(less than 3.5 kHz in the guinea pig, Harrison and. Evans, 1979; Palmer and Russell, 1986), because we were uncertain of the effect on mean firing rate ...
Hearing Research, 45 (1990) 191-202
191
Elsevier HEARES 01354
Diversity of characteristic frequency rate-intensity functions in guinea pig auditory nerve fibres Ian M. Winter *, Donald Robertson and Graeme K. Yates Department of Physiology, The University of Western Australia, Nedlands, Australia
(Received 25 July 1989; accepted 28 November 1989)
Rate-intensity functions at characteristic frequency (CF) were recorded from single fibres in the auditory nerve of anaesthetised guinea pigs. Within the same animal, CF rate-intensity functions, although probably forming a continuum, could be conveniently divided into three groups; (1) Saturating; reach maximum discharge rate within 30 dB of threshold, (2) Sloping-saturation; initially rapid growth in discharge rate leading to a slower growth in discharge rate but not saturating and (3) Straight; approximately constant increase in firing rate per decibel increase in sound pressure up to the maximum sound pressures used. Thresholds for individual fibres were plotted relative to compound action potential thresholds at the appropriate frequency. Fibres with straight CF rate-intensity functions had the highest thresholds. Fibres of the saturating CF rate-intensity type had the lowest thresholds, and the sloping-saturation CF rate-intensity type had thresholds intermediate between saturating and straight. There was a close relationship between the type of CF rate-intensity function exhibited by a fibre and its spontaneous discharge rate. Fibres with saturating CF rate-intensity functions generally had high spontaneous discharge rates (greater than 18/s), whereas those with straight CF rate-intensity functions generally had low spontaneous discharge rates (less than 0.5/s). The majority of fibres with sloping-saturation CF rate-intensity functions had spontaneous rates between 0.5/s and 18/s. There was a negative correlation (r =-0.59) between the logarithm of the spontaneous discharge rate and relative threshold at CF with the lowest spontaneous rate fibres having the highest thresholds and vice-versa.This diversity of CF rate-intensity functions has functional implications for both frequency and intensity coding at high sound pressures in the mammalian auditory system. Guinea pig; Auditory nerve; Rate-intensity function; Spontaneous discharge rate; Threshold; Dynamic range
Introduction Characteristic frequency (CF) rate-intensity functions exhibited b y m a m m a l i a n a u d i t o r y nerve fibres c o m m o n l y have b e e n divided into two types: saturating a n d sloping-saturation. The first type shows a rapid increase in spike discharge rate with increasing s o u n d pressure u n t i l a b o u t 30 dB above threshold, when n o further significant increases or decreases i n discharge rate are observed. The seco n d type is characterized b y the same rapid initial
Correspondence to: Donald Robertson, Department of Physiology, The University of Western Australia, Nedlands, 6009, Western Australia. * Present address: M.R.C. Institute of Hearing Research, University Park, University of Nottingham, Nottingham BG7 2RD, U.K.
growth i n spike discharge rate b u t i n s t e a d of s a t u r a t i n g a b o u t 30 dB above threshold such fibres c o n t i n u e to increase their discharge rate with increases i n s o u n d pressure, albeit at a m u c h reduced rate. T h e relationship b e t w e e n the type of r a t e - i n t e n sity b e h a v i o u r a n d other properties of a u d i t o r y nerve fibres is unclear. Some authors have reported that s a t u r a t i n g or s l o p i n g - s a t u r a t i o n ratei n t e n s i t y f u n c t i o n s c a n be f o u n d i n fibres with either high or low s p o n t a n e o u s discharge rates (SR) (Sachs a n d A b b a s , 1974; P a l m e r a n d Evans, 1980). However, others ( L i b e r m a n , 1988) have reported that only low S R fibres possess slopings a t u r a t i o n r a t e - i n t e n s i t y functions. I n contrast, there is general agreement o n the relationship b e t w e e n d y n a m i c range, SR a n d threshold at CF. A u d i t o r y nerve fibres with the largest d y n a m i c
0378-5955/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)
192
ranges have the lowest spontaneous discharge rates (Evans and Palmer, 1980; Schalk and Sachs, 1980). Low SR fibres generally have the highest CF thresholds whereas fibres with high SR's have the lowest CF thresholds (Liberman, 1978; Kim and Molnar, 1979; Rhode and Smith, 1985). Schalk and Sachs (1980) have also presented limited data which suggests that dynamic range is an increasing function of threshold at CF or a decreasing function of SR. In this paper we show that CF rate-intensity functions from single auditory nerve fibres in the guinea pig can be conveniently divided into three types: saturating, sloping-saturation and a new type not previously reported in the mammalian auditory nerve; straight. The shape of a given fibre's CF rate-intensity function is strongly related to both its threshold at CF and its SR. Some possible implications of the variation in CF rateintensity type recorded from auditory nerve fibres in the guinea pig are discussed and in a companion paper (Yates et al., 1990) we show that the range of CF rate-intensity types may be predicted from a knowledge of the mechanical nonlinearities present in the guinea pig cochlea, as Sachs and Abbas (1974) and Sachs et al. (1989) have predicted for the cat. Methods Recordings were made from ten adult pigmented guinea pigs (250-438 g) anaesthetised with Nembutal (Pentobarbitone sodium, 60 m g / m l ; 25 m g / k g ) and supplemented with H y p n o r m (Fentanyl citrate, 0.315 m g / m l and Fluanisone, 10 m g / m l ) ad libitum to achieve surgical levels of anaesthesia and analgesia. All incisions were infiltrated with Lignocaine. The animals were tracheostomised, artificially respirated, and rectal temperature was maintained at 38.5°C by means of a heating blanket. The animals were placed in a headholder with hollow ear bars and the left external ear was removed. The left bulla was exposed and a coated silver wire was placed under the bony ridge near the round window to monitor the compound action potential (CAP) threshold. Sound stimuli were delivered to the left tympanic membrane via a calibrated Bruel and Kjaer 0.5 inch reverse-driven condenser micro-
100
A
r'~ m "0
8C
60 40
B -.~
6O
c < m "o
80
Be,ore After 100
10
Frequency (kHz)
100
Fig. l. (A) Typical sound calibration showing the sound intensity obtained at 0 dB attenuation for animals 6-10. (B) A representative example of CAP thresholds (animal 7) between 2 and 30 kHz, in 2 kHz steps, before and after opening the cochlea.
