| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
The Journal of Neurophysiology Vol. 79 No. 6 June 1998, pp. 3041-3059
Copyright ©1998 by the American Physiological Society
Department of Psychology, Monash University, Clayton, Victoria 3168, Australia
 |
ABSTRACT |
|---|
 Abstract
 Introduction
Methods
Results
Discussion
References
|
|---|
Heil, Peter and Dexter R. F. Irvine. Functional specialization in auditory cortex: responses to frequency-modulated stimuli in the cat's posterior auditory field. J. Neurophysiol. 79: 3041-3059, 1998. The mammalian auditory cortex contains multiple fields but their functional role is poorly understood. Here we examine the responses of single neurons in the posterior auditory field (P) of barbiturate- and ketamine-anesthetized cats to frequency-modulated (FM) sweeps. FM sweeps traversed the excitatory response area of the neuron under study, and FM direction and the linear rate of change of frequency (RCF) were varied systematically. In some neurons, sweeps of different sound pressure levels (SPLs) also were tested. The response magnitude (number of spikes corrected for spontaneous activity) of nearly all field P neurons varied with RCF. RCF response functions displayed a variety of shapes, but most functions were of low-pass characteristic or peaked at rather low RCFs (<100 kHz/s). Neurons with strong responses to high RCFs (high-pass or nonselective RCF response function characteristics) all displayed spike count
SPL functions to tone burst onsets that were monotonic or weakly nonmonotonic. RCF response functions and best RCFs often changed with SPL. For most neurons, FM directional sensitivity, quantified by a directional sensitivity (DS) index, also varied with RCF and SPL, but the mean and width of the distribution of DS indices across all neurons was independent of RCF. Analysis of response timing revealed that the phasic response of a neuron is triggered when the instantaneous frequency of the sweep reaches a particular value, the effective Fi. For a given neuron, values of effective Fi were independent of RCF, but depended on FM direction and SPL and were associated closely with the boundaries of the neuron's frequency versus amplitude response area. The standard deviation (SD) of the latency of the first spike of the response decreased with RCF. When SD was expressed relative to the rate of change of stimulus frequency, the resulting index of frequency jitter increased with RCF and did so rather uniformly in all neurons and largely independent of SPL. These properties suggest that many FM parameters are represented by, and may be encoded in, orderly temporal patterns across different neurons in addition to the strength of responses. When compared with neurons in primary and anterior auditory fields, field P neurons respond better to relatively slow FMs. Together with previous studies of responses to modulations of amplitude, such as tone onsets, our findings suggest more generally that field P may be best suited for processing signals that vary relatively slowly over time.
The auditory cortex of mammals contains multiple fields distinguished by anatomic and physiological criteria (for reviews, see Aitkin 1990
Surgery
The data for this study were obtained from the same 10 adult cats (5 of each sex; weighing between 2.2 and 4.8 kg) that generated the data of our previous report (Heil and Irvine 1998 Acoustic stimulation and recording procedures
The cat was located in a sound-attenuating chamber. Stimuli were produced digitally (Tucker Davis Technology) and presented to the cat's ears via calibrated sealed sound delivery systems. Each system consisted of a STAX SRS-MK3 transducer in a coupler whose sound delivery tube fitted snugly into the meatal stub. The calibration procedure is described and a representative transfer function of the system illustrated in Heil et al. (1992b)
Data analysis
Response areas were based on the number of spikes discharged during the entire tone burst duration. Contour plots of the response areas (e.g., Figs. 1A and 2A) were generated with commercial software (Surfer, Golden Software). In addition, mean first-spike latency and its standard deviation (SD) were obtained for each frequency/SPL combination. For analysis of FM responses, it was inappropriate to simply count the number of spikes that occurred during the stimulus. Because FM sweep duration was inversely related to RCF, a neuron's response to short-duration/high-RCF stimuli could occur after the ends of such sweeps (e.g., Fig. 2E). Therefore, the response analysis windows included a 100-ms period immediately after each sweep. Furthermore, the number of spikes discharged was corrected for the SR of the neuron. This was necessary because otherwise any spontaneous activity would have contributed more spikes to longer-duration/lower-RCF stimuli. SR was measured in the 100-ms interval preceding each sweep, and the average SR was derived from spike counts in all intervals. The number of spontaneous spikes expected during the response analysis window was then extrapolated from this average. The number of spikes per FM sweep, corrected for SR, was the measure of response R and was plotted against RCF to obtain RCF response functions for each FM direction (e.g., Fig. 1, B and C).
Database
Only a minority of ~200 well-isolated neurons in field P responded to the tone burst stimuli tested (see also Heil and Irvine 1998 Sensitivity to the rate of change of frequency
Start and end frequencies of FM sweeps fell ~1-2 kHz outside the boundaries of a given neuron's response area (Figs. 1A and 2A). The responses of nearly all field P neurons to such FM sweeps were affected by their RCF. A total of 93 RCF response functions for either FM direction were obtained from the 66 neurons studied at one or multiple SPLs. These RCF response functions displayed a variety of shapes. For convenience, these functions were classified into six categories: band-pass (BP), low-pass (LP), high-pass (HP), complex (CX), nonselective (NS), or nonresponsive (NR). A function was classified as BP if responses to RCFs higher and lower than the RCF eliciting the maximum response (best RCF) were <50% of maximum. If only RCFs higher or lower than the best RCF were <50% of maximum, the function was classified as LP or HP, respectively. If the responses to all RCFs tested were within 50% of maximum, the function was classified as NS. Functions that exhibited multiple distinct maxima were classified as CX. If none of the RCFs tested elicited >0.5 spikes/stimulus, the function was classified NR.
Sensitivity to FM direction
The directional sensitivity of field P neurons was assessed by a DS index, which was calculated for certain RCFs at all SPLs tested (see METHODS). For a given neuron, this index generally varied as a function of RCF and SPL. For neuron 96-42/07, for example, and at low SPLs most DS indices were positive, whereas at 55 dB SPL, the index changed systematically from positive to negative values with increasing RCF and at 80 dB SPL, all indices were negative (Fig. 1D). In some cases, the DS magnitude appeared to change systematically with RCF, though without obviously reversing sign (e.g., neurons 96-42/14 and 96-39/04; Figs. 4B and 5B). In other cases, DS indices appeared to scatter unsystematically around some mean, with or without reversing sign (e.g., Figs. 2D, 3B, and 6B). The considerable variation of DS with RCF for most neurons is emphasized by Fig. 9A, which shows all calculated DS indices plotted as a function of RCF. Data obtained from the same neuron and with the same SPL are connected. Note that hardly any lines run horizontally but rather course in all directions. Nevertheless, at any given RCF, DS indices span the entire range without obvious shifts in the distribution. The latter point is more clearly illustrated by Fig. 9B, which plots the mean (
Timing of responses
For RCF response functions of LP characteristic, one could argue that the neuron's response is simply proportional to the time spent by an FM sweep within the neuron's excitatory response area. However, this was not the case. The responses of field P neurons to FM sweeps were generally rather phasic in nature (Figs. 1, E and I, and 2E) and, at least for long-duration/low-RCF sweeps, were shorter than the time spent by the sweep within the neuron's response area. This is evident, for example, from consideration of Fig. 1 A and E or I. At 20 dB SPL, ~20-25% of the duration of the FM sweeps, which cover the 3-8 kHz range, is spent traversing the response area of neuron 96-42/07. However, the duration of that neuron's response to low RCFs is only some 10% of the duration of the stimulus.
