Modulation of auditory processing by cortico-cortical feed-forward and feedback projections

Low-Intensity Ultrasound Causes Direct Excitation of Auditory Cortical Neurons

Cochlear implantation is the first-line treatment for severe and profound hearing loss in children and adults. However, deaf patients with cochlear malformations or with cochlear nerve deficiencies are ineligible for cochlear implants. Meanwhile, the limited spatial selectivity and high risk of invasive craniotomy restrict the wide application of auditory brainstem implants. A noninvasive alternative strategy for safe and effective neuronal stimulation is urgently needed to address this issue. Because of its advantage in neural modulation over electrical stimulation, low-intensity ultrasound (US) is considered a safe modality for eliciting neural activity in the central auditory system. Although the neural modulation ability of low-intensity US has been demonstrated in the human primary somatosensory cortex and primary visual cortex, whether low-intensity US can directly activate auditory cortical neurons is still a topic of debate. To clarify the direct effects on auditory neurons, in the present study, we employed low-intensity US to stimulate auditory cortical neurons in vitro. Our data show that both low-frequency (0.8 MHz) and high-frequency (>27 MHz) US stimulation can elicit the inward current and action potentials in cultured neurons. c-Fos staining results indicate that low-intensity US is efficient for stimulating most neurons. Our study suggests that low-intensity US can excite auditory cortical neurons directly, implying that US-induced neural modulation can be a potential approach for activating the auditory cortex of deaf patients.

Laminar specificity of oscillatory coherence in the auditory cortex

Empirical evidence suggests that, in the auditory cortex (AC), the phase relationship between spikes and local-field potentials (LFPs) plays an important role in the processing of auditory stimuli. Nevertheless, unlike the case of other sensory systems, it remains largely unexplored in the auditory modality whether the properties of the cortical columnar microcircuit shape the dynamics of spike-LFP coherence in a layer-specific manner. In this study, we directly tackle this issue by addressing whether spike-LFP and LFP-stimulus phase synchronization are spatially distributed in the AC during sensory processing, by performing laminar recordings in the cortex of awake short-tailed bats (Carollia perspicillata) while animals listened to conspecific distress vocalizations. We show that, in the AC, spike-LFP and LFP-stimulus synchrony depend significantly on cortical depth, and that sensory stimulation alters the spatial and spectral patterns of spike-LFP phase-locking. We argue that such laminar distribution of coherence could have functional implications for the representation of naturalistic auditory stimuli at a cortical level.

Inhibitory mechanisms shaping delay-tuned combination-sensitivity in the auditory cortex and thalamus of the mustached bat

Delay-tuned auditory neurons of the mustached bat show facilitative responses to a combination of signal elements of a biosonar pulse-echo pair with a specific echo delay. The subcollicular nuclei produce latency-constant phasic on-responding neurons, and the inferior colliculus produces delay-tuned combination-sensitive neurons, designated "FM-FM" neurons. The combination-sensitivity is a facilitated response to the coincidence of the excitatory rebound following glycinergic inhibition to the pulse (1st harmonic) and the short-latency response to the echo (2nd-4th harmonics). The facilitative response of thalamic FM-FM neurons is mediated by glutamate receptors (NMDA and non-NMDA receptors). Different from collicular FM-FM neurons, thalamic ones respond more selectively to pulse-echo pairs than individual signal elements. A number of differences in response properties between collicular and thalamic or cortical FM-FM neurons have been reported. However, differences between thalamic and cortical FM-FM neurons have remained to be studied. Here, we report that GABAergic inhibition controls the duration of burst of spikes of facilitative responses of thalamic FM-FM neurons and sharpens the delay tuning of cortical ones. That is, intra-cortical inhibition sharpens the delay tuning of cortical FM-FM neurons that is potentially broad because of divergent/convergent thalamo-cortical projections. Compared with thalamic neurons, cortical ones tend to show sharper delay tuning, longer response duration, and larger facilitation index. However, those differences are statistically insignificant.

