biophysical simulations of the cochlea and auditory nerve to simple spectrogram-like approximations of the information processing in these structures. For three different stimulus sets, we tested the capacity of these models to predict the time course of single-unit neural responses recorded in ferret primary auditory cortex. We found that simple models based on a log-spaced spectrogram with approximately logarithmic compression perform similarly to the best-performing biophysically detailed models of the auditory periphery, and more consistently well over diverse natural and synthetic sounds. Furthermore, we demonstrated that including approximations of the three categories of auditory nerve fiber in these simple models can substantially improve prediction, particularly when combined with a network encoding model. Our findings imply that the properties of the auditory periphery and central pathway may together result in a simpler than expected functional transformation from ear to cortex. Thus, much of the detailed biological complexity seen in the auditory periphery does not appear to be important for understanding the cortical representation of sound.

Spectral cues and temporal integration during cylinder echo discrimination by bottlenose dolphins (Tursiops truncatus)

Three bottlenose dolphins (Tursiops truncatus) participated in simulated cylinder wall thickness discrimination tasks utilizing electronic "phantom" echoes. The first experiment resulted in psychometric functions (percent correct vs wall thickness difference) similar to those produced by a dolphin performing the task with physical cylinders. In the second experiment, a wide range of cylinder echoes was simulated, with the time separation between echo highlights covering a range from <30 to >300 μs. Dolphin performance and a model of the dolphin auditory periphery suggest that the dolphins used high-frequency, spectral-profiles of the echoes for discrimination and that the utility of spectral cues degraded when the time separation between echo highlights approached and exceeded the dolphin's temporal integration time of ∼264 μs.

Periodicity Pitch Perception

This study presents a computational model to reproduce the biological dynamics of "listening to music." A biologically plausible model of periodicity pitch detection is proposed and simulated. Periodicity pitch is computed across a range of the auditory spectrum. Periodicity pitch is detected from subsets of activated auditory nerve fibers (ANFs). These activate connected model octopus cells, which trigger model neurons detecting onsets and offsets; thence model interval-tuned neurons are innervated at the right interval times; and finally, a set of common interval-detecting neurons indicate pitch. Octopus cells rhythmically spike with the pitch periodicity of the sound. Batteries of interval-tuned neurons stopwatch-like measure the inter-spike intervals of the octopus cells by coding interval durations as first spike latencies (FSLs). The FSL-triggered spikes synchronously coincide through a monolayer spiking neural network at the corresponding receiver pitch neurons.

Phase Locking of Auditory Nerve Fibers: The Role of Lowpass Filtering by Hair Cells

Phase locking of auditory-nerve-fiber (ANF) responses to the temporal fine structure of acoustic stimuli, a hallmark of the auditory system's temporal precision, is important for many aspects of hearing. Previous work has shown that phase-locked period histograms are often well described by exponential transfer functions relating instantaneous stimulus pressure to instantaneous spike rate, with no observed clipping of the histograms. The operating points and slopes of these functions change with stimulus level. The mechanism underlying this apparent gain control is unclear but is distinct from mechanical compression, is independent of refractoriness and spike-rate adaptation, and is apparently instantaneous. Here we show that these findings can be accounted for by a model consisting of a static Boltzmann transducer function yielding a clipped output, followed by a lowpass filter and a static exponential transfer function. Using responses to tones of ANFs from cats of both sexes, we show that, for a given ANF, the period histograms obtained at all stimulus levels for a given stimulus frequency can be described using one set of level-independent model parameters. The model also accounts for changes in the maximum and minimum instantaneous spike rates with changes in stimulus level. Notably, the estimated cutoff frequency is lower for low- than for high-spontaneous-rate ANFs, implying a synapse-specific contribution to lowpass filtering. These findings advance our understanding of ANF phase locking by highlighting the role of peripheral filtering mechanisms in shaping responses of individual ANFs.SIGNIFICANCE STATEMENT Phase locking of auditory-nerve-fiber responses to the temporal fine structure of acoustic stimuli is important for many aspects of hearing. Period histograms typically retain an approximately sinusoidal shape across stimulus levels, with the peripheral auditory system operating as though its overall transfer function is an exponential function whose slope decreases with increasing stimulus level. This apparent gain control can be accounted for by a static saturating transducer function followed by a lowpass filter. In addition to attenuating the AC component, the filter approximately recovers the sinusoidal waveform of the stimulus. The estimated cutoff frequency varies with spontaneous rate, revealing a synaptic contribution to lowpass filtering. These findings highlight the significant impact of peripheral filtering mechanisms on phase locking.