LATENCY AND EFFECTIVE Fi.
The PSTHs illustrated in Figs. 1, E and I, and 2E show that the timing of the phasic response relative to the duration of the FM stimulus shifts systematically as a function of RCF. For example, for upward FM sweeps and for an RCF of 2 kHz/s, the response of neuron 96-42/07 occurs during the first half of the stimulus, whereas with increasing RCF, the response shifts more and more toward the end of the stimulus or even beyond (Fig. 1E). Such a "phase" shift also was observed routinely in the responses of neurons in AI (Heil et al. 1992b EFFECTIVE Fi AND RESPONSE AREA.
For neuron 96-42/07, the values of effective Fi obtained from the slopes k of the relationships between latency and FM duration are identified in Fig. 1A by arrowheads on four horizontal lines. The lines represent the range and SPLs of the FM stimuli and the positions of the arrowheads pointing to the right and left represent the effective Fi to upward and downward FM sweeps, respectively. Note that, in general, effective Fi and CF or best frequency do not coincide. Rather, effective Fi for upward sweeps is always lower than that for downward sweeps. In other words, the response to FM sweeps is triggered when the frequency is modulated toward the best frequency. Effective Fi values change with SPL and roughly follow the neuron's response area boundary, although they do not coincide exactly with that boundary. Very similar data were obtained for neuron 96-27/06 (Fig. 2A).
PRECISION OF SPIKE TIMING AND FREQUENCY JITTER.
The measure of response latency was the mean latency of the first spike within the analysis window relative to stimulus onset. The jitter around the mean on individual presentations was quantified by the standard deviation (SD) of the latency of the first spike. As shown for neurons 96-42/07 and 96-27/06 (Figs. 1, G and K, and 2G), SD generally decreased with increasing RCF, though the slope of such functions became shallower with increasing RCF, even when plotted on double-logarithmic axes. SPL appeared to have little effect on SD.
RELATIONSHIP OF TEMPORAL PRECISION AND RCF RESPONSE FUNCTION CHARACTERISTICS.
Figure 11, A and B, also illustrates that the range of frequency jitter (on a logarithmic scale) covered by the neuronal population is rather wide at low RCFs but narrows as RCF increases. This appears due to fact that those neurons for which the index of frequency jitter was large at low RCFs did not respond at high RCFs. For neuron 96-40/13, for example, and for either FM direction, the frequency jitter at low RCFs is relatively large (in excess of 0.3 kHz at 10 kHz/s; Fig. 3D, cf. Fig. 11). This neuron's RCF response functions for both FM directions show LP characteristic (Fig. 3A), so measures of frequency jitter could not be obtained at high RCFs. In contrast, the frequency jitter of neuron 96-42/14 at low RCFs is relatively small (<0.1 kHz; Fig. 4D) and its RCF response functions are of HP characteristic (Fig. 4A). The frequency jitter of neurons 96-39/04 (Fig. 5D) and 96-39/08 (Fig. 6D) are between those of neurons 96-40/13 and 96-42/14, and their RCF response functions are of BP characteristic (Figs. 5A and 6A). To illustrate the more general validity of this relationship, functions from neurons for which frequency jitter at low RCFs was large are shown by open symbols in Fig. 11, A and B. Note that the proportion of open symbols decreases with increasing RCF: open symbols constitute nearly half of the data points at low RCFs but only a small proportion at high RCFs.
RELATIONSHIP BETWEEN CHARACTERISTICS OF TONE BURST ONSET RESPONSES AND RCF RESPONSE FUNCTIONS.
The responses of neurons in auditory cortex, including field P, to tone bursts are dominated by spikes triggered by tone burst onsets. In essence, the onsets constitute aperiodic unidirectional modulations of SPL (peak pressure), and it is therefore of interest to see whether there are correspondences between a neuron's response to such modulations of amplitude and to those of frequency. When the tone burst rise time is fixed and its SPL is varied, the rate of change of peak pressure is covaried. Although the magnitude of a neuron's onset response is not strictly a function of that rate (Heil 1997b
We have shown here that FM sweeps are effective stimuli for neurons in field P of cat auditory cortex. Because FM sweeps were not used as search stimuli and only those neurons that also responded to tones were tested with FM sweeps, we cannot address the issue of whether there might be neurons in field P that respond selectively to FM stimuli. In the following sections, we will first discuss our basic findings and compare them, where possible, with those of other studies. On a more speculative basis, we then will consider the possible implications of our data for rate and temporal coding of FM stimuli and for functional specializations of different fields within auditory cortex.
Sensitivity to the rate of change of frequency
We have demonstrated here that field P neurons are generally quite sensitive to RCF. RCF response functions displayed a variety of shapes, with LP functions being most prominent. Best RCFs assigned to the various RCF response functions were distributed over at least three orders of magnitude, but low best RCFs predominated. Sensitivity to RCF is not a unique feature of neurons in field P but has been demonstrated in other auditory cortical fields and at more peripheral levels and in a variety of species (e.g., Britt and Starr 1976 Directional sensitivity
We have shown here that the responses of at least some field P neurons are also sensitive to the FM direction. Again, FM directional sensitivity is not unique to field P but can be found at nearly all levels of the auditory pathway as peripheral as the cochlear nucleus (Britt and Starr 1976 Response timing
We have proposed here that the phasic responses of field P neurons to linear FM sweeps of a given SPL and traversing a neuron's response area in a given direction are triggered whenever the instantaneous frequency of the signal reaches a particular value, termed the effective Fi. With some degree of confidence, it can be stated that the effective Fi is largely independent of RCF. This follows from the observations that response latency increased approximately linearly with the duration of the FM sweep and that the delay L0, obtained from the y intercept, corresponded to the neuron's minimum latency. Using the same method of analysis, an invariance of effective Fi with the RCF of linear FM sweeps was proposed previously for AI (Heil et al. 1992b Implications for FM coding
RATE CODING.
With respect to the processing or representation of FM sounds in the auditory system, the emphasis has traditionally been on the analysis of the response magnitudes of the neurons (such as firing rate or other measures of response). The finding that a neuron's response depends, for example, on the direction or on the speed of FM, with some neurons being direction selective by some criterion (e.g., DS > 0.3) or apparently tuned to RCF, has implied that FM direction and RCF are in fact encoded in that response. The response properties often are compared with the corresponding parameters occurring in those sounds in the animal's environment that are deemed of particular importance for the animal, such as vocalizations of conspecifics or of prey or predators (Felsheim and Ostwald 1996 TEMPORAL CODING.
