The frequency selectivity of information-processing channels in the tactile sensory system

The frequency selectivity of the P, NP I, and NP II channels of the four-channel model of mechanoreception for glabrous skin was measured psychophysically by an adaptation tuning curve procedure. The results substantially extend the frequency range over which the frequency selectivity of these channels is known and further confirm the hypothesis that the input stage of each of these channels consists of specific sensory nerve fibers and associated receptors. Specifically, the frequency characteristics of Pacinian nerve fibers, rapidly adapting (RA) nerve fibers, and slowly adapting Type II (SA II) nerve fibers were found to be the peripheral neurophysiological correlates of the P, NP I, and NP II channels, respectively. The finding that the tuning characteristic for a test stimulus of 250 Hz delivered through a small (0.008 cm2) contactor depended dramatically on the duration of the test stimulus whereas the detection threshold did not, provides new evidence in support of the hypothesis that separate NP II and P channels exist.     

Temporal gap detection in tactile channels

The ability of observers to detect temporal gaps in bursts of sinusoids or bursts of band-limited noise was measured to assess the temporal acuity of Pacinian (P) and non-Pacinian (NP) tactile information processing channels. The P channel was isolated by delivering high frequency sinusoids or high frequency noise through a large 1.5-cm² contactor to the thenar eminence. The NP channels were isolated from the P channel by delivering these stimuli as well as stimuli with lower frequencies through a small 0.01-cm² contactor to the same site. Gap detection thresholds were higher for gaps in noise than for gaps in sinusoids but did not differ among conditions designed to isolate P and NP channels. The finding that temporal acuity does not differ among channels supports the hypothesis that, after termination of a stimulus, the P and NP channels exhibit the same amount of neural persistence. Also consistent with this hypothesis are the earlier findings that the enhancement of the sensation magnitude of a stimulus by a prior stimulus (Verrillo and Gescheider, Percept Psychophys 18: 128–136, 1975) and the duration of sensation after the termination of a stimulus (Gescheider et al., J Acoust Soc Am 91: 1690–1696, 1992) are independent of stimulus frequency. One important implication of this hypothesis, if true, is that the presence of temporal summation in the P channel and its absence in the NP channels, results, not from the lack of neural persistence in the NP channels, but instead, in marked contrast to the P channel, from the lack of a mechanism for integrating persistent neural activity over time.     

The Effects of Aging on Information-Processing Channels in the Sense of Touch: I. Absolute Sensitivity

Thresholds for detecting vibrotactile signals of variable frequency applied to the thenar eminence of the hand by small and large contactors were measured in subjects ranging in age from 10 to 89 years. Thresholds were found to increase as a function of age, but the rate of increase was greater after than before the age of 65 years. The rate of loss of vibrotactile sensitivity was substantially greater in the P channel (mediated by Pacinian corpuscles) than in the NP I channel (mediated by rapidly adapting fibers), the NP II channel (mediated by slowly adapting type II fibers), or the NP HI channel (mediated by slowly adapting type I fibers). Women were frequently found to have greater sensitivity than men.     

Temporal gap detection in tactile channels

The ability of observers to detect temporal gaps in bursts of sinusoids or bursts of band-limited noise was measured to assess the temporal acuity of Pacinian (P) and non-Pacinian (NP) tactile information processing channels. The P channel was isolated by delivering high frequency sinusoids or high frequency noise through a large 1.5-cm2 contactor to the thenar eminence. The NP channels were isolated from the P channel by delivering these stimuli as well as stimuli with lower frequencies through a small 0.01-cm2 contactor to the same site. Gap detection thresholds were higher for gaps in noise than for gaps in sinusoids but did not differ among conditions designed to isolate P and NP channels. The finding that temporal acuity does not differ among channels supports the hypothesis that, after termination of a stimulus, the P and NP channels exhibit the same amount of neural persistence. Also consistent with this hypothesis are the earlier findings that the enhancement of the sensation magnitude of a stimulus by a prior stimulus (Verrillo and Gescheider, Percept Psychophys 18: 128-136, 1975) and the duration of sensation after the termination of a stimulus (Gescheider et al., J Acoust Soc Am 91: 1690-1696, 1992) are independent of stimulus frequency. One important implication of this hypothesis, if true, is that the presence of temporal summation in the P channel and its absence in the NP channels, results, not from the lack of neural persistence in the NP channels, but instead, in marked contrast to the P channel, from the lack of a mechanism for integrating persistent neural activity over time.     