Neural fluctuation cues for simultaneous notched-noise masking and profile-analysis tasks: Insights from model midbrain responses

Results of simultaneous notched-noise masking are commonly interpreted as reflecting the bandwidth of underlying auditory filters. This interpretation assumes that listeners detect a tone added to notched-noise based on an increase in energy at the output of an auditory filter. Previous work challenged this assumption by showing that randomly and independently varying (roving) the levels of each stimulus interval does not substantially worsen listener thresholds [Lentz, Richards, and Matiasek (1999). J. Acoust. Soc. Am. 106, 2779-2792]. Lentz et al. further challenged this assumption by showing that filter bandwidths based on notched-noise results were different from those based on a profile-analysis task [Green (1983). Am. Psychol. 38, 133-142; (1988). (Oxford University Press, New York)], although these estimates were later reconciled by emphasizing spectral peaks of the profile-analysis stimulus [Lentz (2006). J. Acoust. Soc. Am. 120, 945-956]. Here, a single physiological model is shown to account for performance in fixed- and roving-level notched-noise tasks and the Lentz et al. profile-analysis task. This model depends on peripheral neural fluctuation cues that are transformed into the average rates of model inferior colliculus neurons. Neural fluctuations are influenced by peripheral filters, synaptic adaptation, cochlear amplification, and saturation of inner hair cells, an element not included in previous theories of envelope-based cues for these tasks. Results suggest reevaluation of the interpretation of performance in these paradigms.

Phase Locking of Auditory-Nerve Fibers Reveals Stereotyped Distortions and an Exponential Transfer Function with a Level-Dependent Slope

Phase locking of auditory-nerve-fiber (ANF) responses to the fine structure of acoustic stimuli is a hallmark of the auditory system's temporal precision and is important for many aspects of hearing. Period histograms from phase-locked ANF responses to low-frequency tones exhibit spike-rate and temporal asymmetries, but otherwise retain an approximately sinusoidal shape as stimulus level increases, even beyond the level at which the mean spike rate saturates. This is intriguing because apical cochlear mechanical vibrations show little compression, and mechanoelectrical transduction in the receptor cells is thought to obey a static sigmoidal nonlinearity, which might be expected to produce peak clipping at moderate and high stimulus levels. Here we analyze phase-locked responses of ANFs from cats of both sexes. We show that the lack of peak clipping is due neither to ANF refractoriness nor to spike-rate adaptation on time scales longer than the stimulus period. We demonstrate that the relationship between instantaneous pressure and instantaneous rate is well described by an exponential function whose slope decreases with increasing stimulus level. Relatively stereotyped harmonic distortions in the input to the exponential can account for the temporal asymmetry of the period histograms, including peak splitting. We show that the model accounts for published membrane-potential waveforms when assuming a power-of-three, but not a power-of-one, relationship to exocytosis. Finally, we demonstrate the relationship between the exponential transfer functions and the sigmoidal pseudotransducer functions obtained in the literature by plotting the maxima and minima of the voltage responses against the maxima and minima of the stimuli.SIGNIFICANCE STATEMENT Phase locking of auditory-nerve-fiber responses to the temporal fine structure of acoustic stimuli is important for many aspects of hearing, but the mechanisms underlying phase locking are not fully understood. Intriguingly, period histograms retain an approximately sinusoidal shape across sound levels, even when the mean rate has saturated. We find that neither refractoriness nor spike-rate adaptation is responsible for this behavior. Instead, the peripheral auditory system operates as though it contains an exponential transfer function whose slope changes with stimulus level. The underlying mechanism is distinct from the comparatively weak cochlear mechanical compression in the cochlear apex, and likely resides in the receptor cells.

Changes in Neuronal Representations of Consonants in the Ascending Auditory System and Their Role in Speech Recognition

A fundamental task of the ascending auditory system is to produce representations that facilitate the recognition of complex sounds. This is particularly challenging in the context of acoustic variability, such as that between different talkers producing the same phoneme. These representations are transformed as information is propagated throughout the ascending auditory system from the inner ear to the auditory cortex (AI). Investigating these transformations and their role in speech recognition is key to understanding hearing impairment and the development of future clinical interventions. Here, we obtained neural responses to an extensive set of natural vowel-consonant-vowel phoneme sequences, each produced by multiple talkers, in three stages of the auditory processing pathway. Auditory nerve (AN) representations were simulated using a model of the peripheral auditory system and extracellular neuronal activity was recorded in the inferior colliculus (IC) and primary auditory cortex (AI) of anaesthetized guinea pigs. A classifier was developed to examine the efficacy of these representations for recognizing the speech sounds. Individual neurons convey progressively less information from AN to AI. Nonetheless, at the population level, representations are sufficiently rich to facilitate recognition of consonants with a high degree of accuracy at all stages indicating a progression from a dense, redundant representation to a sparse, distributed one. We examined the timescale of the neural code for consonant recognition and found that optimal timescales increase throughout the ascending auditory system from a few milliseconds in the periphery to several tens of milliseconds in the cortex. Despite these longer timescales, we found little evidence to suggest that representations up to the level of AI become increasingly invariant to across-talker differences. Instead, our results support the idea that the role of the subcortical auditory system is one of dimensionality expansion, which could provide a basis for flexible classification of arbitrary speech sounds.