Most of these ambiguities also could be resolved by comparing the relative timing rather than, or in addition to, the magnitudes of responses from different neurons. As was outlined above, a neuron's response to FM sweeps is triggered when the instantaneous frequency reaches a particular value, the effective Fi. This frequency appears to be independent of the speed, and hence of the time course, of FM. Effective Fi is associated, more or less closely in different neurons, with the border of the response area, i.e., depends on the width of the response area and on CF. An FM sweep therefore would trigger responses from a population of neurons in a particular temporal sequence (and spatial order, given a tonotopic map), which depends on the neurons' effective Fi and on the time course of the stimulus frequency. In a population of neurons with similar delay L0, the triggering sequence corresponds to the sequence in which the responses actually occur. In effect, the response triggering sequence tracks the signal's changing frequency. Consider an array of five neurons (numbered 1-5) with increasing effective Fi from 1 to 5. Any response triggering sequence in ascending numerical order then could represent an upward sweep, a sequence in descending order a downward sweep. The range of modulation could be represented by the range of effective Fi: a wider frequency excursion might activate neurons 1-5, whereas a more narrow excursion might activate only neurons 2-4. The intervals between response triggering in different neurons could represent the speed of FM, with intervals being short for a high RCF and vice versa. Thus the rate at which responses from different neurons are triggered increases with the rate of change frequency. In the same way, the intervals could represent the form of the modulation. For example, in a map where the effective Fi gradient is logarithmic, exponential sweeps would produce identical intervals while for linear sweeps intervals are the longer the higher the instantaneous frequency.
Areal specializations in auditory cortex
Compared with AI, many neurons in field P have nonmonotonic level functions, i.e., respond well to tones of relatively low and intermediate, but not of high SPL, particularly when tones are shaped with short rise times (Heil and Irvine 1998
INTRODUCTION
Abstract
Introduction
Methods
Results
Discussion
References
; Clarey et al. 1992). However, it is only in some echolocating bats, and only with respect to the processing of echoes or pulse-echo pairs brought about by reflection of the (rather stereotyped) biosonar pulses emitted by these species, that the role of different fields with respect to sound processing is reasonably well understood (for review see, e.g., Suga 1988). Our lack of understanding of the functional roles of the multiple fields in other species is largely due to the fact that cortical areas outside the primary auditory cortex (AI) have received very little attention.
), is one of the better studied areas (e.g., Kitzes and Hollrigel 1996; Orman and Phillips 1984; Phillips and Orman 1984; Phillips et al. 1995, 1996; Schreiner and Urbas 1988). A remarkable feature of field P is its very high percentage of neurons with responses to increments in the sound pressure level (SPL) of tonal stimuli that show nonmonotonic behavior (Kitzes and Hollrigel 1996; Phillips and Orman 1984; Phillips et al. 1995), i.e., with increasing SPL, the number of spikes increases to a maximum, defining a "best SPL" (after Brugge and Merzenich 1973), and then decreases. Many of these cells cease to discharge at high SPLs particularly when tones are shaped with short rise times, leading to circumscribed frequency versus SPL response areas (Heil and Irvine 1998; Phillips and Orman 1984; Phillips et al. 1995). This predominance of nonmonotonicity in field P, and the observation that best SPL was rather stable against variation of tone burst rise time across a range of short rise times, whereas with longer rise times "tuning" to SPL disappeared (Phillips et al. 1995), has led to the proposal that field P might be an area specialized for coding of intensity, particularly of brief sounds, which necessarily have short rise times (Kitzes and Hollrigel 1996; Phillips and Orman 1984; Phillips et al. 1995).
). We have recently reinvestigated this proposal and have arrived at a different conclusion regarding the possible functional role of field P (Heil and Irvine 1998), viz. that field P may be better suited to process slowly varying envelopes, such as onsets of tones of long rise times, than rapidly varying envelopes, such as onsets of tones of high SPL and short rise time (Heil and Irvine 1998). Our reinvestigation of the proposed specialization of field P for intensity coding was driven by the recognition that neurons in this field, like those in AI, respond preferentially to the onset of a tonal stimulus (Phillips and Orman 1984; Phillips et al. 1995, 1996; Reale and Imig 1980), but that paradoxically these onset responses previously have been analyzed with respect to SPL, a parameter that characterizes the steady-state portion of such stimuli. Our reanalysis of the strength and of the timing of the onset responses of field P neurons therefore was focused on the initial time course of the envelope as it had been in our previous studies of AI (Heil 1997a,b; Heil and Irvine 1996).
View larger version (17K):
[in a new window]
FIG. 3.
A: responses of neuron 96-40/13 to upward and downward FM sweeps of 2-10 kHz and 15 dB SPL as a function of RCF. B: resulting directional sensitivity as a function of RCF. Response latency for upward and downward FM sweeps (see key in A) as a function of sweep duration. D: index of frequency jitter for upward and downward FM sweeps (see key in A) as a function of RCF.
; Härtel 1975) and by their major prey, such as birds and rodents. Human speech also contains many FM components and formant transitions (see Pickett 1980). FM sounds can vary along multiple dimensions, such as form of modulation (e.g., linear, exponential, sinusoidal), modulation range or frequency excursion, modulation frequency, "speed" of FM, direction of frequency change, and others. It is therefore nearly impossible to characterize the responses of single neurons along all of those dimensions. To test our hypothesis and to allow for the most straightforward comparison with neuronal responses to FM stimuli obtained from other auditory cortical fields in this species (Heil et al. 1992b,c; Mendelson and Cynader 1985; Mendelson et al. 1993; Phillips et al. 1985; Tian and Rauschecker 1994) and comparison with neuronal responses to aperiodic AM (i.e., to tone burst onsets) (Heil and Irvine 1998), we have selected aperiodic linear FM sweeps of either direction of FM and parametrically varied FM speed, i.e., the rate of change of frequency (RCF).
,c; Kowalski et al. 1995; Mendelson and Cynader 1985; Mendelson et al. 1993; Shamma et al. 1993; Tian and Rauschecker 1994). Alternatively, or additionally, one could envisage that information about FM parameters is encoded in the temporal sequence of response initiation of different neurons. To shed more light on the feasibility of rate and temporal coding of FM stimuli, we therefore have analyzed aspects of the timing of the responses of field P neurons to FM sweeps in addition to the strength of these responses.
METHODS
Abstract
Introduction
Methods
Results
Discussion
References
). For detailed accounts of anesthesia and surgery, the reader is referred to that study. Briefly, anesthesia was introduced with 40 mg/kg ip pentobarbitone sodium and maintained at a surgical level throughout the experiment with either intravenous pentobarbitone (cats 96-13, 96-26, 96-27, 96-29, 96-33, and 96-37) or an intramuscular cocktail of ketamine and xylazine (cats 96-39, 96-40, 96-42, and 96-43). The electrocardiogram was monitored continuously, and rectal temperature was held near 38°C. The trachea was cannulated, and a head holder fixed to the skull. A round-window electrode and a length of fine-bore polyethylene tubing, allowing static pressure equalization within the middle ear, were inserted in each bulla, which then were resealed. The skull and the dura overlying field P in the caudal bank of the posterior ectosylvian sulcus (PES) and parts of AI on the middle ectosylvian gyrus were removed. A specially designed Perspex chamber was fixed to the skull surrounding the opening. The chamber was filled with warm saline and sealed with a glass plate, on which a small hydraulic microdrive was mounted and which housed the glass-insulated tungsten microelectrode.
. The frequency response was flat within ±5 dB up to ~12 kHz, beyond which the output decreased by ~5 dB/octave. Compound action potential thresholds were monitored via the round window electrodes.