Unsupervised speech recognition through spike-timing-dependent plasticity in a convolutional spiking neural network

Speech recognition (SR) has been improved significantly by artificial neural networks (ANNs), but ANNs have the drawbacks of biologically implausibility and excessive power consumption because of the nonlocal transfer of real-valued errors and weights. While spiking neural networks (SNNs) have the potential to solve these drawbacks of ANNs due to their efficient spike communication and their natural way to utilize kinds of synaptic plasticity rules found in brain for weight modification. However, existing SNN models for SR either had bad performance, or were trained in biologically implausible ways. In this paper, we present a biologically inspired convolutional SNN model for SR. The network adopts the time-to-first-spike coding scheme for fast and efficient information processing. A biological learning rule, spike-timing-dependent plasticity (STDP), is used to adjust the synaptic weights of convolutional neurons to form receptive fields in an unsupervised way. In the convolutional structure, the strategy of local weight sharing is introduced and could lead to better feature extraction of speech signals than global weight sharing. We first evaluated the SNN model with a linear support vector machine (SVM) on the TIDIGITS dataset and it got the performance of 97.5%, comparable to the best results of ANNs. Deep analysis on network outputs showed that, not only are the output data more linearly separable, but they also have fewer dimensions and become sparse. To further confirm the validity of our model, we trained it on a more difficult recognition task based on the TIMIT dataset, and it got a high performance of 93.8%. Moreover, a linear spike-based classifier-tempotron-can also achieve high accuracies very close to that of SVM on both the two tasks. These demonstrate that an STDP-based convolutional SNN model equipped with local weight sharing and temporal coding is capable of solving the SR task accurately and efficiently.

A central mechanism enhances pain perception of noxious thermal stimulus changes

Pain perception temporarily exaggerates abrupt thermal stimulus changes revealing a mechanism for nociceptive temporal contrast enhancement (TCE). Although the mechanism is unknown, a non-linear model with perceptual feedback accurately simulates the phenomenon. Here we test if a mechanism in the central nervous system underlies thermal TCE. Our model successfully predicted an optimal stimulus, incorporating a transient temperature offset (step-up/step-down), with maximal TCE, resulting in psychophysically verified large decrements in pain response ("offset-analgesia"; mean analgesia: 85%, n = 20 subjects). Next, this stimulus was delivered using two thermodes, one delivering the longer duration baseline temperature pulse and the other superimposing a short higher temperature pulse. The two stimuli were applied simultaneously either near or far on the same arm, or on opposite arms. Spatial separation across multiple peripheral receptive fields ensures the composite stimulus timecourse is first reconstituted in the central nervous system. Following ipsilateral stimulus cessation on the high temperature thermode, but before cessation of the low temperature stimulus properties of TCE were observed both for individual subjects and in group-mean responses. This demonstrates a central integration mechanism is sufficient to evoke painful thermal TCE, an essential step in transforming transient afferent nociceptive signals into a stable pain perception.

Acuity in ranging based on delay-tuned combination-sensitive neurons in the auditory cortex of mustached bats

A 1.0-ms echo delay from an emitted bio-sonar pulse at 25 °C corresponds to a 17.3-cm target distance. In the auditory cortex of the mustached bat, Pteronotus parnellii, neurons tuned to a specific delay (best delay) of an echo from an emitted pulse are clustered in the FF, dorsal fringe and ventral fringe areas. ("FF" stands for the frequency-modulated components of a pulse and its echo.) Those delay-tuned neurons are systematically arranged in the FF area according to their best delays and form a 18-ms-long delay axis. Using the neurophysiological data, the theoretical acuity at a 75% correct level was computed as just-noticeable changes in (a) the location of maximally responding delay-tuned neurons, (b) the location of the center of all responses in the FF area, and (c) the weighted sum of responses of all delay-tuned neurons. The acuity is range-dependent: the shorter the target range, the higher the acuity is. The just-noticeable changes in target range are 7.57-46.2, 0.50-2.32 and 0.22-2.53 mm at the target ranges of up to 140 cm for (a), (b) and (c), respectively. When the dorsal and ventral fringe areas are included in the computation, the just-noticeable changes become smaller than those in the FF area alone. Those acuities computed are comparable to certain behavioral acuities.