phone fitted into the hollow ear bar. For the first five experiments the maximum sound pressure varied between 96 and 116 dB SPL over the frequency range 8 - 2 0 kHz. For the remaining experiments the sound system was typically flat + 3.5 dB, with m a x i m u m sound pressures varying between 85 and 93 dB SPL in the frequency range 8 - 2 0 kHz (Fig. 1A), although detailed SPL frequency-response calibrations were not obtained in all animals. A sound calibration was not obtained for frequencies above 20 kHz. CAP thresholds (2-30 kHz, in 2 kHz steps) were monitored throughout the experiment and nerve-fibre recordings were discontinued from frequency regions in which thresholds had deteriorated by more than 5 dB from the initial CAP audiogram. A greater than 5 dB deterioration across all frequency regions resulted in the termination of the experiment. Recordings were obtained from auditory nerve fibres using the modiolar approach developed by Alder and Johnstone (1978). Briefly, the cochlea was opened by careful shaving with a modified
193 scalpel blade. A small hole was then made in the modiolar wall with a fine hand-held pick. A typical example of the CAP audiogram measured before and after such an opening of the cochlea is shown in Fig. lB. Threshold changes were typically small, and in the example shown are within measurement error (+_ 3 dB). Glass micropipettes filled with 2 M potassium acetate, and with resistances of 40-90 M~2 were inserted into the modiolar hole under visual control. The electrode was advanced using a stepping motor microdrive located outside the sound-attenuating room. A 50 ms burst of broadband noise was used as a search stimulus. On contact with a fibre, its spontaneous discharge rate was estimated in the absence of acoustic stimulation, over a 5 s period. A frequency versus threshold curve (FTC) was then determined around the fibre's CF using a threshold tracking algorithm written by one of the authors (GKY). Briefly, the experimenter selected the range of frequencies, in 0.5 or 1.0 kHz steps, for which a threshold would be estimated. A 50 ms tone-burst (rise-fall time 0.5-1 ms) was presented at a rate of 10/s. The number of spikes occurring during the stimulus period was compared with the number of spikes occurring in an equivalent period immediately following stimulus offset. This latter period was used as our ongoing estimate of spontaneous rate for the purposes of estimating threshold. For each frequency a threshold criterion was set at an increase in discharge rate of the RMS of (20 spikes/s and 20% of the spontaneous rate). For fibres with spontaneous rates less than 30 this approximates to a 20 spikes/s increase while for spontaneous rates greater than 100 it approximates to 20% of spontaneous. As the statistical likelihood of ever exactly determining threshold was small, an upper and lower limit of acceptable discharge rates which could be called threshold was defined. The upper and lower limits were symmetrical around the threshold criterion and allowed a variation of half the criterion threshold. Error limits were calculated for the upper and lower threshold limits and if the spike rate was less than the upper limit minus error and more than the lower limit plus error, threshold was located. This threshold tracking program was used to estimate fibre CF and where there were two or more adjacent frequencies
of equal lowest threshold, CF was taken to be the mean of those frequencies. Rate-intensity functions were then collected at CF and, in some experiments, at a frequency below CF. The intensity of a 100 ms tone-burst (rise-fall time 0.5-1.0 ms; 400 ms inter-stimulus interval) was varied pseudo-randomly in 3 dB steps over a 99 dB range (0-99 dB attenuation). All rate-intensity functions presented in this paper were measured at the fibre CF. Discharge rate in rate-intensity functions was normalised using Eqn. 1 (Sachs and Abbas, 1974). Normalised rate =
R ( d) - R (sp)
R(sat) - R(sp)
(1)
R(d) is the driven discharge rate, R(sp) is the spontaneous discharge rate and R(sat) is the saturated discharge rate. R(sat) was estimated over the five highest sound pressures used. R(sat) for sloping-saturation and straight rate-intensity functions was taken as the highest discharge rate. R(sp) was estimated by averaging the discharge rates over the lowest four or five sound intensities used. Sound pressure was normalised relative to threshold at CF for each fibre. Results
Recordings were obtained from a total of 229 fibres in ten animals, with CFs ranging from 2.5 kHz to 28 kHz. Qualitatively, all the phenomena discussed in detail below were observed across the full extent of this CF range. However, for quantitative analysis, we have used only those data which satisfied a number of stringent criteria. First, fibres were included only if their CFs were in frequency regions that showed a less than 5 dB increase in CAP threshold relative to the first CAP measurements in that animal. Second, we excluded from quantitative analysis all fibres with CFs below 8 kHz. This is for two reasons; fibre's with CFs between approximately 4 kHz and 8 k H z were infrequently found with our use of the modiolar approach, and secondly, we chose to avoid frequency regions in which phase-locking occurs (less than 3.5 kHz in the guinea pig, Harrison and Evans, 1979; Palmer and Russell, 1986), because we were uncertain of the effect on mean firing rate
194
1
7° 0
o.