Hairy Skin: Psychophysical Channels and Their Physiological Substrates

Experiments were conducted in which threshold-frequency characteristics were measured on the hairy skin of the forearm of human observers. Thresholds were measured with two stimulus probe areas (2.9 and 0.008 cm²) at three skin-surface temperatures (15d`, 30d`, and 40d`C). The results suggest that whereas glabrous skin uses four distinct channels of information, only three channels may be involved in mediating the sense of touch for hairy skin. The three channels are defined as P_h (Pacinian, hairy skin), NP_{h low} (non-Pacinian, hairy skin, low frequencies) and NP_{h mid} (non-Pacinian, hairy skin, middle frequencies). In addition, it is proposed that the neural substrates for the three psychophysically characterized channels are, respectively, the Pacinian corpuscle (PC) nerve fibers, the slowly adapting type II (SAII) fibers, and the rapidly adapting (RA) fibers.     

Pattern-specific contrast threshold elevation

Visual channels are defined psychophysically; stimuli that interact share information in the same channel, and those that do not interact are processed in different channels. Channels are often investigated by means of adaptation to one stimulus, testing contrast threshold elevation with one (or more) others. Much recent work has tested the tuning of channels for orientation and spatial frequency, using simple line gratings. This study examined the pattern-specificity of such adaptation, testing the hypothesis that the fundamental operators of the Lie Transformation Group Theory of Neuropsychology (LTG/NP) define psychophysical channels. In Experiment I the three basic pattern pairs of LTG/NP were used as adaptation and test stimuli in a conventional contrast threshold-elevation experiment. Threshold elevation was pattern-specific, thus supporting the hypothesis. In subsequent experiments various 'fractured' patterns, and patterns generated by combinations of Lie operators were used both for adaptation and test. The results were mixed; some supported the original hypothesis, but many did not. Relations between local contour orientations in adaptation and test patterns could explain some results, but not all. The hypothesis that adaptation occurs to the oriented spatial-frequency components of the test patterns, on the other hand, gave a good fit to the data. It is concluded that there is pattern-specificity in contrast threshold elevation, but it is a form of specificity that can be explained without recourse to a model of geometrical pattern processing, at least for the simple patterns used here.     

Time Course and Action Spectrum of Vibrotactile Adaptation

In a series of experiments designed to explore the processes underlying adaptation of the sense of flutter-vibration, vibrotactile threshold was measured on the pad of the index finger, using Békésy tracking. Unadapted thresholds were first measured, for a number of frequencies (4-90 Hz) and contactor sizes (1-8 mm diameter). As expected, these measurements indicated the presence of (1) a Pacinian system possessing spatial summation and increasing in sensitivity, as frequency was raised, at the rate of 12 dB/octave; and (2) a non-Pacinian system showing little spatial summation, and with a frequency characteristic matching that of the NP I mechanism of Bolanowski et al. (1988). These baseline data of Experiment 1 guided the selection of stimulus parameters for subsequent experiments, in which threshold for a test stimulus was measured before, during, and after periods of vibrotactile adaptation.

In Experiment 2, test stimuli of 10 H_z and 50 H_z were combined factorially with 30-dB SL adapting stimuli of the same two frequencies. When the test stimulus was 10 Hz, the two adapting frequencies were equally effective in raising threshold; however, when the 50-Hz test stimulus was used, the 50-Hz adapting stimulus raised threshold by a greater amount than did the 10-Hz adapter. These results confirm on the finger the independence of adaptation in Pacinian and non-Pacinian channels, a result previously established on the thenar by other workers. For all four frequency combinations, threshold rose exponentially with a time constant of 1.5-2 min.

In Experiment 3, an action spectrum was determined, showing the adapting amplitude needed at each of a series of frequencies to raise the threshold of a 10-Hz stimulus by 10 dB; this spectrum was essentially flat from 30 to 90 Hz. The results, taken in conjunction with what is known about rapidly adapting cutaneous mechanoreceptors, imply that the effectiveness of an adapting stimulus is not determined solely by the amount of activity it generates in first-order afferents.     