The Effects of Urethane on Rat Outer Hair Cells

The cochlea converts sound vibration into electrical impulses and amplifies the low-level sound signal. Urethane, a widely used anesthetic in animal research, has been shown to reduce the neural responses to auditory stimuli. However, the effects of urethane on cochlea, especially on the function of outer hair cells, remain largely unknown. In the present study, we compared the cochlear microphonic responses between awake and urethane-anesthetized rats. The results revealed that the amplitude of the cochlear microphonic was decreased by urethane, resulting in an increase in the threshold at all of the sound frequencies examined. To deduce the possible mechanism underlying the urethane-induced decrease in cochlear sensitivity, we examined the electrical response properties of isolated outer hair cells using whole-cell patch-clamp recording. We found that urethane hyperpolarizes the outer hair cell membrane potential in a dose-dependent manner and elicits larger outward current. This urethane-induced outward current was blocked by strychnine, an antagonist of the α9 subunit of the nicotinic acetylcholine receptor. Meanwhile, the function of the outer hair cell motor protein, prestin, was not affected. These results suggest that urethane anesthesia is expected to decrease the responses of outer hair cells, whereas the frequency selectivity of cochlea remains unchanged.

Bat auditory cortex – model for general mammalian auditory computation or special design solution for active time perception?

Audition in bats serves passive orientation, alerting functions and communication as it does in other vertebrates. In addition, bats have evolved echolocation for orientation and prey detection and capture. This put a selective pressure on the auditory system in regard to echolocation-relevant temporal computation and frequency analysis. The present review attempts to evaluate in which respect the processing modules of bat auditory cortex (AC) are a model for typical mammalian AC function or are designed for echolocation-unique purposes. We conclude that, while cortical area arrangement and cortical frequency processing does not deviate greatly from that of other mammals, the echo delay time-sensitive dorsal cortex regions contain special designs for very powerful time perception. Different bat species have either a unique chronotopic cortex topography or a distributed salt-and-pepper representation of echo delay. The two designs seem to enable similar behavioural performance.

Sound-by-sound thalamic stimulation modulates midbrain auditory excitability and relative binaural sensitivity in frogs

Descending circuitry can modulate auditory processing, biasing sensitivity to particular stimulus parameters and locations. Using awake in vivo single unit recordings, this study tested whether electrical stimulation of the thalamus modulates auditory excitability and relative binaural sensitivity in neurons of the amphibian midbrain. In addition, by using electrical stimuli that were either longer than the acoustic stimuli (i.e., seconds) or presented on a sound-by-sound basis (ms), experiments addressed whether the form of modulation depended on the temporal structure of the electrical stimulus. Following long duration electrical stimulation (3-10 s of 20 Hz square pulses), excitability (spikes/acoustic stimulus) to free-field noise stimuli decreased by 32%, but returned over 600 s. In contrast, sound-by-sound electrical stimulation using a single 2 ms duration electrical pulse 25 ms before each noise stimulus caused faster and varied forms of modulation: modulation lasted <2 s and, in different cells, excitability either decreased, increased or shifted in latency. Within cells, the modulatory effect of sound-by-sound electrical stimulation varied between different acoustic stimuli, including for different male calls, suggesting modulation is specific to certain stimulus attributes. For binaural units, modulation depended on the ear of input, as sound-by-sound electrical stimulation preceding dichotic acoustic stimulation caused asymmetric modulatory effects: sensitivity shifted for sounds at only one ear, or by different relative amounts for both ears. This caused a change in the relative difference in binaural sensitivity. Thus, sound-by-sound electrical stimulation revealed fast and ear-specific (i.e., lateralized) auditory modulation that is potentially suited to shifts in auditory attention during sound segregation in the auditory scene.

