Universal automated classification of the acoustic startle reflex using machine learning

The startle reflex (SR), a robust, motor response elicited by an intense auditory, visual, or somatosensory stimulus has been widely used as a tool to assess psychophysiology in humans and animals for almost a century in diverse fields such as schizophrenia, bipolar disorder, hearing loss, and tinnitus. Previously, SR waveforms have been ignored, or assessed with basic statistical techniques and/or simple template matching paradigms. This has led to considerable variability in SR studies from different laboratories, and species. In an effort to standardize SR assessment methods, we developed a machine learning algorithm and workflow to automatically classify SR waveforms in virtually any animal model including mice, rats, guinea pigs, and gerbils obtained with various paradigms and modalities from several laboratories. The universal features common to SR waveforms of various species and paradigms are examined and discussed in the context of each animal model. The procedure describes common results using the SR across species and how to fully implement the open-source R implementation. Since SR is widely used to investigate toxicological or pharmaceutical efficacy, a detailed and universal SR waveform classification protocol should be developed to aid in standardizing SR assessment procedures across different laboratories and species. This machine learning-based method will improve data reliability and translatability between labs that use the startle reflex paradigm.

Long-Lasting forward Suppression of Spontaneous Firing in Auditory Neurons: Implication to the Residual Inhibition of Tinnitus

Tinnitus is the perception of a sound that has no external source. Sound stimuli can suppress spontaneous firing in auditory neurons long after stimulus offset. It is unknown how changes in sound stimulus parameters affect this forward suppression. Using in vivo extracellular recording in awake mice, we found that about 40 % of spontaneously active inferior colliculus (IC) neurons exhibited forward suppression of spontaneous activity after sound offset. The duration of this suppression increased with sound duration and lasted about 40 s following a 30-s stimulus offset. Pure tones presented at the neuron's characteristic frequency (CF) were more effective in triggering suppression compared to non-CF or wideband noise stimuli. In contrast, non-CF stimuli often induced forward facilitation. About one third of IC neurons exhibited shorter suppression durations with each subsequent sound presentation. These characteristics of forward suppression are similar to the psychoacoustic properties of residual inhibition of tinnitus: a phenomenon of brief (about 30 s) suppression of tinnitus observed in tinnitus patients after sound presentations. Because elevated spontaneous firing in central auditory neurons has been linked to tinnitus, forward suppression of this firing with sound might be an underlying mechanism of residual inhibition.

Prepulse inhibition of the acoustic startle reflex vs. auditory brainstem response for hearing assessment

The high prevalence of noise-induced and age-related hearing loss in the general population has warranted the use of animal models to study the etiology of these pathologies. Quick and accurate auditory threshold determination is a prerequisite for experimental manipulations targeting hearing loss in animal models. The standard auditory brainstem response (ABR) measurement is fairly quick and translational across species, but is limited by the need for anesthesia and a lack of perceptual assessment. The goal of this study was to develop a new method of hearing assessment utilizing prepulse inhibition (PPI) of the acoustic startle reflex, a commonly used tool that measures detection thresholds in awake animals, and can be performed on multiple animals simultaneously. We found that in control mice PPI audiometric functions are similar to both ABR and traditional operant conditioning audiograms. The hearing thresholds assessed with PPI audiometry in sound exposed mice were also similar to those detected by ABR thresholds one day after exposure. However, three months after exposure PPI threshold shifts were still evident at and near the frequency of exposure whereas ABR thresholds recovered to the pre-exposed level. In contrast, PPI audiometry and ABR wave one amplitudes detected similar losses. PPI audiometry provides a high throughput automated behavioral screening tool of hearing in awake animals. Overall, PPI audiometry and ABR assessments of the auditory system are robust techniques with distinct advantages and limitations, which when combined, can provide ample information about the functionality of the auditory system.