Multidimensional stimulus encoding in the auditory nerve of the barn owl

Auditory perception depends on multi-dimensional information in acoustic signals that must be encoded by auditory nerve fibers (ANF). These dimensions are represented by filters with different frequency selectivities. Multiple models have been suggested; however, the identification of relevant filters and type of interactions has been elusive, limiting progress in modeling the cochlear output. Spike-triggered covariance analysis of barn owl ANF responses was used to determine the number of relevant stimulus filters and estimate the nonlinearity that produces responses from filter outputs. This confirmed that ANF responses depend on multiple filters. The first, most dominant filter was the spike-triggered average, which was excitatory for all neurons. The second and third filters could be either suppressive or excitatory with center frequencies above or below that of the first filter. The nonlinear function mapping the first two filter outputs to the spiking probability ranged from restricted to nearly circular-symmetric, reflecting different modes of interaction between stimulus dimensions across the sample. This shows that stimulus encoding in ANFs of the barn owl is multidimensional and exhibits diversity over the population, suggesting that models must allow for variable numbers of filters and types of interactions between filters to describe how sound is encoded in ANFs.

Parallel Distribution of an Inner Hair Cell and Auditory Nerve Model for Real-Time Application

This paper summarizes recent efforts in implementing a model of the ear's inner hair cell and auditory nerve on a neuromorphic hardware platform, the SpiNNaker machine. This exploits the massive parallelism of the target architecture to obtain real-time modeling of a biologically realistic number of human auditory nerve fibres. We show how this model can be integrated with additional modules that simulate previous stages of the early auditory pathway running on the same hardware architecture, thus producing a full-scale spiking auditory nerve output from a single sound stimulus. The results of the SpiNNaker implementation are shown to be comparable with a MATLAB version of the same model algorithms, while removing the inherent performance limitations associated with an increase in auditory model scale that are seen in the conventional computer simulations. Finally, we outline the potential for using this system as part of a full-scale, real-time digital model of the complete human auditory pathway on the SpiNNaker platform.

Modeling Pitch Perception With an Active Auditory Model Extended by Octopus Cells

Pitch is an essential category for musical sensations. Models of pitch perception are vividly discussed up to date. Most of them rely on definitions of mathematical methods in the spectral or temporal domain. Our proposed pitch perception model is composed of an active auditory model extended by octopus cells. The active auditory model is the same as used in the Stimulation based on Auditory Modeling (SAM), a successful cochlear implant sound processing strategy extended here by modeling the functional behavior of the octopus cells in the ventral cochlear nucleus and by modeling their connections to the auditory nerve fibers (ANFs). The neurophysiological parameterization of the extended model is fully described in the time domain. The model is based on latency-phase en- and decoding as octopus cells are latency-phase rectifiers in their local receptive fields. Pitch is ubiquitously represented by cascaded firing sweeps of octopus cells. Based on the firing patterns of octopus cells, inter-spike interval histograms can be aggregated, in which the place of the global maximum is assumed to encode the pitch.

The effects of the activation of the inner-hair-cell basolateral K+ channels on auditory nerve responses

The basolateral membrane of the mammalian inner hair cell (IHC) expresses large voltage and Ca²⁺ gated outward K⁺ currents. To quantify how the voltage-dependent activation of the K⁺ channels affects the functionality of the auditory nerve innervating the IHC, this study adopts a model of mechanical-to-neural transduction in which the basolateral K⁺ conductances of the IHC can be made voltage-dependent or not. The model shows that the voltage-dependent activation of the K⁺ channels (i) enhances the phase-locking properties of the auditory fiber (AF) responses; (ii) enables the auditory nerve to encode a large dynamic range of sound levels; (iii) enables the AF responses to synchronize precisely with the envelope of amplitude modulated stimuli; and (iv), is responsible for the steep offset responses of the AFs. These results suggest that the basolateral K⁺ channels play a major role in determining the well-known response properties of the AFs and challenge the classical view that describes the IHC membrane as an electrical low-pass filter. In contrast to previous models of the IHC-AF complex, this study ascribes many of the AF response properties to fairly basic mechanisms in the IHC membrane rather than to complex mechanisms in the synapse.

The contribution of visual information to the perception of speech in noise with and without informative temporal fine structure

Understanding what is said in demanding listening situations is assisted greatly by looking at the face of a talker. Previous studies have observed that normal-hearing listeners can benefit from this visual information when a talker's voice is presented in background noise. These benefits have also been observed in quiet listening conditions in cochlear-implant users, whose device does not convey the informative temporal fine structure cues in speech, and when normal-hearing individuals listen to speech processed to remove these informative temporal fine structure cues. The current study (1) characterised the benefits of visual information when listening in background noise; and (2) used sine-wave vocoding to compare the size of the visual benefit when speech is presented with or without informative temporal fine structure. The accuracy with which normal-hearing individuals reported words in spoken sentences was assessed across three experiments. The availability of visual information and informative temporal fine structure cues was varied within and across the experiments. The results showed that visual benefit was observed using open- and closed-set tests of speech perception. The size of the benefit increased when informative temporal fine structure cues were removed. This finding suggests that visual information may play an important role in the ability of cochlear-implant users to understand speech in many everyday situations. Models of audio-visual integration were able to account for the additional benefit of visual information when speech was degraded and suggested that auditory and visual information was being integrated in a similar way in all conditions. The modelling results were consistent with the notion that audio-visual benefit is derived from the optimal combination of auditory and visual sensory cues.