Neural maps for target range in the auditory cortex of echolocating bats

Computational brain maps as opposed to maps of receptor surfaces strongly reflect functional neuronal design principles. In echolocating bats, computational maps are established that topographically represent the distance of objects. These target range maps are derived from the temporal delay between emitted call and returning echo and constitute a regular representation of time (chronotopy). Basic features of these maps are innate, and in different bat species the map size and precision varies. An inherent advantage of target range maps is the implementation of mechanisms for lateral inhibition and excitatory feedback. Both can help to focus target ranging depending on the actual echolocation situation. However, these maps are not absolutely necessary for bat echolocation since there are bat species without cortical target-distance maps, which use alternative ensemble computation mechanisms.

Tuning shifts of the auditory system by corticocortical and corticofugal projections and conditioning

The central auditory system consists of the lemniscal and nonlemniscal systems. The thalamic lemniscal and nonlemniscal auditory nuclei are different from each other in response properties and neural connectivities. The cortical auditory areas receiving the projections from these thalamic nuclei interact with each other through corticocortical projections and project down to the subcortical auditory nuclei. This corticofugal (descending) system forms multiple feedback loops with the ascending system. The corticocortical and corticofugal projections modulate auditory signal processing and play an essential role in the plasticity of the auditory system. Focal electric stimulation - comparable to repetitive tonal stimulation - of the lemniscal system evokes three major types of changes in the physiological properties, such as the tuning to specific values of acoustic parameters of cortical and subcortical auditory neurons through different combinations of facilitation and inhibition. For such changes, a neuromodulator, acetylcholine, plays an essential role. Electric stimulation of the nonlemniscal system evokes changes in the lemniscal system that is different from those evoked by the lemniscal stimulation. Auditory signals ascending from the lemniscal and nonlemniscal thalamic nuclei to the cortical auditory areas appear to be selected or adjusted by a "differential" gating mechanism. Conditioning for associative learning and pseudo-conditioning for nonassociative learning respectively elicit tone-specific and nonspecific plastic changes. The lemniscal, corticofugal and cholinergic systems are involved in eliciting the former, but not the latter. The current article reviews the recent progress in the research of corticocortical and corticofugal modulations of the auditory system and its plasticity elicited by conditioning and pseudo-conditioning.

Developmental neuroplasticity after cochlear implantation

Cortical development is dependent on stimulus-driven learning. The absence of sensory input from birth, as occurs in congenital deafness, affects normal growth and connectivity needed to form a functional sensory system, resulting in deficits in oral language learning. Cochlear implants bypass cochlear damage by directly stimulating the auditory nerve and brain, making it possible to avoid many of the deleterious effects of sensory deprivation. Congenitally deaf animals and children who receive implants provide a platform to examine the characteristics of cortical plasticity in the auditory system. In this review, we discuss the existence of time limits for, and mechanistic constraints on, sensitive periods for cochlear implantation and describe the effects of multimodal and cognitive reorganization that result from long-term auditory deprivation.

Corticocortical interactions between and within three cortical auditory areas specialized for time-domain signal processing

In auditory cortex of the mustached bat, the FF (F means frequency modulation), dorsal fringe (DF), and ventral fringe (VF) areas consist of "combination-sensitive" neurons tuned to the pair of an emitted biosonar pulse and its echo with a specific delay (best delay: BD). The DF and VF areas are hierarchically at a higher level than the FF area. Focal electric stimulation of the FF area evokes "centrifugal" BD shifts of DF neurons, i.e., shifts away from the BD of the stimulated FF neurons, whereas stimulation of the DF neurons evokes "centripetal" BD shifts of FF neurons, i.e., shifts toward the BD of the stimulated DF neurons. In our current studies, we found that the feedforward projection from FF neurons evokes centrifugal BD shifts of VF neurons, that the feedback projection from VF neurons evokes centripetal BD shifts of FF neurons, that the contralateral projection from DF neurons evokes centripetal BD shifts of DF neurons, and that the centripetal BD shifts evoked by the DF and VF neurons are 2.5 times larger than the centrifugal BD shifts evoked by the FF neurons. The centrifugal BD shifts shape the selective neural representation of a specific target distance, whereas the centripetal BD shifts expand the representation of the selected specific target distance to focus on the processing of the target information at a specific distance. The centrifugal and centripetal BD shifts evoked by the feedforward and feedback projections promote finer analysis of a target at shorter distances.