Time course and action spectrum of vibrotactile adaptation

In a series of experiments designed to explore the processes underlying adaptation of the sense of flutter-vibration, vibrotactile threshold was measured on the pad of the index finger, using Békésy tracking. Unadapted thresholds were first measured, for a number of frequencies (4-90 Hz) and contactor sizes (1-8 mm diameter). As expected, these measurements indicated the presence of (1) a Pacinian system possessing spatial summation and increasing in sensitivity, as frequency was raised, at the rate of 12 dB/octave; and (2) a non-Pacinian system showing little spatial summation, and with a frequency characteristic matching that of the NP I mechanism of Bolanowski et al. (1988). These baseline data of Experiment 1 guided the selection of stimulus parameters for subsequent experiments, in which threshold for a test stimulus was measured before, during, and after periods of vibrotactile adaptation. In Experiment 2, test stimuli of 10 Hz and 50 Hz were combined factorially with 30-dB SL adapting stimuli of the same two frequencies. When the test stimulus was 10 Hz, the two adapting frequencies were equally effective in raising threshold; however, when the 50-Hz test stimulus was used, the 50-Hz adapting stimulus raised threshold by a greater amount than did the 10-Hz adapter. These results confirm on the finger the independence of adaptation in Pacinian and non-Pacinian channels, a result previously established on the thenar by other workers. For all four frequency combinations, threshold rose exponentially with a time constant of 1.5-2 min. In Experiment 3, an action spectrum was determined, showing the adapting amplitude needed at each of a series of frequencies to raise the threshold of a 10-Hz stimulus by 10 dB; this spectrum was essentially flat from 30 to 90 Hz. The results, taken in conjunction with what is known about rapidly adapting cutaneous mechanoreceptors, imply that the effectiveness of an adapting stimulus is not determined solely by the amount of activity it generates in first-order afferents.     

Vibrotactile thresholds of the Non-Pacinian I channel: I. Methodological issues

Thresholds of the Non-Pacinian I (NP I) channel were measured using a two-interval forced-choice paradigm, a technique independent of the subject's criterion. The studies were performed using the terminal phalanx of the human middle finger with a 40-Hz vibratory stimulus. Unlike most of the previous experiments performed in our laboratory, a contactor surround was not used. This was done to enable comparison with population models of mechanoreceptive fibers in the literature. Since the Pacinian (P) channel and NP I channel have similar vibrotactile thresholds at 40?Hz, a forward-masking procedure was used to elevate the thresholds of the P channel with respect to the NP I channel. While it has been established that the Pacinian fibers are entrained at high stimulus levels, the P channel can be perceptually masked using a 250-Hz stimulus presented prior to the 40-Hz test stimulus. The masking functions were found to be approximately linear on log-log axes and the threshold shifts were found to increase as the masking-stimulus levels increased. The results are discussed in relation to previous studies that were performed at various stimulation sites by using a contactor surround or not. A companion paper presents the variation of NP I-channel thresholds, measured using the methods described herein, and addresses the effects of stimulation along the proximo-distal axis of the phalanx. The companion paper also discusses the predictions of a computational model, recently proposed, in light of the empirical results presented.     

A four-channel analysis of the tactile sensitivity of the fingertip: frequency selectivity,spatial summation,and temporal summation

Thresholds were measured for the detection of vibratory stimuli of variable frequency and duration applied to the index fingertip and thenar eminence through contactors of different sizes. The effects of stimulus frequency could be accounted for by the frequency characteristics of the Pacinian (P), non-Pacinian (NP) I, and NP III channels previously determined for the thenar eminence (Bolanowski et al., J Acoust Soc Am 84 : 1680-1694, 1988; Gescheider et al., Somatosens Mot Res 18: 191- 201, 2001). The effect of changing stimulus duration was also essentially identical for both sites, demonstrating the same amount of temporal summation in the P channel. Although the effect of changing stimulus frequency and changing stimulus duration did not differ for the two sites, the effect of varying the size of the stimulus was significantly greater for the thenar eminence than for the fingertip. The attenuated amount of spatial summation on the fingertip was interpreted as an indication that the mechanism of spatial summation consists of the operations of both neural integration and probability summation.     