Functional modeling of the human auditory brainstem response to broadband stimulation

Population responses such as the auditory brainstem response (ABR) are commonly used for hearing screening, but the relationship between single-unit physiology and scalp-recorded population responses are not well understood. Computational models that integrate physiologically realistic models of single-unit auditory-nerve (AN), cochlear nucleus (CN) and inferior colliculus (IC) cells with models of broadband peripheral excitation can be used to simulate ABRs and thereby link detailed knowledge of animal physiology to human applications. Existing functional ABR models fail to capture the empirically observed 1.2-2 ms ABR wave-V latency-vs-intensity decrease that is thought to arise from level-dependent changes in cochlear excitation and firing synchrony across different tonotopic sections. This paper proposes an approach where level-dependent cochlear excitation patterns, which reflect human cochlear filter tuning parameters, drive AN fibers to yield realistic level-dependent properties of the ABR wave-V. The number of free model parameters is minimal, producing a model in which various sources of hearing-impairment can easily be simulated on an individualized and frequency-dependent basis. The model fits latency-vs-intensity functions observed in human ABRs and otoacoustic emissions while maintaining rate-level and threshold characteristics of single-unit AN fibers. The simulations help to reveal which tonotopic regions dominate ABR waveform peaks at different stimulus intensities.

Modeling auditory coding: from sound to spikes

Models are valuable tools to assess how deeply we understand complex systems: only if we are able to replicate the output of a system based on the function of its subcomponents can we assume that we have probably grasped its principles of operation. On the other hand, discrepancies between model results and measurements reveal gaps in our current knowledge, which can in turn be targeted by matched experiments. Models of the auditory periphery have improved greatly during the last decades, and account for many phenomena observed in experiments. While the cochlea is only partly accessible in experiments, models can extrapolate its behavior without gap from base to apex and with arbitrary input signals. With models we can for example evaluate speech coding with large speech databases, which is not possible experimentally, and models have been tuned to replicate features of the human hearing organ, for which practically no invasive electrophysiological measurements are available. Auditory models have become instrumental in evaluating models of neuronal sound processing in the auditory brainstem and even at higher levels, where they are used to provide realistic input, and finally, models can be used to illustrate how such a complicated system as the inner ear works by visualizing its responses. The big advantage there is that intermediate steps in various domains (mechanical, electrical, and chemical) are available, such that a consistent picture of the evolvement of its output can be drawn. However, it must be kept in mind that no model is able to replicate all physiological characteristics (yet) and therefore it is critical to choose the most appropriate model—or models—for every research question. To facilitate this task, this paper not only reviews three recent auditory models, it also introduces a framework that allows researchers to easily switch between models. It also provides uniform evaluation and visualization scripts, which allow for direct comparisons between models.

Subcortical representation of musical dyads: individual differences and neural generators

When two notes are played simultaneously they form a musical dyad. The sensation of pleasantness, or "consonance", of a dyad is likely driven by the harmonic relation of the frequency components of the combined spectrum of the two notes. Previous work has demonstrated a relation between individual preference for consonant over dissonant dyads, and the strength of neural temporal coding of the harmonicity of consonant relative to dissonant dyads as measured using the electrophysiological "frequency-following response" (FFR). However, this work also demonstrated that both these variables correlate strongly with musical experience. The current study was designed to determine whether the relation between consonance preference and neural temporal coding is maintained when controlling for musical experience. The results demonstrate that strength of neural coding of harmonicity is predictive of individual preference for consonance even for non-musicians. An additional purpose of the current study was to assess the cochlear generation site of the FFR to low-frequency dyads. By comparing the reduction in FFR strength when high-pass masking noise was added to the output of a model of the auditory periphery, the results provide evidence for the FFR to low-frequency dyads resulting in part from basal cochlear generators.