6 mm, at distances of ~0.5-1.5 mm from the center of the sulcus and at various dorsoventral positions. Tone bursts, clicks, and noise bursts were used as search stimuli and were delivered at various SPLs and varying laterality. Once a neuron was well isolated and found to be responsive to tone bursts, initial estimates of its characteristic frequency (CF; frequency of lowest response threshold), its response area in frequency versus SPL coordinates (henceforth termed response area), and its preferred laterality of stimulus presentation [viz., monaural ipsilateral, monaural contralateral, or binaural with identical stimuli at each ear (i.e., diotic)] were obtained using audiovisual criteria. Then quantitative data on the neuron's response area were obtained, generally with diotic stimulation. All tone bursts were of 200 ms duration and were shaped with cosine-squared rise/fall functions of 8.5 ms (or in some cases 1.7 ms) rise and fall times. Tone amplitudes ranged from below threshold to 90 dB SPL (re 20 µPa) generally in steps of 10 dB. In some instances, and over near-threshold ranges, step size was 5 dB. The frequency range and step size on a linear scale were selected so that they covered the neuron's response area with reasonable resolution (see Figs. 1A and 2A) (Heil and Irvine 1998). Frequency/SPL combinations were presented at 1 Hz in pseudorandom order with 15 repetitions, each with a different pseudorandom order. Intermixed with the tone burst trials were trials (1 of 20) during which no stimulus was presented and which were used to obtain a measure of the neuron's spontaneous discharge rate (SR).
View larger version (48K):
[in a new window]
FIG. 1.
Responses of neuron 96-42/07 to tones and to FM sweeps. A: response area, obtained with diotically presented tone bursts shaped with 8.5-ms cosine-squared rise and fall times. Amplitude and frequency steps were 10 dB and 0.5 kHz, respectively. Measure of response is the total number of spikes per stimulus recorded during each tone (200 ms). Isoresponse areas are shown by different shades of gray as identified by the scale (right). Plots were generated with commercial software (Surfer, Golden Software), using the kriging interpolation algorithm and a ratio of search radii reflecting the relative spacing of frequency and amplitude steps. Four horizontal lines represent the range (3-8 kHz) and SPLs (20, 30, 55, and 80 dB SPL) of the FM stimuli. Positions of the arrowheads on these lines pointing to the right and left represent the effective Fi to upward and downward FM sweeps, respectively, as determined from analysis of response latency to FM sweeps and as described in RESULTS. B and C: responses to upward and downward FM sweeps plotted as a function of rate of change of frequency (RCF). Different symbols identify different sound pressure levels (SPLs; see key in B). Negative values (below broken line) arise when the number of spikes recorded is less than that expected from spontaneous activity. D: directional sensitivity as a function of RCF and SPL (symbols as in B). E and I: poststimulus time histograms (PSTHs) of the responses of this neuron to 15 repetitions of upward and downward FM sweeps of 20 dB SPL and different RCF. Each PSTH starts at stimulus onset and ends 100 ms after stimulus offset. Stimulus offset is identified by a small marker below the baseline of each PSTH. Duration of the FM portion of the stimulus, given by 
F/RCF, decreases with increasing RCF. Because in this case F = 5 kHz (see A), and each stimulus has rise and fall times of 5 ms each, the time scales of the PSTHs decrease from 2,610 ms for a RCF of 2 kHz/s (top PSTHs) to 113 ms for a RCF of 1,970 kHz/s (bottom PSTHs). F and J: response latency for upward and downward FM sweeps plotted as a function of FM sweep duration (i.e., stimulus duration minus rise and fall times) and SPL (symbols as in B). G and K: SD of response latency for upward and downward FM sweeps plotted as a function of RCF and SPL (symbols as in B). H and L: index of frequency jitter for upward and downward FM sweeps plotted as a function of RCF and SPL (symbols as in B).
View larger version (34K):
[in a new window]
FIG. 2.
Responses of neuron 96-27/06 to tones and to FM sweeps. A: response area, obtained with diotically presented tone bursts shaped with 1.7-ms cosine-squared rise and fall times. Amplitude and frequency steps were 10 dB and 1 kHz, respectively. Other conventions as in Fig. 1A. B-D: responses to upward (B) and downward (C) FM sweeps, and directional sensitivity (D) plotted as a function of RCF. Different symbols identify different SPLs (see key in B). E: PSTHs of the responses to 10 repetitions of 20 dB SPL downward FM sweeps from 14 to 5 kHz at different RCFs. Other conventions as in Fig. 1, E and I. F: response latency for downward FM sweeps plotted as a function of sweep duration and SPL (symbols as in B). G and H: SD of response latency and corresponding frequency jitter for downward FM sweeps plotted as a function of RCF and SPL (symbols as in B).
, or to FM stimuli, or both, were recorded. The selection of parameters for FM sweeps was based on the neuron's response area. FM sweeps were shaped with linear rise and fall times of 5 ms, during which the frequency was held constant. The start and end frequencies present during the rise and fall time, respectively, generally fell outside the neuron's frequency response range. At the signal's full amplitude the frequency changed linearly from either a low to a high frequency (upward) or vice versa (downward). The rate of change of frequency (RCF) was varied parametrically, in some cases from as low as 0.5 kHz/s to as high as ~4,000 kHz/s, in equal steps on a logarithmic scale for each FM direction. Because start and end frequencies were fixed, FM sweep duration decreased with increasing RCF. FM sweeps longer than 3 s could not be generated by the system. Up to 20 repetitions of FM sweeps of different RCF and direction were presented in pseudorandom order. In several cases, FM sweeps of different SPLs were tested. The times of occurrence of discriminated spikes were stored on disk with 10-µs resolution for off-line analysis.
; Heil et al. 1992a-c; Kowalski et al. 1995; Mendelson and Cynader 1985; Phillips et al. 1985), a measure of a neuron's directional sensitivity (DS) was obtained by a comparison of the responses to upward (Rup) and downward FM sweeps (Rdown)
This index can vary between +1 and
1, where the sign indicates the preference for upward (+) or downward () sweeps, and the magnitude the degree of preference. A disadvantage of this index is that it is affected drastically by small changes in R, when the values of R are low. For example, when for a given pair of FM sweeps only a single spike is recorded in any number of repetitions for each FM direction, DS is zero. But, when for one direction no such spike occurs, DS is either 1 (when Rdown is 0) or 1 (when Rup is 0). Therefore, a DS index was calculated for a given RCF only when the response to at least one FM direction was 
0.5 spikes per stimulus.
RESULTS
Abstract
Introduction
Methods
Results
Discussion
References
), and response areas, which were used to select FM parameters, were obtained from 80 neurons. Sixty-six of those neurons were studied with FM sweeps over a range of RCFs for both FM directions at at least one SPL. The CFs of these neurons ranged from 0.4 to 18 kHz, with a median of 3.4 kHz. Eighteen neurons were studied at two (12 neurons) or more (6 neurons) SPLs. The data obtained from neurons 96-42/07 and 96-27/06 are among the most extensive and are illustrated in Figs. 1 and 2. These figures will be used throughout the results section to outline analytic procedures and some of the basic findings.