A four-channel analysis of the tactile sensitivity of the fingertip: frequency selectivity,spatial summation,and temporal summation

Thresholds were measured for the detection of vibratory stimuli of variable frequency and duration applied to the index fingertip and thenar eminence through contactors of different sizes. The effects of stimulus frequency could be accounted for by the frequency characteristics of the Pacinian (P), non-Pacinian (NP) I, and NP III channels previously determined for the thenar eminence (Bolanowski et al., J Acoust Soc Am 84: 1680-1694, 1988; Gescheider et al., Somatosens Mot Res 18: 191-201, 2001). The effect of changing stimulus duration was also essentially identical for both sites, demonstrating the same amount of temporal summation in the P channel. Although the effect of changing stimulus frequency and changing stimulus duration did not differ for the two sites, the effect of varying the size of the stimulus was significantly greater for the thenar eminence than for the fingertip. The attenuated amount of spatial summation on the fingertip was interpreted as an indication that the mechanism of spatial summation consists of the operations of both neural integration and probability summation.     

Vibrotactile thresholds of the Non-Pacinian I channel: I. Methodological issues

Thresholds of the Non-Pacinian I (NP I) channel were measured using a two-interval forced-choice paradigm, a technique independent of the subject's criterion. The studies were performed using the terminal phalanx of the human middle finger with a 40-Hz vibratory stimulus. Unlike most of the previous experiments performed in our laboratory, a contactor surround was not used. This was done to enable comparison with population models of mechanoreceptive fibers in the literature. Since the Pacinian (P) channel and NP I channel have similar vibrotactile thresholds at 40?Hz, a forward-masking procedure was used to elevate the thresholds of the P channel with respect to the NP I channel. While it has been established that the Pacinian fibers are entrained at high stimulus levels, the P channel can be perceptually masked using a 250-Hz stimulus presented prior to the 40-Hz test stimulus. The masking functions were found to be approximately linear on log-log axes and the threshold shifts were found to increase as the masking-stimulus levels increased. The results are discussed in relation to previous studies that were performed at various stimulation sites by using a contactor surround or not. A companion paper presents the variation of NP I-channel thresholds, measured using the methods described herein, and addresses the effects of stimulation along the proximo-distal axis of the phalanx. The companion paper also discusses the predictions of a computational model, recently proposed, in light of the empirical results presented.     

Frequency responses of cat rapidly adapting mechanoreceptive fibers

The frequency responses of 11 rapidly adapting (RA) fibers in cat were studied by representing the average firing rate as a function of sinusoidal stimulus amplitude and stimulus frequency. Specifically, rate-intensity functions at different stimulation frequencies were fitted by four-parameter (a0, a1, a2, a3), piece-wise linear functions using nonlinear regression (n = 59; R2 > 0.877). Rate-intensity functions at intermediate frequencies were found by linear interpolation. The result of this analysis is rate-amplitude-frequency functions plotted as two-dimensional surfaces. The surfaces consist of five regions separated and sufficiently defined by four space curves. At 14 different frequencies, the statistical distribution of each rate-intensity-function parameter could be approximated by a particular lognormal distribution (n = 56; R2 > 0.796). The Kolmogorov-Smirnov test fails to reject this hypothesis for each combination of frequency and parameter (56 tests; p > 0.39). Therefore, at a given frequency, the variation of the parameters can be represented by lognormal distributions with specific means and standard deviations. Responses of six RA fibers, which are different from the data-set used for modeling, were compared with the stochastic model at different frequencies. The parameters of those fibers were tested against the null hypotheses that they were sampled from the particular parameter distributions dictated by the model. The Kolmogorov-Smirnov test fails to reject all the hypotheses at the alpha = 0.05 level (44 tests). At the alpha = 0.10 level, only a few test parameters were found to be departing from the model (a0 and a1 at 5 Hz; a2 at 20 Hz; a2 and a3 at 50 Hz). The remaining test parameters could be accurately described by the model. Having confirmed the validity of the model, the logarithmic means and the logarithmic standard deviations of the lognormally distributed rate-intensity-function parameters were estimated in the frequency range of 4-200 Hz. The rate-amplitude-frequency surfaces sampled from the established stochastic model completely characterize the rate responses of RA fibers to sinusoidal stimuli and are superior to tuning curves which require selecting criterion responses. The current rate-response model is promising for future computational work, especially on population modeling.     