A model of synaptic vesicle-pool depletion and replenishment can account for the interspike interval distributions and nonrenewal properties of spontaneous spike trains of auditory-nerve fibers

In mammalian auditory systems, the spiking characteristics of each primary afferent (type I auditory-nerve fiber; ANF) are mainly determined by a single ribbon synapse in a single receptor cell (inner hair cell; IHC). ANF spike trains therefore provide a window into the operation of these synapses and cells. It was demonstrated previously (Heil et al., 2007) that the distribution of interspike intervals (ISIs) of cat ANFs during spontaneous activity can be modeled as resulting from refractoriness operating on a non-Poisson stochastic point process of excitation (transmitter release events from the IHC). Here, we investigate nonrenewal properties of these cat-ANF spontaneous spike trains, manifest as negative serial ISI correlations and reduced spike-count variability over short timescales. A previously discussed excitatory process, the constrained failure of events from a homogeneous Poisson point process, can account for these properties, but does not offer a parsimonious explanation for certain trends in the data. We then investigate a three-parameter model of vesicle-pool depletion and replenishment and find that it accounts for all experimental observations, including the ISI distributions, with only the release probability varying between spike trains. The maximum number of units (single vesicles or groups of simultaneously released vesicles) in the readily releasable pool and their replenishment time constant can be assumed to be constant (∼4 and 13.5 ms, respectively). We suggest that the organization of the IHC ribbon synapses not only enables sustained release of neurotransmitter but also imposes temporal regularity on the release process, particularly when operating at high rates.

Modeling the influence of short term depression in vesicle release and stochastic calcium channel gating on auditory nerve spontaneous firing statistics

We propose several modifications to an existing computational model of stochastic vesicle release in inner hair cell ribbon synapses, with the aim of producing simulated auditory nerve fiber spiking data that more closely matches empirical data. Specifically, we studied the inter-spike-interval (ISI) distribution, and long and short term ISI correlations in spontaneous spiking in post-synaptic auditory nerve fibers. We introduced short term plasticity to the pre-synaptic release probability, in a manner analogous to standard stochastic models of cortical short term synaptic depression. This modification resulted in a similar distribution of vesicle release intervals to that estimated from empirical data. We also introduced a biophysical stochastic model of calcium channel opening and closing, but showed that this model is insufficient for generating a match with empirically observed spike correlations. However, by combining a phenomenological model of channel noise and our short term depression model, we generated short and long term correlations in auditory nerve spontaneous activity that qualitatively match empirical data.

Why do I hear but not understand? Stochastic undersampling as a model of degraded neural encoding of speech

Hearing impairment is a serious disease with increasing prevalence. It is defined based on increased audiometric thresholds but increased thresholds are only partly responsible for the greater difficulty understanding speech in noisy environments experienced by some older listeners or by hearing-impaired listeners. Identifying the additional factors and mechanisms that impair intelligibility is fundamental to understanding hearing impairment but these factors remain uncertain. Traditionally, these additional factors have been sought in the way the speech spectrum is encoded in the pattern of impaired mechanical cochlear responses. Recent studies, however, are steering the focus toward impaired encoding of the speech waveform in the auditory nerve. In our recent work, we gave evidence that a significant factor might be the loss of afferent auditory nerve fibers, a pathology that comes with aging or noise overexposure. Our approach was based on a signal-processing analogy whereby the auditory nerve may be regarded as a stochastic sampler of the sound waveform and deafferentation may be described in terms of waveform undersampling. We showed that stochastic undersampling simultaneously degrades the encoding of soft and rapid waveform features, and that this degrades speech intelligibility in noise more than in quiet without significant increases in audiometric thresholds. Here, we review our recent work in a broader context and argue that the stochastic undersampling analogy may be extended to study the perceptual consequences of various different hearing pathologies and their treatment.

Contribution of auditory nerve fibers to compound action potential of the auditory nerve

Sound-evoked compound action potential (CAP), which captures the synchronous activation of the auditory nerve fibers (ANFs), is commonly used to probe deafness in experimental and clinical settings. All ANFs are believed to contribute to CAP threshold and amplitude: low sound pressure levels activate the high-spontaneous rate (SR) fibers, and increasing levels gradually recruit medium- and then low-SR fibers. In this study, we quantitatively analyze the contribution of the ANFs to CAP 6 days after 30-min infusion of ouabain into the round window niche. Anatomic examination showed a progressive ablation of ANFs following increasing concentration of ouabain. CAP amplitude and threshold plotted against loss of ANFs revealed three ANF pools: 1) a highly ouabain-sensitive pool, which does not participate in either CAP threshold or amplitude, 2) a less sensitive pool, which only encoded CAP amplitude, and 3) a ouabain-resistant pool, required for CAP threshold and amplitude. Remarkably, distribution of the three pools was similar to the SR-based ANF distribution (low-, medium-, and high-SR fibers), suggesting that the low-SR fiber loss leaves the CAP unaffected. Single-unit recordings from the auditory nerve confirmed this hypothesis and further showed that it is due to the delayed and broad first spike latency distribution of low-SR fibers. In addition to unraveling the neural mechanisms that encode CAP, our computational simulation of an assembly of guinea pig ANFs generalizes and extends our experimental findings to different species of mammals. Altogether, our data demonstrate that substantial ANF loss can coexist with normal hearing threshold and even unchanged CAP amplitude.