) and those of neuron 96-40/13 to both FM directions (Fig. 3A) represent examples of LP functions. BP characteristics are depicted by most of the other RCF response functions of neuron 96-42/07 (Fig. 1, B and C), by most functions of neuron 96-27/06 (Fig. 2, B and C), and by the two functions of neurons 96-39/04 (Fig. 5A) and 96-39/08 (Fig. 6A). For neuron 96-27/06 (Fig. 2B), the functions to upward sweeps of 20-40 dB SPL and for neuron 96-42/07 (Fig. 1B) to upward FM sweeps of 80 dB SPL were classified as NR. All four RCF response functions obtained from neuron 96-42/14 were of HP characteristics (Fig. 4A).
View larger version (19K):
[in a new window]
FIG. 5.
A-D: responses of neuron 96-39/05 to upward and downward FM sweeps of 0.1-6 kHz and 50 dB SPL. Organization of the figure is analogous to that of Fig. 3.
View larger version (19K):
[in a new window]
FIG. 6.
A-D: responses of neuron 96-39/08 to upward and downward FM sweeps of 2.5-6.5 kHz and 35 dB SPL. Organization of the figure is analogous to that of Fig. 3.
View larger version (25K):
[in a new window]
FIG. 4.
A-D: responses of neuron 96-42/14 to upward and downward FM sweeps of 0.1-12 kHz and 40 and 50 dB SPL (see keys). Organization of the figure is analogous to that of Fig. 3.
level function for CF tones. However, because the number of different SPLs that could be tested before the cell was lost was limited, systematic relationships cannot be ruled out. Best RCF for upward sweeps was not significantly correlated with the width of the response area at the corresponding SPL or with CF, while for downward sweeps there were weak correlations of the logarithm of best RCF with those of response area width and CF (r = 0.310 and r = 0.234, respectively; n = 79; P < 0.05).
View larger version (29K):
[in a new window]
FIG. 7.
Distribution of "best RCF" for upward (top) and downward FM sweeps (bottom) for neurons with RCF response functions that fell into band-pass (BP), low-pass (LP), high-pass (HP), and complex (CX) categories. Histograms include measures at all SPLs tested in a given neuron. Side maxima, obtained from CX functions, also are included (see key).
View larger version (16K):
[in a new window]
FIG. 8.
Average number of spikes per FM sweep per neuron for upward and downward FM sweeps plotted as a function of RCF. See RESULTS for further information.
) ± 1 SD of DS indices averaged within RCF octave bands as a function of RCF. Note that the distributions of DS values are rather symmetrical around zero (- - -) and of similar width along the entire RCF range, except at the extreme ends, where only very few data points were collected.
View larger version (33K):
[in a new window]
FIG. 9.
Directional sensitivity of field P neurons. A: directional sensitivity (DS) indices plotted as a function of RCF. Data obtained from the same neuron and with FM sweeps of the same SPL are connected. B: mean ( ) ± 1 SD (
) of DS indices in A averaged within octave bands of RCF and plotted as a function of RCF. C: scatterplot of mean DS (see RESULTS) against DS at best RCF. - - -, unity slope and 0,0 origin. Box demarcates DS magnitudes of 0.3, often arbitrarily selected as the lower limit for directional selectivity. D: mean DS plotted as a function of the SPL of FM sweeps. Data from the same neuron are identified by the same symbol and connected by lines. - - - in A, B, and D mark the boundary between preferences for upward and downward sweeps (positive and negative DS indices, respectively).
; Tian and Rauschecker 1994). Using this criterion, ~58% of points would qualify as reflecting direction selectivity when based on the response at best RCF and ~53% when based on mean DS. Nevertheless, highest directional sensitivities were not necessarily associated with the best RCF. The data illustrated in Figs. 4 and 5 represent examples where the best RCFs are in fact correlated with the lowest directional sensitivities (cf. Figs. 4, A with B, and 5, Awith B).
).
). There it was argued that such a shift would occur if the responses of a given neuron to FM sweeps in a given direction were in fact triggered at, or evoked by, the same instantaneous frequency (termed effective Fi), independent of RCF, and if the responses occurred with some fixed delay L0 after the occurrence of that frequency. The response latency LRCF then would be given by
where RT is the rise time of the FM stimulus (viz., 5-ms; during which the frequency was constant) and DRCF the duration of the FM portion of the stimulus, which is inversely proportional to RCF. Thus LRCF should increase linearly with DRCF, and the effective Fi can be calculated from the slope k of that relationship. For upward and downward FM sweeps, effective Fi would be given by the terms (FL + k*
F) and (FH k*F), respectively, where FL and FH are the lowest and highest instantaneous frequencies of the sweep and F is the modulation depth, given by F = (FH FL).
) might there be systematic deviations from linearity. Linear functions also provided good approximations of the data for other neurons (see Figs. 2F and 3C-6C), although in one case, there was clearly a systematic deviation from linearity (not shown). Only 3 of 96 correlation coefficients were <0.95. The medians of the correlation coefficients for upward and downward sweeps were 0.996 and 0.997, respectively. The value of the delay L0, obtained from the y intercepts of the linear regressions, was correlated with the near-minimum latency ofthe neuron recorded with tone bursts (r = 0.523;n = 97; P < 0.0001). The slope and origin of that relationship were not different from 1 and 0, respectively (not shown).
View larger version (13K):
[in a new window]
FIG. 10.
Effective Fi, obtained from the slope of the linear relationship between response latency and FM duration, for upward and downward FM sweeps (see key in A) plotted as a function of CF (A) and of the near-frequency border of the response area at the equivalent SPL (B and C).
1 in a log-log plot, this index of frequency jitter remains constant. Conversely, if a neuron's SD to FM sweeps were constant, i.e., independent of RCF, the frequency jitter would increase as a function of RCF with a slope of 1 in a log-log plot (Fig. 11A, - - -). Note that this conversion is independent of whether or not the neuron's response is triggered at the same instantaneous frequency at different RCFs.
View larger version (33K):
[in a new window]
FIG. 11.
Index of frequency jitter, obtained from the SD of the response latency to FM sweeps, for upward (A) and downward (B) sweeps plotted as a function of RCF. Data obtained from the same neuron and with FM sweeps of the same SPL are connected. Functions from neurons with large frequency jitter at low RCFs are shown by open circles. Note that the proportion of open circles decreases with increasing RCF. Dashed line in A represents the frequency jitter resulting from a hypothetical RCF-invariant SD of 0.1 ms. C: frequency jitter averaged within octave bands of RCF and plotted as a function of RCF. D: scatterplot of frequency jitter obtained from a given neuron and for a given RCF and SPL to downward sweeps against that to upward sweeps. - - -, unity slope. Most point (viz. 219) lie above, and 7 and 115 points on and below that line, respectively.
). Use of this measure is exploratory, as we have insufficient data to establish the relationship between minimum SD for FM stimuli and tone bursts.
View larger version (16K):
[in a new window]
FIG. 12.