Asymmetric response properties of rapidly adapting mechanoreceptive fibers in the rat glabrous skin

Previous histological and neurophysiological studies have shown that the innervation density of rapidly adapting (RA) mechanoreceptive fibers increases towards the fingertip. Since the psychophysical detection threshold depends on the contribution of several RA fibers, a high innervation density would imply lower thresholds. However, our previous human study showed that psychophysical detection thresholds for the Non-Pacinian I channel mediated by RA fibers do not improve towards the fingertip. By recording single-unit spike activity from rat RA fibers, here we tested the hypothesis that the responsiveness of RA fibers is asymmetric in the proximo-distal axis which may counterbalance the effects of innervation density. RA fibers (n?=?32) innervating the digital glabrous skin of rat hind paw were stimulated with 40-Hz sinusoidal mechanical bursts at five different stimulus locations relative to the receptive field (RF) center (two distal, one RF center, two proximal). Different contactor sizes (area: 0.39, 1.63, 2.96?mm²) were used. Rate-intensity functions were constructed based on average firing rates, and the absolute spike threshold and the entrainment threshold were obtained for each RA fiber. Thresholds for proximal stimulus locations were found to be significantly higher than those for distal stimulus locations, which suggests that the mechanical stimulus is transmitted better towards the proximal direction. The effect of contactor size was not significant. Mechanical impedance of the rat digital glabrous skin was further measured and a lumped-parameter model was proposed to interpret the relationship between the asymmetric response properties of RA fibers and the mechanical properties of the skin.     

Distribution of the intensity-characteristic parameters of cat rapidly adapting mechanoreceptive fibers

Modeling population responses of nerve fibers requires statistical characterization of fiber-response properties. The rate/intensity characteristics of cat rapidly adapting (RA) fibers were fitted by four-parameter, piece-wise linear functions using nonlinear regression (n = 14; R2 > 0.958). The parameters were tested against the null hypothesis that they are log normally distributed. The test fail to reject this hypothesis (Kolmogorov-Smirnov p>0.477). However, a significant statistical difference was found between the specific lognormal distributions obtained from monkey (Johnson, J Neurophysiol 37: 48-72, 1974) and cat for all four parameters (Kolmogorov-Smirnov, p<0.0075, p<0.05, p<0.0001, p<0.00007). Although the stimulus contactor size was not the same in monkey and cat studies, the differences between monkey and cat fibers are attributed to anatomical differences in the glabrous sin of both species. Modeling studies suggest that the absolute firing thresholds of RA fibers have a right-skewed distribution because of the anatomical constraints present in both species' skin. Meissner corpuscles, which are the sensory end-organs of RA fibers, are likely to be found deeper in the skin within dermal papilla, therefore, the thresholds can be elevated. However, the thresholds are bounded at lower end, probably due to the epidermal junction that acts as a superficial mechanical barrier for these corpuscles.     

Adaptation in auditory hair cells

The narrow stimulus limits of hair cell transduction, equivalent to a total excursion of about 100nm at the tip of the hair bundle, demand tight regulation of the mechanical input to ensure that the mechanoelectrical transducer (MET) channels operate in their linear range. This control is provided by multiple components of Ca(2+)-dependent adaptation. A slow mechanism limits the mechanical stimulus through the action of one or more unconventional myosins. There is also a fast, sub-millisecond, Ca(2+) regulation of the MET channel, which can generate resonance and confer tuning on transduction. Changing the conductance or kinetics of the MET channels can vary their resonant frequency. The tuning information conveyed in transduction may combine with the somatic motility of outer hair cells to produce an active process that supplies amplification and augments frequency selectivity in the mammalian cochlea.     

