Quantifying 2-factor phase relations in non-linear responses from low characteristic-frequency auditory-nerve fibers

Auditory-nerve excitation by two response factors that can be in antiphase has been hypothesized by Kiang (1990) on the basis of non-linear interference in responses to tones (Kiang et al., 1969). The general conditions for antiphasic responses and the relevance of the hypothesis for other auditory stimuli are unknown. Clarification was sought in a systematic modeling study of published data on level-dependent non-linear responses from low characteristics-frequency (CF) auditory-nerve fibers for a broad variety of acoustic stimuli. The MBPNL non-linear I/O model of cochlear frequency analysis (Goldstein, 1990), which incorporates the 2-factor hypothesis, was used to stimulate the reported non-linear phenomena. It was found that experiments with paired click stimuli (Goblick and Pfeiffer, 1969) and with octave-band complex tones (Horst et al., 1990), in addition to experiments with single clicks or tones, are sensitive to the phase difference between factors. Surprisingly, the paired-click transient responses require a quadrature phase, while the complex-tone steady-state responsesrequire an antiphase relation. The MBPNL model simulations of all low-CF data surveyed, for simple and complex stimuli, are consistent with a quadrature phase for transient responses and antiphase relation for steady-state responses. It is hypothesized that some adaptive, low-CF, cochlear mechanism, not described by the basic MBPNL model, produces a temporal transition of the ‘2-factor’ response from an initial quadrature relation (tip leading) to a final antiphase relation. New experimental and modeling research guided by this working hypothesis is proposed.     

Coherence analysis of scalp responses to amplitude-modulated tones

Magnitude squared coherence is a method of auditory evoked potential (AEP) frequency-domain analysis, measuring the degree to which the AEP is determined by the stimulus as a function of frequency. In 10 normal-hearing human subjects, scalp responses to an amplitude-modulated (AM) tone (500 Hz carrier modulated with a 40 Hz envelope) were recorded. Critical value criteria were utilized in the statistical analysis of coherence-intensity functions for determination of threshold responses. When compared with subjective determination of AEP threshold from time-domain waveforms, coherence analysis provided a significantly (p less than 0.001) more sensitive threshold measure. Coherence analysis provides information on the spectral content of the response, and allows for the objective determination of threshold which may potentially be utilized in a more expeditious threshold measurement scheme.     

Representation of voice pitch in discharge patterns of auditory-nerve fibers

Responses of populations of auditory-nerve fibers were measured for synthesized consonant-vowel stimuli. This paper explores the encoding of fundamental frequency (pitch) in these responses. Post-stimulus time (PST) histograms were computed from 25 ms segments of the spike trains. Discrete Fourier transforms with a 40 Hz resolution were computed from the histograms. Two representations of pitch are considered. The first representation is based on the pitch-related temporal properties of the speech signal. Histograms for individual units can show envelope modulations directly related to the pitch period. These modulations reflect the responses of these fibers to a number of stimulus harmonics near fiber CF. Responses of fibers near formant frequencies are dominated by a single large harmonic component, and thus show small or no pitch-related enveloped modulations. Envelope modulations are reduced in the presence of background noise. The second representation uses both temporal properties of auditory-nerve responses and cochlear place to encode the pitch-related harmonic structure of speech. As a measure of the response of the population of fibers to each harmonic of 40 Hz the magnitude of the component of the Fourier transform at that frequency was averaged across all fibers whose characteristic frequencies were within one-fourth octave of that harmonic. We call this measure the average localized synchronized rate (ALSR). The ALSR provides a good representation of stimulus spectrum, even in the presence of background noise. From the harmonic structure of the ALSR, we are able to extract the stimulus pitch frequency. The relationship of these two representations to pitch perception in both acoustic and electrical stimulation (via cochlear implants) is discussed.     

Representation of whispered vowels in discharge patterns of auditory-nerve fibers

We have previously shown that the spectra of speech sounds can be represented in the temporal patterns of discharge of populations of auditory-nerve fibers. These results were obtained using perfectly periodic stimuli, for which a temporal representation is straightforward. In order to see if our results could be generalized to nonperiodic stimuli, we have studied responses to a whispered vowel with formant frequencies typical of /ε/. The whispered vowel was generated by exciting a vocal tract model with noise; this signal was therefore aperiodic. Temporal patterns of responses to the vowel in populations of auditory-nerve fibers were analyzed using interval histograms. Fourier transforms of these histograms show large components at frequencies near the formant frequencies of the vowel. With these Fourier transform components as a measure of the temporal response, a temporal-place representation of the response of populations of fibers preserves the spectral features of the aperiodic vowel stimulus. Profiles of average rate versus characteristic frequency for fibers with spontaneous rates greater than $\text{1}\text{s}$ show little if any formant-related structure. On the other hand, such profiles for fibers with spontaneous rates less than $\text{1}\text{s}$ may show peaks in the region of the formants.     