Methodological optimization of tinnitus assessment using prepulse inhibition of the acoustic startle reflex

Recently prepulse inhibition of the acoustic startle reflex (ASR) became a popular technique for tinnitus assessment in laboratory animals. This method confers a significant advantage over the previously used time-consuming behavioral approaches utilizing basic mechanisms of conditioning. Although this technique has been successfully used to assess tinnitus in different laboratory animals, many of the finer details of this methodology have not been described enough to be replicated, but are critical for tinnitus assessment. Here we provide detail description of key procedures and methodological issues that provide guidance for newcomers with the process of learning to correctly apply gap detection techniques for tinnitus assessment in laboratory animals. The major categories of these issues include: refinement of hardware for best performance, optimization of stimulus parameters, behavioral considerations, and identification of optimal strategies for data analysis. This article is part of a Special Issue entitled: Tinnitus Neuroscience.

Suppression of spontaneous firing in inferior colliculus neurons during sound processing

Spontaneous activity is a well-known neural phenomenon that occurs throughout the brain and is essential for normal development of auditory circuits and for processing of sounds. Spontaneous activity could interfere with sound processing by reducing the signal-to-noise ratio. Multiple studies have reported that spontaneous activity in auditory neurons can be suppressed by sound stimuli. The goal of this study was to determine the stimulus conditions that cause this suppression and to identify possible underlying mechanisms. Experiments were conducted in the inferior colliculus (IC) of awake little brown bats using extracellular and intracellular recording techniques. The majority of IC neurons (82%) fired spontaneously, with a median spontaneous firing rate of 6 spikes/s. After offset of a 4 ms sound, more than half of these neurons exhibited suppression of spontaneous firing that lasted hundreds of milliseconds. The duration of suppression increased with sound level. Intracellular recordings showed that a short (<50 ms) membrane hyperpolarization was often present during the beginning of suppression, but it was never observed during the remainder of the suppression. Beyond the initial 50 ms period, the absence of significant changes in input resistance during suppression suggests that suppression is presynaptic in origin. Namely, it may occur on presynaptic terminals and/or elsewhere on presynaptic neurons. Suppression of spontaneous firing may serve as a mechanism for enhancing signal-to-noise ratios during signal processing.

Timing of sound-evoked potentials and spike responses in the inferior colliculus of awake bats

Neurons in the inferior colliculus (IC), one of the major integrative centers of the auditory system, process acoustic information converging from almost all nuclei of the auditory brain stem. During this integration, excitatory and inhibitory inputs arrive to auditory neurons at different time delays. Result of this integration determines timing of IC neuron firing. In the mammalian IC, the range of the first spike latencies is very large (5-50 ms). At present, a contribution of excitatory and inhibitory inputs in controlling neurons' firing in the IC is still under debate. In the present study we assess the role of excitation and inhibition in determining first spike response latency in the IC. Postsynaptic responses were recorded to pure tones presented at neuron's characteristic frequency or to downward frequency modulated sweeps in awake bats. There are three main results emerging from the present study: (1) the most common response pattern in the IC is hyperpolarization followed by depolarization followed by hyperpolarization, (2) latencies of depolarizing or hyperpolarizing responses to tonal stimuli are short (3-7 ms) whereas the first spike latencies may vary to a great extent (4-26 ms) from one neuron to another, and (3) high threshold hyperpolarization preceded long latency spikes in IC neurons exhibiting paradoxical latency shift. Our data also show that the onset hyperpolarizing potentials in the IC have very small jitter (<100 micros) across repeated stimulus presentations. The results of this study suggest that inhibition, arriving earlier than excitation, may play a role as a mechanism for delaying the first spike latency in IC neurons.

Intracellular recording reveals temporal integration in inferior colliculus neurons of awake bats

The central nucleus of the inferior colliculus (IC) is a major integrative center in the central auditory system. It receives information from both the ascending and descending auditory pathways. To determine how single IC neurons integrate information over a wide range of sound frequencies and sound levels, we examined their intracellular responses to frequency-modulated (FM) sounds in awake little brown bats (Myotis lucifugus). Postsynaptic potentials were recorded in response to downward FM sweeps of the range typical for little brown bats (80-20 kHz) and to three FM subcomponents (80-60, 60-40, and 40-20 kHz). The majority of recorded neurons responded to the 80- to 20-kHz downward FM sweep with a complex response. In this response an initial hyperpolarization was followed by depolarization with or without spike followed by hyperpolarization. Intracellular recordings in response to three FM subcomponents revealed that these neurons receive excitatory and inhibitory inputs from a wide range of sound frequencies. One third of IC neurons performed nearly linear temporal summation across a wide range of sound frequencies, whereas two thirds of IC neurons exhibited nonlinear summation with different degrees of nonlinearity. Some IC neurons showed different latencies of postsynaptic potentials in response to different FM subcomponents. Often responses to the later FM subcomponent occurred before responses to the earlier ones. This phenomenon may be responsible for response selectivity of IC neurons to FM sweeps.