Frequency responses of cat rapidly adapting mechanoreceptive fibers

The frequency responses of 11 rapidly adapting (RA) fibers in cat were studied by representing the average firing rate as a function of sinusoidal stimulus amplitude and stimulus frequency. Specifically, rate-intensity functions at different stimulation frequencies were fitted by four-parameter (a₀, a₁, a₂, a₃), piece-wise linear functions using nonlinear regression (n = 59; R² > 0.877). Rate-intensity functions at intermediate frequencies were found by linear interpolation. The result of this analysis is rate–amplitude–frequency functions plotted as two-dimensional surfaces. The surfaces consist of five regions separated and sufficiently defined by four space curves. At 14 different frequencies, the statistical distribution of each rate-intensity-function parameter could be approximated by a particular lognormal distribution (n = 56; R² > 0.796). The Kolmogorov–Smirnov test fails to reject this hypothesis for each combination of frequency and parameter (56 tests; p > 0.39). Therefore, at a given frequency, the variation of the parameters can be represented by lognormal distributions with specific means and standard deviations. Responses of six RA fibers, which are different from the data-set used for modeling, were compared with the stochastic model at different frequencies. The parameters of those fibers were tested against the null hypotheses that they were sampled from the particular parameter distributions dictated by the model. The Kolmogorov–Smirnov test fails to reject all the hypotheses at the α = 0.05 level (44 tests). At the α = 0.10 level, only a few test parameters were found to be departing from the model (a₀ and a₁ at 5?Hz; a₂ at 20?Hz; a₂ and a₃ at 50?Hz). The remaining test parameters could be accurately described by the model. Having confirmed the validity of the model, the logarithmic means and the logarithmic standard deviations of the lognormally distributed rate-intensity-function parameters were estimated in the frequency range of 4–200?Hz. The rate–amplitude–frequency surfaces sampled from the established stochastic model completely characterize the rate responses of RA fibers to sinusoidal stimuli and are superior to tuning curves which require selecting criterion responses. The current rate-response model is promising for future computational work, especially on population modeling.     

Signal processing at mammalian carotid body chemoreceptors

Mammalian carotid bodies are richly vascularized chemosensory organs that sense blood levels of O₂, CO₂/H⁺, and glucose and maintain homeostatic regulation of these levels via the reflex control of ventilation. Carotid bodies consist of innervated clusters of type I (or glomus) cells in intimate association with glial-like type II cells. Carotid bodies make afferent connections with fibers from sensory neurons in the petrosal ganglia and receive efferent inhibitory innervation from parasympathetic neurons located in the carotid sinus and glossopharyngeal nerves. There are synapses between type I (chemosensory) cells and petrosal afferent terminals, as well as between neighboring type I cells. There is a broad array of neurotransmitters and neuromodulators and their ionotropic and metabotropic receptors in the carotid body. This allows for complex processing of sensory stimuli (e.g., hypoxia and acid hypercapnia) involving both autocrine and paracrine signaling pathways. This review summarizes and evaluates current knowledge of these pathways and presents an integrated working model on information processing in carotid bodies. Included in this model is a novel hypothesis for a potential role of type II cells as an amplifier for the release of a key excitatory carotid body neurotransmitter, ATP, via P2Y purinoceptors and pannexin-1 channels.     

Adaptation-induced plasticity of orientation tuning in adult visual cortex

A key emergent property of the primary visual cortex (V1) is the orientation selectivity of its neurons. The extent to which adult visual cortical neurons can exhibit changes in orientation selectivity is unknown. Here we use single-unit recording and intrinsic signal imaging in V1 of adult cats to demonstrate systematic repulsive shifts in orientation preference following short-term exposure (adaptation) to one stimulus orientation. In contrast to the common view of adaptation as a passive process by which responses around the adapting orientation are reduced, we show that changes in orientation tuning also occur due to response increases at orientations away from the adapting stimulus. Adaptation-induced orientation plasticity is thus an active time-dependent process that involves network interactions and includes both response depression and enhancement.     

Manifestations of dynamic coding of the amplitude-modulated sounds on the level of auditory nerve fibres

Average firing rate of the auditory nerve fiber as function of the level of the tone with the frequency equal to characteristic frequency of the fibers, can be defined as an input-output characteristic. It is known that the steepening of the input-output characteristic of the real auditory nerve fiber is more, and the width is less than the spontaneous activity of the fiber. The latter characterizes fiber's ability to generate spikes, if the stimulus is absent. However it is known, that the real auditory nerve fibers with low spontaneous activity reproduce amplitude modulation of the signals much better, than the fibers with high spontaneous activity. From the results of simulation experiments, it follows that the dynamic properties of the auditory nerve fibers, providing fine tuning or adaptation of a fiber threshold under the stimulus level but not the static input-output characteristics, are the reason of fibers reproduction of stimuli amplitude modulations. However the auditory nerve fibers with high spontaneous activity due to abrupt input-output characteristic are capable to reproduce modulations of sounds whose levels are lower than a threshold of the fiber, if a weak signal adds to a weak broadband noise. This is a phenomenon of stochastic resonance found in the reactions of auditory nerve fibers.