Stimulus properties influencing the responses of inferior colliculus neurons to amplitude-modulated sounds

The temporal pattern of the responses of neurons in the inferior colliculus of the anesthetized rat were studied using continuous tone or noise carrier signals, amplitude modulated by pseudorandom noise. Period histograms of the responses, cross-correlated with the pseudorandom noise, gave an estimate of the unit's impulse responses to modulation. The amplitude-modulation rate transfer function (MTF) was obtained by Fourier transforming the correlograms. At sound levels within approximately 15 dB of the unit threshold, the MTFs were near lowpass functions between 6 and 200 Hz but became more bandpass-like as the intensity was increased. There was a steep decline in the response to modulation at modulation frequencies above 200 Hz for all stimulus intensities. For the bandpass-type MTFs the greatest modulation of the discharge pattern occurred at modulation frequencies between 10 and 200 Hz with a maximum in the distribution of MTF peak values between 100 and 120 Hz. There was no consistent relationship with characteristic frequency of either the position of the MTF peak or the high-frequency cutoff of the MTF. The cross-correlograms obtained at high stimulus intensities (30-60 dB above threshold) often showed a negative peak, representing a decrease in the probability of firing in response to intensity increments in the stimulus, and denoting a nonmonotonic rate-intensity function. The MTFs for units responding to amplitude-modulated broadband noise were often flatter in the low frequency region than those generated with tone carriers at corresponding intensities. For some units addition of a broadband noise background to the modulated tone changed the response characteristic of the MTF from bandpass to lowpass and shifted the MTF peak to a lower modulation frequency. The results demonstrate that although neurons in the inferior colliculus are selectively sensitive to the modulation frequency of dynamic stimuli, the response characteristics are not invariant, but instead are closely dependent on the conditions under which the modulation is presented.     

Steady-state evoked responses to sinusoidally amplitude-modulated sounds recorded in man

Steady-state potentials evoked in response to binaural, sinusoidally amplitude-modulated (AM) pure tones and broadband noise signals were recorded differentially from position F4 and the ipsilateral mastoid on the human scalp. The responses elicited by the AM stimuli were approximately periodic waveforms whose energy was predominantly at the modulation frequency of the stimulus. The magnitude of responses was between 0.1 and 4 microV for modulation frequencies between 2 and 400 Hz imposed on a 1-kHz carrier signal. The magnitude of the responses increased linearly with log modulation depth for low (4 Hz) and high (80 Hz) modulation rates. The response magnitude also increased linearly with the mean intensity of the sound for intensities up to 60 dB above the subject's pure tone threshold; at higher levels the response saturated. The relationship between response magnitude and modulation frequency (the modulation transfer function) was a lowpass function for both pure tone and broadband noise carrier signals. The modulation transfer functions were similar to those obtained from human psychophysical measurements where spectral cues are either unavailable or not used by the subject. The responses also contained a significant component at the second harmonic of the modulation frequency. The magnitude of this component was greatest at modulation rates between 5 and 20 Hz. The responses elicited by ipsilateral and contralateral monaural stimulation were approximately equal in magnitude, and binaural stimulation produced a potential 30% greater than the individual monaural responses. It is suggested that the evoked response represents the entrained neural activity to temporal amplitude fluctuations, and reflects the psychophysically measured performance of the auditory system for the detection and analysis of amplitude modulation.     

Spontaneous rates, thresholds and tuning of auditory-nerve fibers in the gerbil: comparisons to cat data

Characteristics of 245 auditory nerve fibers in eleven Mongolian gerbils are described in terms of spontaneous rates, thresholds, and tuning curves. The animals were reared in a low-noise environment and had similar hearing thresholds across frequency. Tuning curves were obtained with an algorithm developed to characterize the tuning of auditory fibers in cat, thereby allowing direct comparisons to published data from cat. The results demonstrate that basic similarities exist between gerbil and cat data, although some minor differences are also apparent. Tuning curve bandwidths, as measured 10 and 40 dB above the thresholds at the characteristic frequency (CF), follow trends found in cat data. Like cat, auditory nerve fibers in the gerbil have a range of spontaneous rates. In individual gerbils, fibers associated with low spontaneous rates have higher thresholds than do fibers of similar CF with high rates. Five of the eleven gerbils showed profiles of spontaneous rate across frequency reminiscent of those obtained from quiet-raised young cats. The profiles of the remaining gerbils tended to be compressed to a smaller range of spontaneous rates for characteristic frequencies above about five kHz, much like older cats with unknown noise histories. The cause of the spontaneous compression is not obvious. The correspondence between cat and gerbil with regard to spontaneous rate and CF threshold implies the presence of fundamental mechanisms that are common to mammalian auditory systems.     