Relationship of RCF response function characteristics and temporal precision. A: RCF response function category for upward and downward FM sweeps (see key) plotted against the minimum SD obtained from the neuron's responses to tone burst onsets. Numbers at some of the symbols for BP and CX type functions identify "best RCFs" (in kHz/s) and, for CX, also the positions of side maxima. See RESULTS for further description. B: best RCFs for upward and downward FM sweeps plotted against minimum SD.
; Heil and Irvine 1998), characteristics of its level functions are informative: when for a fixed rise time the onset response increases monotonically with SPL, it also increases monotonically with rate of change of peak pressure; when the response changes nonmonotonically with SPL, it changes nonmonotonically with rate of change of peak pressure. For the purpose of a crude comparison with RCF response function categories, we have classified the level functions of field P neurons, obtained at CF and with short rise times (8.5 ms; cosine-squared rise function), as strongly nonmonotonic, if the neuron ceased to discharge at high SPLs (90 dB SPL), or as monotonic/weakly nonmonotonic, if the neuron had not ceased to discharge at 90 dB SPL. Figure 13 shows the distribution of RCF response function categories (pooled for the 2 FM directions) for those two classes of level functions. Neurons with RCF response functions that were classified as NS and as HP, i.e., that responded well or particularly well to high RCFs, were associated exclusively with the monotonic/weakly nonmonotonic category, i.e., they also responded well to rather high rates of change of peak pressure. Similarly, neurons with RCF response functions that were classified as LP and as BP, i.e., that responded well to low and intermediate, but not to high RCFs, were associated most often with the strongly nonmonotonic category, i.e., they also did not respond well to high rates of change of peak pressure. Overall, however, the correspondence between the shapes of a neuron's level function for CF tones and those of its RCF functions at some SPL is limited. The RCF response function category cannot be reliably predicted on the basis of the shape of the level function.
View larger version (24K):
[in a new window]
FIG. 13.
Relationship of the degree of nonmonotonicity of level functions to tone bursts and RCF response function category. Level functions were classified as either strongly nonmonotonic or as monotonic/weakly nonmonotonic. See RESULTS for further description.
DISCUSSION
Abstract
Introduction
Methods
Results
Discussion
References
; Felsheim and Ostwald 1996; Heil et al. 1992a-c; Kowalski et al. 1995; Mendelson and Cynader 1985; Mendelson et al. 1993; Møller 1971, 1972; Sinex and Geisler 1981; Suga 1973; Tian and Rauschecker 1994). A direct comparison of most of these data with those of the present study is not straightforward, given the differences in, for example, stimulus paradigms, data analysis, response measures, or temporal response patterns of the neurons recorded (e.g., phasic or sustained). A few studies, however, have employed stimuli similar to those used in our study, viz., linear FM sweeps of different RCF that completely traversed the neuron's excitatory response area, and have explored the (phasic) responses of single or multiunits in other fields of cat auditory cortex, viz. in AI (Heil et al. 1992b,c; Mendelson and Cynader 1985) and in the anterior auditory field (AAF) (Tian and Rauschecker 1994). Most neurons in these fields responded best to rather high RCFs, most often displaying high-pass RCF response functions or band-pass functions with peaks at relatively high RCFs. The distributions of best RCFs in AI and AAF were such that the number of units increased rapidly with increasing best RCF (for AI, see Fig. 5 in Heil et al. 1992b; for AAF, see Fig. 11 in Tian and Rauschecker 1994). When these histograms are compared with those for field P (Fig. 7), it appears that on average best RCFs in field P are considerably lower than in either AI or AAF. However, to judge the magnitude of this difference, it has to be kept in mind that the data in the different studies were obtained from neuronal populations of different CF and that best RCF may increase with CF (Heil et al. 1992a; Tian and Rauschecker 1994; present study).
) and in cat AI (Mendelson et al. 1993). Consistent with this interpretation, the latter authors also observed that best RCF increased with SPL, as does bandwidth. Second, the frequency representation in the cochlea [see, e.g., Liberman (1982) for cat and Manley et al. (1987) for chick] and in central auditory nuclei [see, e.g., Merzenich et al. (1975); Rajan et al. (1993) for cat AI; Heil and Scheich (1985, 1992) for chick field L] is better described by a logarithmic than by a linear relationship so that, at least to a first approximation, frequency octaves tend to occupy similar distances along the basilar papilla or along the tonotopic axis of the central neuronal structures. Therefore, and as was pointed out by Mendelson et al. (1993) and by Felsheim and Ostwald (1996), exponential, but not linear, FM sweeps produce a close to uniform motion of excitation along the basilar membrane and the tonotopic axis of central neuronal structures. A linear FM sweep of any given RCF would thus correspond to a higher rate in octaves per time when the sweep covers a low-frequency range as compared with a high-frequency range.
,c) unfortunately do not allow this comparison, as they were obtained from neurons along high-frequency (15-22 kHz) isofrequency domains, and consequently there was almost no overlap in the CF distribution with that of our field P sample (see Database). The only study of cat auditory cortex from which data on the CF-RCF relationship for linear FM sweeps can be extracted is that of Tian and Rauschecker (1994). Their Fig. 12A indicates, for upward FM sweeps, median RCFs of ~85 and 130 kHz/s for AAF neurons with CFs <5 kHz and between 5 and 10 kHz, respectively. The corresponding values from our field P sample are 9 and 12 kHz/s. Although this comparison is qualified by various differences in methodology between the two studies, it strongly supports the view that RCFs in field P are lower than those in AAF at least.
reported a positive correlation between best RCF and threshold. Likewise, in our study of the topographic organization of various response properties along the isofrequency domain of cat AI (Heil et al. 1992c), we observed that patches of low thresholds to CF tones and patches of low best RCFs were congruous. Low thresholds, in turn, were correlated with strongly nonmonotonic level functions (Heil et al. 1994; Schreiner et al. 1992). Hence, low-threshold nonmonotonic functions to CF tone bursts reflect sensitivity to slow, and suppression/inhibition of responses to rapid, AM. Although the RCF response function cannot be predicted reliably from the shape of the level function, these correspondences and correlations suggest some common mechanisms. In particular, they suggest that the inhibition that causes level functions to be nonmonotonic (e.g., Fuzessery and Hall 1996) may be one of the factors contributing to a neuron's RCF sensitivity.
; Erulkar et al. 1968). Only for auditory nerve fibers and some cochlear nucleus neurons are the PSTHs recorded with FM sweeps of opposite direction roughly mirror-images of one another (Britt and Starr 1976; Sinex and Geisler 1981), such that the numbers of spikes elicited by FM sweeps of opposite direction are roughly the same but the temporal patterns are reversed. The present study was not designed to explore the mechanisms behind the FM directional sensitivity of field P neurons, which must await further study. Basic mechanisms discussed for the generation of FM directional sensitivity include differential effects of spatiotemporal summation of excitatory postsynaptic potentials (EPSPs) caused by orderly (i.e., tonotopic) arrangements of inputs along dendrites (e.g., Erulkar et al. 1968; Fuzessery 1994; Heil et al. 1992a), ideas prompted by the theoretical work on the electrical properties of dendritic trees of Rall (1964; see also review by Segev 1992), and differential effects of temporal summation of EPSPs and inhibitory postsynaptic potentials caused by asymmetric lateral inhibitory areas (e.g., Fuzessery 1994; Heil et al. 1992a; Kowalski et al. 1995; Shamma et al. 1993; Shannon-Hartman et al. 1992; Suga 1965, 1973; for review, see Fuzessery 1994).
used linear FM sweeps of 2-kHz excursion, embedded in constant-frequency segments corresponding to the start and end frequencies of the sweep, and varied the center frequency of the sweep relative to the neuron's response area. They found, in cat AI, that a neuron responded best when the frequency was modulated toward its CF. Consequently, a neuron preferred upward sweeps when the modulation range was below its CF and downward sweeps when that range was above its CF and showed no directional preference when the range was around CF. This behavior, which appears to be closely related to our finding (see further) that a neuron's response is triggered when the frequency is modulated toward the CF, was independent of the directional sensitivity revealed with FM sweeps that traversed the neuron's entire frequency response range (Phillips et al. 1985). Versnel and Nelken (1994) found, in ferret AI, that directional sensitivities obtained with linear versus logarithmic FM sweeps were uncorrelated.