A hardware model of the auditory periphery to transduce acoustic signals into neural activity

To improve the performance of cochlear implants, we have integrated a microdevice into a model of the auditory periphery with the goal of creating a microprocessor. We constructed an artificial peripheral auditory system using a hybrid model in which polyvinylidene difluoride was used as a piezoelectric sensor to convert mechanical stimuli into electric signals. To produce frequency selectivity, the slit on a stainless steel base plate was designed such that the local resonance frequency of the membrane over the slit reflected the transfer function. In the acoustic sensor, electric signals were generated based on the piezoelectric effect from local stress in the membrane. The electrodes on the resonating plate produced relatively large electric output signals. The signals were fed into a computer model that mimicked some functions of inner hair cells, inner hair cell–auditory nerve synapses, and auditory nerve fibers. In general, the responses of the model to pure-tone burst and complex stimuli accurately represented the discharge rates of high-spontaneous-rate auditory nerve fibers across a range of frequencies greater than 1 kHz and middle to high sound pressure levels. Thus, the model provides a tool to understand information processing in the peripheral auditory system and a basic design for connecting artificial acoustic sensors to the peripheral auditory nervous system. Finally, we discuss the need for stimulus control with an appropriate model of the auditory periphery based on auditory brainstem responses that were electrically evoked by different temporal pulse patterns with the same pulse number.

Effect of metabolic presbyacusis on cochlear responses: a simulation approach using a physiologically-based model

In the presented model, electrical, acoustical, and mechanical elements of the cochlea are explicitly integrated into a signal transmission line where these elements convey physiological interpretations of the human cochlear structures. As a result, this physiologically-motivated model enables simulation of specific cochlear lesions such as presbyacusis. The hypothesis is that high-frequency hearing loss in older adults may be due to metabolic presbyacusis whereby age-related cellular/chemical degenerations in the lateral wall of the cochlea cause a reduction in the endocochlear potential. The simulations quantitatively confirm this hypothesis and emphasize that even if the outer and inner hair cells are totally active and intact, metabolic presbyacusis alone can significantly deteriorate the cochlear functionality. Specifically, in the model, as the endocochlear potential decreases, the transduction mechanism produces less receptor current such that there is a reduction in the battery of the somatic motor. This leads to a drastic decrease in cochlear amplification and frequency sensitivity, as well as changes in position-frequency map (tuning pattern) of the cochlea. In addition, the simulations show that the age-related reduction of the endocochlear potential significantly inhibits the firing rate of the auditory nerve which might contribute to the decline of temporal resolution in the aging auditory system.

From birdsong to human speech recognition: bayesian inference on a hierarchy of nonlinear dynamical systems

Our knowledge about the computational mechanisms underlying human learning and recognition of sound sequences, especially speech, is still very limited. One difficulty in deciphering the exact means by which humans recognize speech is that there are scarce experimental findings at a neuronal, microscopic level. Here, we show that our neuronal-computational understanding of speech learning and recognition may be vastly improved by looking at an animal model, i.e., the songbird, which faces the same challenge as humans: to learn and decode complex auditory input, in an online fashion. Motivated by striking similarities between the human and songbird neural recognition systems at the macroscopic level, we assumed that the human brain uses the same computational principles at a microscopic level and translated a birdsong model into a novel human sound learning and recognition model with an emphasis on speech. We show that the resulting Bayesian model with a hierarchy of nonlinear dynamical systems can learn speech samples such as words rapidly and recognize them robustly, even in adverse conditions. In addition, we show that recognition can be performed even when words are spoken by different speakers and with different accents—an everyday situation in which current state-of-the-art speech recognition models often fail. The model can also be used to qualitatively explain behavioral data on human speech learning and derive predictions for future experiments.

Neuroscience still lacks a concrete explanation of how humans recognize speech. Even though neuroimaging techniques are helpful in determining the brain areas involved in speech recognition, there are rarely mechanistic explanations at a neuronal level. Here, we assume that songbirds and humans solve a very similar task: extracting information from sound wave modulations produced by a singing bird or a speaking human. Given strong evidence that both humans and songbirds, although genetically very distant, converged to a similar solution, we combined the vast amount of neurobiological findings for songbirds with nonlinear dynamical systems theory to develop a hierarchical, Bayesian model which explains fundamental functions in recognition of sound sequences. We found that the resulting model is good at learning and recognizing human speech. We suggest that this translated model can be used to qualitatively explain or predict experimental data, and the underlying mechanism can be used to construct improved automatic speech recognition algorithms.