Rapid adaptation of auditory-nerve fibers: fine structure at high stimulus intensities

Simple phenomenological models of rapid adaptation do not account for details in the response of auditory-nerve fibers to the onset of high-intensity sound stimulus. Simple refractory effects, in combination with higher-order discharge-history effects appear necessary to fully account for the oscillations and plateaus observed in post-stimulus time histograms.     

Auditory neurophonic responses to amplitude-modulated tones: transfer functions and forward masking

Two auditory neurophonic responses - one recorded from the scalp (frequency following response or FFR) and one from the auditory nerve (auditory nerve neurophonic or ANN) - were obtained following stimulation of the cat cochlea with amplitude-modulated (AM) high-frequency tones. The carrier frequencies varied between 2 and 30 kHz. The modulation frequencies varied between 400 and 3000 Hz. The AM responses were compared with pure-tone neurophonic responses. The AM response waveforms were found to have a similar spectral composition, similar rates of adaptation, and similar rates of recovery from forward masking as the comparable pure-tone responses. As with the pure-tone neurophonics, an unmodulated masking stimulus can produce prolonged depression of the probe response. The amount and duration of this depression is dependent upon the level and frequency of the masker. The frequency dependence of the depression is demonstrated by forward masked tuning curves which indicate that the AM responses arise from fiber populations which have restricted characteristic frequency distributions centered on the carrier frequency. Response amplitude as a function of stimulus level (I/O) functions, response amplitude as a function of carrier frequency (carrier transfer functions or CTF) and response amplitude as a function of modulation frequency (modulation transfer functions or MTF) were also measured. It was found that the I/O functions were saturating monotonic functions of stimulus intensity, CTFs were flat for carrier frequencies from 6 to 30 kHz, and MTFs were flat for modulation frequencies from 100 to 1500 Hz. These results are compared with similar data for single units and compound action potentials.     

Human auditory steady-state responses to amplitude-modulated tones: phase and latency measurements

Human auditory steady-state responses were recorded to four stimuli, with carrier frequencies (f(c)) of 750, 1500, 3000 and 6000 Hz, presented simultaneously at 60 dB SPL. Each carrier frequency was modulated by a specific modulation frequency (f(m)) of 80.6, 85.5, 90.3 or 95.2 Hz. By using four different recording conditions we obtained responses for all permutations of f(m) and f(c). The phase delays (P) of the responses were unwrapped and converted to latency (L) using the equation: L=P/(360xf(m)). The number of cycles of the stimulus that occurred prior to the recorded response was estimated by analyzing the effect of modulation frequency on the responses. These calculations provided latencies of 20.7, 17.7, 16.1 and 16.1 ms for carrier frequencies 750, 1500, 3000 and 6000 Hz. This latency difference of about 4.5 ms between low and high carrier frequencies remained constant over many different manipulations of the stimuli: faster modulation rates (150-190 Hz), binaural rather than monaural presentation, different intensities, stimuli presented alone or in conjunction with other stimuli, and modulation frequencies that were separated by as little as 0.24 Hz. This frequency-related delay is greater than that measured using transient evoked potentials, most likely because of differences in how transient and steady-state responses are generated and how their latencies are determined.     

Rate/level functions of auditory-nerve fibers in young and quiet-aged gerbils

Steady-state rate/level functions of single auditory-nerve fibers to characteristic frequency (CF) tone bursts were measured in quiet-aged (35-37 months) and young control (4-7 months) gerbils. Rate/level functions of aged gerbils are different from those of young controls in that the thresholds are shifted to higher sound levels, but otherwise the shapes of the aged and young rate/level functions are similar. Specifically, there is little difference in the slope of the dynamic range portion of the rate/level functions when comparing aged gerbils to young controls. This is in contrast to whole-nerve input/output (I/O) functions of aged gerbils, which exhibit slopes that are less steep than those of the young controls (Hellstrom and Schmiedt, 1990b). Thus, it is likely that the deterioration of the CAP I/O function in aged animals is not due to a deterioration of rate/level functions in single units, but rather to other factors such as spiral ganglion cell degeneration or a loss of synchrony.     