Stimulation rate influences frequency tuning characteristics of inferior colliculus neurons in the little brown bat, Myotis lucifugus

Previous studies of frequency selectivity have investigated unit's responses to tonal stimuli widely separated in time to minimize inter-stimulus interaction. The results of such studies are assumed to accurately portray the cell's frequency selectivity. The goal of the present study was to investigate the frequency tuning characteristics of neurons in the inferior colliculus (IC) of the little brown bat (Myotis lucifugus) to tone pulses presented at higher rates. Our results indicate that the frequency response properties of central auditory neurons at low stimulation rates do not necessarily reflect the units' frequency response properties to sounds presented at higher, more behaviorally relevant rates. Specifically, IC neurons often show greater frequency selectivity at higher stimulation rates, which presumably confers a greater perceptual frequency resolution.

Oscillation may play a role in time domain central auditory processing

To study how sound intensity altered the temporal response pattern of a unit, we recorded from 92 single neurons in the inferior colliculus (IC) of the little brown bat and investigated their firing patterns in response to brief tone pulses (2 msec duration) at the characteristic frequency of the unit over a wide dynamic range (10-90 dB sound pressure level). We found two unusual response characteristics at high sound levels in approximately one-third of the IC neurons investigated. For 16 IC neurons (17%), an increase in sound level not only elicited a shorter response latency and an increase in spike count but also transformed the firing pattern of the unit from phasic to periodic; this pattern was more pronounced at higher sound levels. The firing periodicity was unit specific, ranging from 1.3 to 6.7 msec. Twenty-seven IC neurons (29%) exhibited a longer response latency at higher sound levels compared with lower sound levels [i.e., paradoxical latency shift (PLS)]. The majority of this population showed a one or more quantum increase in latency when sound level was elevated. The quantum shift was also unit specific, ranging from 1.2 to 8.2 msec. We further investigated the firing patterns of 14 IC neurons showing PLS before, during, and after iontophoretic application of bicuculline. For 12 of these neurons, drug application abolished the PLS and transformed the firing patterns of the unit at high sound levels from phasic into sustained periodic discharges. Our results suggest that neural oscillation in combination with ordinary inhibition may be responsible for the creation of PLSs shown previously to be important for temporal information processing.

Temporal dynamics of acoustic stimuli enhance amplitude tuning of inferior colliculus neurons

Sounds in real-world situations seldom occur in isolation. In spite of this, most studies in the auditory system have employed sounds that serve to isolate physiological responses, namely, at low rates of stimulation. It is unclear, however, whether the basic response properties of a neuron derived thereof, such as its amplitude and frequency selectivities, are applicable to real-world situations where sounds occur in rapid succession. In the present study, we investigated one of the basic response properties of neurons in the bat inferior colliculus (IC), i.e., the rate-level function, to tone pulses in three different configurations: individual tone pulses of constant amplitude at different rates of stimulation, random-amplitude pulse trains, and dynamic-amplitude-modulated pulse trains the temporal pattern of which was similar to what bats encounter in a behavioral context. We reported that for the majority of IC neurons, amplitude selectivity to tone pulses was dependent on the rate of stimulation. In general, the selectivity was greater at high rates or in a behavioral context than at low rates. For a small population of IC neurons, however, the rate of stimulation had little or no effect on their rate-level functions. Thus for IC neurons, responses to sounds presented at low rates may or may not be used to predict the responses to the same stimuli presented at high rates or in a behavioral context. The possible neural mechanisms underlying the rate-dependent effects are discussed.