Making use of auditory models for better mimicking of normal hearing processes with cochlear implants: the SAM coding strategy

Mimicking the human ear on the basis of auditory models has become a viable approach in many applications by now. However, only a few attempts have been made to extend the scope of physiological ear models to be employed in cochlear implants (CI). Contemporary CI systems rely on much simpler filter banks and simulate the natural signal processing of a healthy cochlea to only a very limited extent. When looking at rehabilitation outcomes, current systems seem to have reached their peak potential, which signals the need for better algorithms and/or technologies. In this paper, we present a novel sound processing strategy, SAM (Stimulation based on Auditory Modeling), that is based on neurophysiological models of the human ear and can be employed in auditory prostheses. It incorporates active cochlear filtering (basilar membrane and outer hair cells) along with the mechanoelectrical transduction of the inner hair cells, so that several psychoacoustic phenomena are accounted for inherently. Although possible, current implementation does not make use of parallel stimulation of the electrodes, which matches state-of-the-art CI hardware. This paper elaborates on SAM's signal processing and provides a computational evaluation of the strategy. Results show that aspects of normal cochlear processing that are missing in common strategies can be replicated by SAM. This is supposed to improve overall CI user performance, which we have at least partly proven in a pilot study with implantees.

On the controversy about the sharpness of human cochlear tuning

In signal processing terms, the operation of the mammalian cochlea in the inner ear may be likened to a bank of filters. Based on otoacoustic emission evidence, it has been recently claimed that cochlear tuning is sharper for human than for other mammals. The claim was corroborated with a behavioral method that involves the masking of pure tones with forward notched noises (NN). Using this method, it has been further claimed that human cochlear tuning is sharper than suggested by earlier behavioral studies. These claims are controversial. Here, we contribute to the controversy by theoretically assessing the accuracy of the NN method at inferring the bandwidth (BW) of nonlinear cochlear filters. Behavioral forward masking was mimicked using a computer model of the squared basilar membrane response followed by a temporal integrator. Isoresponse and isolevel versions of the forward masking NN method were applied to infer the already known BW of the cochlear filter used in the model. We show that isolevel methods were overall more accurate than isoresponse methods. We also show that BWs for NNs and sinusoids equate only for isolevel methods and when the levels of the two stimuli are appropriately scaled. Lastly, we show that the inferred BW depends on the method version (isolevel BW was twice as broad as isoresponse BW at 40 dB SPL) and on the stimulus level (isoresponse and isolevel BW decreased and increased, respectively, with increasing level over the level range where cochlear responses went from linear to compressive). We suggest that the latter may contribute to explaining the reported differences in cochlear tuning across behavioral studies and species. We further suggest that given the well-established nonlinear nature of cochlear responses, even greater care must be exercised when using a single BW value to describe and compare cochlear tuning.

Pitch shifts on mistuned harmonics in the presence and absence of corresponding in-tune components

Mistuning a harmonic produces an exaggerated change in its pitch, a component-pitch shift. The origin of these pitch shifts was explored by manipulations intended to alter the grouping status of a mistuned target component in a periodic complex tone. In experiment 1, which used diotic presentation, reinstating the corresponding harmonic (in-tune counterpart) caused the pitch shifts on the mistuned target largely to disappear for components 3 and 4, although they remained for component 2. A computational model of component-pitch shifts, based on harmonic cancellation, was unable to explain the near-complete loss of pitch shifts when the counterpart was present; only small changes occurred. In experiment 2, the complex tone and mistuned component 4 were presented in the left ear and the in-tune counterpart was presented in the right. The in-tune counterpart again reduced component-pitch shifts, but they were restored when a captor complex into which the counterpart fitted as harmonic 3 was added in the right ear; presumably by providing an alternative grouping possibility for the counterpart. It is proposed that component-pitch shifts occur only if the mistuned component is selected to contribute to the complex-tone percept; these shifts are eliminated if it is displaced by a better candidate.

A frequency-selective feedback model of auditory efferent suppression and its implications for the recognition of speech in noise

The potential contribution of the peripheral auditory efferent system to our understanding of speech in a background of competing noise was studied using a computer model of the auditory periphery and assessed using an automatic speech recognition system. A previous study had shown that a fixed efferent attenuation applied to all channels of a multi-channel model could improve the recognition of connected digit triplets in noise [G. J. Brown, R. T. Ferry, and R. Meddis, J. Acoust. Soc. Am. 127, 943-954 (2010)]. In the current study an anatomically justified feedback loop was used to automatically regulate separate attenuation values for each auditory channel. This arrangement resulted in a further enhancement of speech recognition over fixed-attenuation conditions. Comparisons between multi-talker babble and pink noise interference conditions suggest that the benefit originates from the model's ability to modify the amount of suppression in each channel separately according to the spectral shape of the interfering sounds.

A neurocognitive model of recognition and pitch segregation

This paper describes a neurocognitive model of pitch segregation in which it is proposed that recognition mechanisms initiate early in auditory processing pathways so that long-term memory templates may be employed to segregate and integrate auditory features. In this model neural representations of pitch height are primed by the location and pattern of excitation across auditory filter channels in relation to long-term memory templates for common stimuli. Since waveform driven pitch mechanisms may produce information at multiple frequencies for tonal stimuli, pitch priming was assumed to include competitive inhibition that would allow only one pitch estimation at any time. Consequently concurrent pitch information must be relayed to short-term memory via a parallel mechanism that employs pitch information contained in the long-term memory template of the chord. Pure tones, harmonic complexes and two pitch chords of harmonic complexes were correctly classified by the correlation of templates comprising auditory nerve excitation and off-frequency inhibition with the excitation patterns of stimuli. The model then replicated behavioral data for pitch matching of concurrent vowels. Comparison of model outputs to the behavioral data suggests that inability to recognize a stimulus was associated with poor pitch segregation due to the use of inappropriate pitch priming strategies.