Towards a unifying basis of auditory thresholds: distributions of the first-spike latencies of auditory-nerve fibers

Detecting sounds in quiet is the simplest task performed by the auditory system, but the neural mechanisms underlying perceptual detection thresholds for sounds in quiet are still not understood. Heil and Neubauer [Heil, P., Neubauer, H., 2003. A unifying basis of auditory thresholds based on temporal summation. Proc. Natl. Acad. Sci. USA 100, 6151-6156] have provided evidence for a simple probabilistic model according to which the stimulus, at any point in time, has a certain probability of exceeding threshold and being detected. Consequently, as stimulus duration increases, the cumulative probability of detection events increases, performance improves, and threshold amplitude decreases. The origin of these processes was traced to the first synapse in the auditory system, between the inner hair cell and the afferent auditory-nerve fiber (ANF). Here we provide further support for this probabilistic "continuous-look" model. It is derived from analyses of the distributions of the latencies of the first spikes of cat ANFs with very low spontaneous discharge rates to tones of different amplitudes. The first spikes in these fibers can be considered detection events. We show that, as predicted, the distributions can be explained by the joint probability of the occurrence of three independent sub-events, where the probability of each of those occurring is proportional to the low-pass filtered stimulus amplitude. The "temporal integration" functions of individual ANFs, derived from their first-spike latencies, are remarkably similar in shape to "temporal integration" functions, which relate threshold sound pressure level at the perceptual level to stimulus duration. This further supports a close link between the mechanisms determining the timing of the first (and other) evoked spikes at the level of the auditory nerve and detection thresholds at the perceptual level. The possible origin and some functional consequences of the expansive power-law non-linearity are discussed.     

Response modulation of auditory-nerve fibers by am stimuli: effects of average intensity

Auditory-nerve responses were obtained for characteristic frequency tones which were amplitude modulated by sinusoids. Response modulation (RM) was determined from folded histograms which were synchronized to the modulating wave form. As the average intensity increased from threshold, RM increased to a maximum and then decreased, and the shape of the RM function resembled that described previously for incremental responses. However, unlike the latter, the RM function could not he predicted directly from the steady-state rate-intensity function. In general, the maximum RM occurred at a higher intensity than predicted, and RM occurred over a wider range of average intensities than predicted. The results are interpreted as reflecting a dynamic response characteristic with an operating range that exceeds that determined from the steady-state rate-intensity function.     

Evidence for expansive power functions in the generation of the discharges of 'low- and medium-spontaneous' auditory-nerve fibers

Re-analysis of data from Geisler et al. [J. Acoust. Soc. Am. 77, 1102-1109, 1985] indicates that the slopes of the intensity versus discharge-rate curves of auditory nerve (AN) fibers decrease systematically with increasing spontaneous discharge rate. For 'high-spontaneous' fibers, the slope is usually less than 0.5 dB/dB, while for 'low-spontaneous' fibers the slopes reach values greater than 4.0 dB/dB. A two-stage model accounts for this behavior. The first stage is a static non-linearity based on the measured intensity-voltage characteristic of inner hair cells. The second stage, representing action-potential generation, is linear for high-spontaneous fibers, but a squaring function for low- and medium-spontaneous fibers. The output of the model displays realistic slopes for its various intensity-rate curves. There are suggestions that a nonlinearity of still higher power is needed to simulate accurately the behavior of AN fibers having the lowest spontaneous rates (less than 0.1/s). The model also accounts for other observed differences between the discharge patterns of the different fiber classes.     

Activity patterns of primary auditory-nerve fibres in chickens: development of fundamental properties.

We have examined the activity patterns of single auditory-nerve fibers in the chicken and tested for possible changes during post-hatching development. For this purpose, we recorded from fibres in the cochlear ganglion of chickens of two age groups (about P2 and P21) and investigated their spontaneous and sound-evoked activity patterns under nembutal-chloralhydrate anaesthesia. The spontaneous activity of primary auditory neurones was irregular, the average rates were between 20.5 (P2) and 23 (P21) spikes/s. Many low-frequency fibres from both age groups showed preferred intervals in their spontaneous activity. Tuning characteristics, including the range of characteristic frequencies, the presence of primary and two-tone suppression, the slopes of tuning-curve flanks and Q10dB values were similar to those previously reported for the starling and were statistically indistinguishable between the two age groups. However, there was a difference in fibre thresholds at the highest frequencies. Systematic differences were also present between the two age groups with regard to some characteristics of the rate-intensity functions. These data indicate that whereas the tuning properties of primary auditory fibres of the chicken cochlea are mature as early as post-hatching day 2, the intensity functions are not.     