) and in chick field L (Heil et al. 1992a). Consequently, the DS index derived from a stimulus pair of a given (e.g., the best) RCF is not necessarily representative of the neuron's RCF-dependent directional sensitivity. However, if the DS index at the best RCF is used for the purpose of comparison, ~58% of field P neurons would be classified as direction-selective (DS > 0.3). This number compares well with the 65% obtained from single units in cat AI by Mendelson and Cynader (1985). It apparently also compares well with the 66% in cat AAF, reported by Tian and Rauschecker (1994). In this study, however, it is unclear which FM sweep pairs were selected to derive the index. Tian and Rauschecker (1994) illustrate the distribution of directional selectivities (DS > 0.3) over RCF and propose that neurons were more frequently direction-selective at lower RCFs (see their Fig. 15), a finding deemed consistent with psychophysical observations, made by them, suggesting that discrimination between different FM directions becomes increasingly difficult with higher RCFs. However, most AAF cells in their sample preferred rather high RCFs and responded only poorly to low RCFs (see their Figs. 9-13). This suggests that the high directional selectivities would be based on rather weak responses. Furthermore, as pointed out in the present study (see also Heil et al. 1992b; Phillips et al. 1985), the DS index is very sensitive to small differences in the response to upward versus downward FM sweeps, when the responses themselves are small. Tian and Rauschecker (1994) do not state whether they have excluded DS indices based on very weak responses. So it seems possible that the decreasing frequency of directional selectivities with increasing RCF observed by them might be an artifact caused by this vulnerability of the DS measure. In this context, it is noteworthy that in chick field L directional sensitivities, obtained at best RCF, increased with increasing RCF (Heil et al. 1992a). As shown here for field P (Fig. 9, A and B), the directional sensitivity of the neuronal population appears independent of RCF: at any given RCF DS indices range from 1 to 1, with a similar mean (near 0) and width of their distribution. It will be interesting to see whether and how DS indices relate to psychophysical measures of the discriminability of the FM direction at different RCFs.
; see also Heil 1997c). Response triggering by a particular instantaneous frequency also has been demonstrated at other levels of the auditory pathway, in different species and with other forms of FM (Felsheim and Ostwald 1996; Heil et al. 1992a; Schulze et al. 1997; Whitfield and Evans 1965). These studies also have shown that, when the FM starts outside and then enters a neuron's excitatory response area, the responses are triggered when the frequency is modulated toward the CF (see also Tian and Rauschecker 1994). Effective Fi appears to be more or less closely associated with the near-frequency border of the neuron's response area (Figs. 1A, 2A, and 10) (Felsheim and Ostwald 1996; Heil 1997c; Heil et al. 1992a,b; Schulze et al. 1997), and thus depends not only on FM direction but also on SPL. In AI, it was found that effective Fi was correlated very closely with the frequency where the slope of the neuron's response function to tones of different frequencies but identical amplitudes (frequency response function, FRF) was steepest, i.e., where the magnitude of the first derivative of the FRF was maximal (Heil 1997c; Schulze et al. 1997). That point along the FRF identifies the frequency around which the neuron is most sensitive to a change in the frequency of a tone burst. The tight correlation between that frequency and the effective Fi seen with FM sweeps may therefore not be coincidental. In any event, the match of effective Fi with the borders of the response area or the frequency of maximum FRF slope emphasizes their significance, as opposed to that of the best frequency, for processing of FM sounds, at least of those that start outside and traverse a neuron's response area, as discussed previously (Heil 1997c). Because the effective Fi for FM sweeps of a given direction and SPL is independent of the speed of FM for linear (Heil 1997c; Heil et al. 1992b; Schulze et al. 1997; present study) and for exponential (Felsheim and Ostwald 1996) sweeps and independent of the modulation frequency for sinusoidal FM (Heil et al. 1992a; Whitfield and Evans 1965), it is probably fair to say that the effective Fi is independent of the time course of the modulation. The mechanisms that underlie this phenomenon are unknown, but, as will be outlined below, it may have important implications for the neural representation of FM stimuli.
). At any given RCF, there was a range of frequency jitter, but neurons with precision of spike timing at low RCFs that was poor often did not respond to high RCFs (Fig. 11, A and B). A similar relationship emerged when the precision of spike timing to tone burst onsets was compared with the neuron's RCF response functions (Fig. 12). The possible implications of these findings for the neural representation of FM stimuli will be discussed further.
; Fuzessery 1994; Gaese and Ostwald 1995; Heil 1992a,b; Poon et al. 1991; Tian and Rauschecker 1994). Yet, it is obvious that such sounds, although important, constitute only a very small proportion of the diverse range of sounds in the environment that we, and many animals, are exposed to and that we can discriminate and identify. Indeed, the correspondence between the apparent tuning properties of neurons in species such as rats or cats and the equivalent acoustic parameters in such vocalizations is rather poor (Felsheim and Ostwald 1996; Gaese and Ostwald 1995; Heil et al. 1992a,b; Poon et al. 1991; Tian and Rauschecker 1994).
) and, as detailed earlier, for cat AI (Phillips et al. 1985). These results demonstrate that a neuron's response to FM sweeps is some joint function of several parameters, including at least RCF, direction, SPL, and modulation range, but possibly also other parameters, such as form of modulation (see Versnel and Nelken 1994). Clearly, the responses of several neurons would have to be compared to resolve the ambiguities with regard to these signal properties that are inherent in the response of a single neuron.
,b; Mendelson et al. 1993; Phillips et al. 1985; Tian and Rauschecker 1994; present study) but also at more peripheral levels (e.g., Britt and Starr 1976; Felsheim and Ostwald 1996; Poon et al. 1991). Through the involvement of different neurons, it seems possible to track FM sounds in which the frequency changes on time scales too rapid to be encoded in changes in the firing rate of single neurons, a problem emphasized by Bialek and Rieke (1992). Such latency-place codes may not only operate for FM signals but could be involved in other sensory processes (for review, see Cariani 1995).
).
; Heil and Irvine 1998).