Cochleo- and tonotopic organization of the second auditory cortical area in the cat

The cochleo- and tonotopic organization of the second auditory area (AII) was investigated in cats anaesthetized with pentobarbital using a combination of macro- and microelectrode recording technique. The results obtained following electrical stimulation of the neural fibres innervating different regions of the organ of Corti indicate the existence of two complete representations of the cochlea in area AII: one in the dorsocaudal portion, the other in its ventrorostral portion. These two cortical representations of the cochlea differ in size and spatial orientation. The dorsocaudal projection area extends over a distance of 2.6-3.2 mm from the basal to the apical focus and is arc-shaped. The spatial orientation of cochlea representation within the dorsocaudal region of AII is similar to that described in AI, in that stimulation of the cochlea base results in maximal responses in the more rostral portion of AII and stimulation of the apex evokes cortical responses more caudally. The ventrorostral region within AII is smaller (1.4-2.5 mm length), and has the opposite cochleotopic orientation (base and apex stimulation represented caudally and rostrally, respectively). In both AII zones, there was a proportionally greater cortical representation of basilar membrane than of middle and apical portions. Although two distinct zones with the overall cochleotopic pattern described above were noted in all cats, their precise size and location considerably varied in different animals. Using microelectrode recordings, a cortical tonotopic organization can be observed that was consistent with and expanded on the earlier cochleotopic data. Within the dorsocaudal region of AII, neurons with higher best frequency responses were located in more rostral regions, while those with lower best frequencies were located caudally. An orderly progression of best frequency responses was noted as serial recordings carried out along the full extent of the representation. Neurons within the ventrorostral region of AII also displayed an orderly progression of best frequencies, but in the opposite direction, with higher best frequencies noted more caudally and lower best frequencies more rostrally.

Encoding of sound duration by neurons in the auditory cortex of the little brown bat, Myotis lucifugus

Responses of 117 single- or multi-units in the auditory cortex (AC) of bats (Myotis lucifugus) to tone bursts of different stimulus durations (1-400 ms) were studied over a wide range of stimulus intensities to determine how stimulus duration is represented in the AC. 36% of AC neurons responded more strongly to short stimulus durations showing short-pass duration response functions, 31% responded equally to all pulse durations (i.e., all-pass), 18% responded preferentially to stimuli having longer durations (i.e., long-pass), and 15% responded to a narrow range of stimulus durations (i.e., band-pass). Neurons showing long-pass and short-pass duration response functions were narrowly distributed within two horizontal slabs of the cortex, over the rostrocaudal extent of the AC. The effects of stimulus level on duration selectivity were evaluated for 17 AC neurons. For 65% of these units, an increase in stimulus intensity resulted in a progressive decrease in the best duration. In light of the unusual intensity-dependent duration responses of AC neurons, we hypothesized that the response selectivities of AC neurons is different from that in the brainstem. This hypothesis was validated by results of study of the duration response characteristics of single neurons in the inferior colliculus.

Peculiarities of inhibition in cat auditory cortex neurons evoked by tonal stimuli of various durations

The extra- and intracellular responses of 262 neurons in A1 to tones of best frequency with durations ranging from 10 ms to 1.2 min were studied acute experiments on ketamine-anesthetized cats. Following the generation of action potentials in response to the tone stimulus, inhibition of both the background and the auditory stimulus-evoked spike activity were observed in 91% of the investigated neurons. The duration of this inhibition corresponded to the stimulus duration. For the remaining neurons (9%) an inhibition of the stimulus-evoked spike activity alone was seen, also corresponding to the stimulus duration. Maximal inhibition of the spike activity occurred for the first 100-200 ms of the inhibitory response (the period which equalled the time of development of an IPSP in a cell). During this period of IPSP development, the membrane resistance of the neuron was reduced to 60-90% of its initial value. Varying the duration of the acoustic signal within a range of 10-200 ms was accompanied by a change in the IPSP duration and inhibition of the spike activity of the neuron. Whenever the tone lasted more than 200 ms, the membrane potential of the neuron was restored to the resting potential. However, during this period, the responsiveness of the neuron was lower than that initially observed. Measurement of the membrane resistance during the inhibitory pause that was not accompanied by hyperpolarization produced an index with an average 17% lower than the initial value for 87% of the neurons.(ABSTRACT TRUNCATED AT 250 WORDS)