The frequency following response (FFR) may reflect pitch-bearing information but is not a direct representation of pitch

The frequency following response (FFR), a scalp-recorded measure of phase-locked brainstem activity, is often assumed to reflect the pitch of sounds as perceived by humans. In two experiments, we investigated the characteristics of the FFR evoked by complex tones. FFR waveforms to alternating-polarity stimuli were averaged for each polarity and added, to enhance envelope, or subtracted, to enhance temporal fine structure information. In experiment 1, frequency-shifted complex tones, with all harmonics shifted by the same amount in Hertz, were presented diotically. Only the autocorrelation functions (ACFs) of the subtraction-FFR waveforms showed a peak at a delay shifted in the direction of the expected pitch shifts. This expected pitch shift was also present in the ACFs of the output of an auditory nerve model. In experiment 2, the components of a harmonic complex with harmonic numbers 2, 3, and 4 were presented either to the same ear ("mono") or the third harmonic was presented contralaterally to the ear receiving the even harmonics ("dichotic"). In the latter case, a pitch corresponding to the missing fundamental was still perceived. Monaural control conditions presenting only the even harmonics ("2 + 4") or only the third harmonic ("3") were also tested. Both the subtraction and the addition waveforms showed that (1) the FFR magnitude spectra for "dichotic" were similar to the sum of the spectra for the two monaural control conditions and lacked peaks at the fundamental frequency and other distortion products visible for "mono" and (2) ACFs for "dichotic" were similar to those for "2 + 4" and dissimilar to those for "mono." The results indicate that the neural responses reflected in the FFR preserve monaural temporal information that may be important for pitch, but provide no evidence for any additional processing over and above that already present in the auditory periphery, and do not directly represent the pitch of dichotic stimuli.

The influence of different segments of the ongoing envelope on sensitivity to interaural time delays

The auditory system is sensitive to interaural timing disparities in the fine structure and the envelope of sounds, each contributing important cues for lateralization. In this study, psychophysical measurements were conducted with customized envelope waveforms in order to investigate the isolated effect of different segments of a periodic, ongoing envelope on lateralization. One envelope cycle was composed of the four segments attack flank, hold duration, decay flank, and pause duration, which were independently varied to customize the envelope waveform. The envelope waveforms were applied to a 4-kHz sinusoidal carrier, and just noticeable envelope interaural time differences were measured in six normal hearing subjects. The results indicate that attack durations and pause durations prior to the attack are the most important stimulus characteristics for processing envelope timing disparities. The results were compared to predictions of three binaural lateralization models based on the normalized cross correlation coefficient. Two of the models included an additional stage to mimic neural adaptation prior to binaural interaction, involving either a single short time constant (5 ms) or a combination of five time constants up to 500 ms. It was shown that the model with the single short time constant accounted best for the data.

A finite-sample, distribution-free, probabilistic lower bound on mutual information

For any memoryless communication channel with a binary-valued input and a one-dimensional real-valued output, we introduce a probabilistic lower bound on the mutual information given empirical observations on the channel. The bound is built on the Dvoretzky-Kiefer-Wolfowitz inequality and is distribution free. A quadratic time algorithm is described for computing the bound and its corresponding class-conditional distribution functions. We compare our approach to existing techniques and show the superiority of our bound to a method inspired by Fano's inequality where the continuous random variable is discretized.

Auditory information coding by modeled cochlear nucleus neurons

In this paper we use information theory to quantify the information in the output spike trains of modeled cochlear nucleus globular bushy cells (GBCs). GBCs are part of the sound localization pathway. They are known for their precise temporal processing, and they code amplitude modulations with high fidelity. Here we investigated the information transmission for a natural sound, a recorded vowel. We conclude that the maximum information transmission rate for a single neuron was close to 1,050 bits/s, which corresponds to a value of approximately 5.8 bits per spike. For quasi-periodic signals like voiced speech, the transmitted information saturated as word duration increased. In general, approximately 80% of the available information from the spike trains was transmitted within about 20 ms. Transmitted information for speech signals concentrated around formant frequency regions. The efficiency of neural coding was above 60% up to the highest temporal resolution we investigated (20 μs). The increase in transmitted information to that precision indicates that these neurons are able to code information with extremely high fidelity, which is required for sound localization. On the other hand, only 20% of the information was captured when the temporal resolution was reduced to 4 ms. As the temporal resolution of most speech recognition systems is limited to less than 10 ms, this massive information loss might be one of the reasons which are responsible for the lack of noise robustness of these systems.