, viz. that discriminability of FM direction decreases with increasing RCF. Assuming that the responses of neurons in the human auditory cortex to FM stimuli show a similar behavior as those reported here for cat field P and that humans might rely on a temporal tracking mechanism, one would expect that, for linear FM sweeps of a given modulation range, direction discriminability should deteriorate with increasing RCF. That discriminability also might be expected to be largely independent of SPL and for a given RCF, should improve with increasing modulation range. In contrast, if the discriminability relied on directional sensitivities in the neuronal responses, the distribution of DS indices (Fig. 9, A and B) would suggest that performance should be independent of RCF.
; Kitzes and Hollrigel 1996; Phillips and Orman 1984; Phillips et al. 1995). However, field P neurons, like those of AI, respond predominantly to tone onset, i.e., that part of the stimulus that constitutes an aperiodic AM. Nonmonotonic level functions therefore reflect good responses to relatively slow or intermediate AM and poor responses to relatively fast AM (Heil and Irvine 1998). On average then, and compared with AI, field P responds better to slowly varying envelopes. Similarly, Schreiner and Urbas (1988) found that neuron clusters in field P preferred lower modulation frequencies of sinusoidally amplitude-modulated tones than clusters in AI and AAF. Note that for such tones the envelope varies more rapidly the higher the modulation frequency. The present study has revealed that, on average, field P neurons also respond well to slow, but not to fast, FMs. Taken together, these neuronal properties therefore suggest that field P may be best suited or specialized for processing relatively slowly time-varying signals.
; Heil and Irvine 1997, 1998), the resolution of the proposed temporal tracking mechanism for changes in stimulus frequency or amplitude depends on the precision of spike timing. As the precision is relatively poor in field P (Heil and Irvine 1998; Phillips et al. 1995), the field's resolution of details in the stimulus time course would be rather limited.
| |
ACKNOWLEDGEMENTS |
|---|
We are grateful to J. F. Cassell, E. Hartley, V. N. Park, D. Sketchley, R. R. Williams, and members of the general departmental staff for excellent technical support and to Dr. R. Rajan for assistance in preliminary experiments.
The study was supported by the National Health and Medical Research Council of Australia.
| |
FOOTNOTES |
|---|
Address reprint requests to P. Heil.
Received 5 August 1997; accepted in final form 10 February 1998.
| |
REFERENCES |
|---|
|
Abstract
Introduction
Methods
Results
Discussion
References
|
|---|
This article has been cited by other articles:
|
|
 |
|
  T. A. Brown and R. V. Harrison Responses of Neurons in Chinchilla Auditory Cortex to Frequency-Modulated Tones J Neurophysiol, April 1, 2009; 101(4): 2017 - 2029. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. A. Razak and Z. M. Fuzessery Facilitatory Mechanisms Underlying Selectivity for the Direction and Rate of Frequency Modulated Sweeps in the Auditory Cortex J. Neurosci., September 24, 2008; 28(39): 9806 - 9816. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Qin, J. Wang, and Y. Sato Heterogeneous Neuronal Responses to Frequency-Modulated Tones in the Primary Auditory Cortex of Awake Cats J Neurophysiol, September 1, 2008; 100(3): 1622 - 1634. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. K. Pandya, D. L. Rathbun, R. Moucha, N. D. Engineer, and M. P. Kilgard Spectral and Temporal Processing in Rat Posterior Auditory Cortex Cereb Cortex, February 1, 2008; 18(2): 301 - 314. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. A. Atencio, D. T. Blake, F. Strata, S. W. Cheung, M. M. Merzenich, and C. E. Schreiner Frequency-Modulation Encoding in the Primary Auditory Cortex of the Awake Owl Monkey J Neurophysiol, October 1, 2007; 98(4): 2182 - 2195. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Malhotra and S. G. Lomber Sound Localization During Homotopic and Heterotopic Bilateral Cooling Deactivation of Primary and Nonprimary Auditory Cortical Areas in the Cat J Neurophysiol, January 1, 2007; 97(1): 26 - 43. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Luo, Y. Wang, D. Poeppel, and J. Z. Simon Concurrent Encoding of Frequency and Amplitude Modulation in Human Auditory Cortex: MEG Evidence J Neurophysiol, November 1, 2006; 96(5): 2712 - 2723. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. A. Razak and Z. M. Fuzessery Neural Mechanisms Underlying Selectivity for the Rate and Direction of Frequency-Modulated Sweeps in the Auditory Cortex of the Pallid Bat J Neurophysiol, September 1, 2006; 96(3): 1303 - 1319. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. M. Fuzessery, M. D. Richardson, and M. S. Coburn Neural Mechanisms Underlying Selectivity for the Rate and Direction of Frequency-Modulated Sweeps in the Inferior Colliculus of the Pallid Bat J Neurophysiol, September 1, 2006; 96(3): 1320 - 1336. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. N. O'Connor, C. I. Petkov, and M. L. Sutter Adaptive Stimulus Optimization for Auditory Cortical Neurons J Neurophysiol, December 1, 2005; 94(6): 4051 - 4067. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. K. Bizley, F. R. Nodal, I. Nelken, and A. J. King Functional Organization of Ferret Auditory Cortex Cereb Cortex, October 1, 2005; 15(10): 1637 - 1653. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Godey, C. A. Atencio, B. H. Bonham, C. E. Schreiner, and S. W. Cheung Functional Organization of Squirrel Monkey Primary Auditory Cortex: Responses to Frequency-Modulation Sweeps J Neurophysiol, August 1, 2005; 94(2): 1299 - 1311. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Brechmann and H. Scheich Hemispheric shifts of sound representation in auditory cortex with conceptual listening Cereb Cortex, May 1, 2005; 15(5): 578 - 587. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Fishbach, Y. Yeshurun, and I. Nelken Neural Model for Physiological Responses to Frequency and Amplitude Transitions Uncovers Topographical Order in the Auditory Cortex J Neurophysiol, December 1, 2003; 90(6): 3663 - 3678. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. E. O'Neill and W. O. Brimijoin Directional Selectivity for FM Sweeps in the Suprageniculate Nucleus of the Mustached Bat Medial Geniculate Body J Neurophysiol, July 1, 2002; 88(1): 172 - 187. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. A Hall, I. S Johnsrude, M. P Haggard, A. R Palmer, M. A Akeroyd, and A. Q. Summerfield Spectral and Temporal Processing in Human Auditory Cortex Cereb Cortex, February 1, 2002; 12(2): 140 - 149. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Brechmann, F. Baumgart, and H. Scheich Sound-Level-Dependent Representation of Frequency Modulations in Human Auditory Cortex: A Low-Noise fMRI Study J Neurophysiol, January 1, 2002; 87(1): 423 - 433. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. C. Loftus and M. L. Sutter Spectrotemporal Organization of Excitatory and Inhibitory Receptive Fields of Cat Posterior Auditory Field Neurons J Neurophysiol, July 1, 2001; 86(1): 475 - 491. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Fishbach, I. Nelken, and Y. Yeshurun Auditory Edge Detection: A Neural Model for Physiological and Psychoacoustical Responses to Amplitude Transients J Neurophysiol, June 1, 2001; 85(6): 2303 - 2323. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. I. Sanderson and J. A. Simmons Neural Responses to Overlapping FM Sounds in the Inferior Colliculus of Echolocating Bats J Neurophysiol, April 1, 2000; 83(4): 1840 - 1855. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