Cochlear microphonic responses of the peripheral auditory system to frequency-varying signals

Cochlear partition displacement responses to rising and falling frequency sweeps were inferred from cochlear microphonic potentials recorded from three basal turn locations in the guinea pig cochlea. Relative phase measures of microphonic potentials recorded from the three locations suggested that displacements of the partition toward either scala vestibuli or scala tympani occurred closely together in time for rising sweeps and were dispersed in time for falling sweeps. These differences in peripheral response patterns to sweeps may explain, in part, asymmetric neural discharges elicited from higher neural centers.     

Basic properties of auditory-nerve responses from a "simple' ear: the basilar papilla of the frog

Spike discharges initiated by mammalian inner hair cells are produced by a complicated system involving both mechanical and neural components that normally operate in a bi-directional configuration involving multiple feedback loops. In contrast, the frog basilar papilla has the equivalent of inner hair cells, but lacks outer hair cells; it has no efferent system, and no basilar membrane. This suggests that the frog basilar papilla lacks some of the mechanical and neural feedback paths characteristic of the mammalian system. Detailed measurements of tuning curves, spontaneous activity and responses to tones an clicks reveal large parametric differences between frog and mammals in spontaneous rate, absolute refractory time, long-term adaptation and phase locking. Responses to tone bursts are qualitatively similar, but parametrically quite different. More focused examinations of these effects will be able to exploit the differences in adaptation to long- versus short-duration stimuli could be caused by depletion of afferent neurotransmitter or by activation of feedback loops involving the efferent system. In the basilar papilla, any differences in adaptation must result from changes in the afferent pathway alone.     

Effects of contralateral sound on auditory-nerve responses. I. Contributions of cochlear efferents

The response properties of single auditory-nerve fibers in barbiturate-anaesthetized cats were recorded with and without simultaneous presentation of sound to the contralateral ear. The tendons to the middle ear muscles on both sides were cut before all experiments, and contralateral stimuli were restricted to levels below the threshold for crosstalk to the ipsilateral ear. Contralateral tones and broad-band noise were found to suppress the responses of auditory-nerve afferents to ipsilateral tones at their characteristic frequency (CF), but not to tones off CF. The suppression due to contralateral sound required approximately 100-200 ms to develop and to decay. When the contralateral stimuli were tones at the CF, the strongest suppression was observed in low- and medium-spontaneous-rate units with CFs between 1 and 2 kHz. The suppressive effect of contralateral sound completely disappeared immediately after severing the entire olivocochlear bundle (OCB) within the internal auditory meatus. the completeness of the OCB cuts was assessed histologically. Most of the suppressive effect remained after lesions to the OCB in the floor of the IVth ventricle which eliminated the crossed olivocochlear projection but spared most of the uncrossed projection. It is argued that this suppressive effect of contralateral sound is mediated by the uncrossed olivocochlear efferents to the outer hair cells.     

Effects of electrical stimulation of efferent olivocochlear neurons on cat auditory-nerve fibers. II. Spontaneous rate

In order to increase our understanding of cochlear mechanisms, we measured changes in the rate of spontaneous firing (SR) of single auditory-nerve fibers in response to the stimulation of medial olivocochlear efferents in cats. During the first second of efferent stimulation, SR was depressed by up to 35%, except in one very sensitive animal in which depressions up to 80% were found. With data from this aberrant cat excluded, the SR depression, on the average, increased as auditory-nerve fiber sensitivity increased, increased as the original SR decreased (data were not obtained for SRs less than two spikes/sec), and had a broad maximum at CFs of about 10 kHz. After the efferent stimulation was turned off, there was an "overshoot" in which the SR increased past the original rate in some fibers. The "overshoot" was larger for fibers with lower SRs and for fibers which showed larger "adaptation" in the efferent-induced depression of SR. The data on SR depression during efferent stimulation are consistent with two hypotheses: (1) that the stronger than usual efferent suppression of "spontaneous" rate found in some very sensitive fibers occurs because the "spontaneous" firing was, in part, a response to sound, and (2) that "true spontaneous" firing is reduced by the efferent-induced hyperpolarization of outer hair cells (OHCs) being electrically coupled through the endocochlear potential to inner hair cells (IHCs). It is suggested that (1) the efferent-induced suppression of "true spontaneous" activity is largest at CFs near 10 kHz because this CF region receives the greatest OHC innervation from medial efferents and the efferent-induced change in OHCs is electrically coupled to IHCs, whereas (2) the efferent suppression of responses to sound is largest at lower CFs because the efferent endings on OHCs act to decrease the motion of the basilar membrane and this change is propagated apically from the active efferent synapses on OHCs.