I
n
~o I0(
0(~ O.
j
300,
2oo. (3. 100.
o A
do
s0
"00.S, _ 0.4, 0 Z
"
~o
dB Attenuation
F
0.2,
-20
0
20
40
60
do
,~0
dB Arbitrary Ref.
Fig. 2. Representative examples of the three types of rate-intensity function. (A) saturating, (B) sloping-saturation and (C) straight. For all units, CF varies between 16 and 24 kHz, maximum sound level varies between 104 and 115 dB SPL. ( D - F ) Normalised rate-intensity functions for the same fibres in A - C . (D) saturating, (E) sloping-saturation and (F) straight.
when there existed a cycle-by-cycle variation in firing probability. Finally, rate-intensity functions were analysed only if sufficient repetitions of the stimulus (minimum of 5 repetitions) were obtained at all sound pressures used (usually 34). The signal-to-noise ratio also had to be excellent throughout the recording period such that no spurious triggering was obtained. This selection procedure left us with a total of 170 fibres with CFs ranging from 8 to 24.5 kHz for which
threshold at CF, spontaneous rate, and F T C data were obtained. For 89 of these fibres, we also obtained accurate measurements of rate-intensity functions at CF. The typical variation found in rate-intensity shape for guinea pig auditory nerve fibres is illustrated in Fig. 2. The fibres in Fig. 2 are representative of their type and are not all from the same animal. Sound pressure is plotted in dB attenuation and CFs vary between 16 and 24 kHz.
195 The maximum sound pressure at 0 dB attenuation varied between 104 and 116 dB SPL in the frequency range from 16 to 20 kHz. Saturating rate-intensit3~ functions are shown in Fig. 2A. These fibres were the most frequently encountered. However, a reliable estimate of the relative numbers of the different rate-intensity types was not attempted in this study, since our experiments were deliberately biased in trying to record, from each CF region encountered, a variety of rate-intensity types. Fibres with sloping-saturation rate-intensity functions (Fig. 2B) show an initial rapid rise in discharge rate similar to the saturating type, but instead of saturating, this second type continued to increase discharge rate with further increase in sound pressure, albeit at a much reduced rate. Both saturating and sloping-saturation rate-intensity functions have been described in detail by other workers and the overall features of these rate-intensity functions in our data are in good agreement with these previous reports (Sachs and Abbas, 1974; Palmer and Evans, 1980). However, the third type we describe here, 'straight' (Fig. 2C), has not previously been reported in studies of the mammalian auditory nerve response to CF tones, although it is possible that the 'crossed ramp' fibres of N o m o t o et al. (1964) may have been of the straight type. Previous studies have demonstrated straight RI functions but only at stimulus frequencies well above CF (Sachs and Abbas, 1974; Evans, 1975). Straight rate-intensity functions showed an almost constant increase in discharge rate with each dB increase in sound pressure. Units characterised by straight or sloping-saturation rate-intensity functions showed no signs of saturation at the highest sound pressures used in this study (over 110 dB SPL for some fibres). In Figs 2D, E and F the same rate-intensity functions in Figs. 2A, B and C are shown normalised with respect to both CF threshold determined from the FTC, and m a x i m u m recorded driven discharge rate. This emphasizes the differences between the different rate-intensity types in both dynamic range and slope. Dynamic range has often been quantified by the range in dB between a fixed high and low percentage of m a x i m u m driven rate (where m a x i m u m driven rate is defined
as saturated discharge rate minus SR). We have not attempted such a quantification. Clearly this type of analysis is inappropriate for both slopingsaturation and straight rate-intensity functions as neither type has usually saturated at the highest sound pressures used.
Relationship between threshold, rate-intensity type and Qloa8 It is evident+ from the data shown in Figs. 2 A - C that there is a relationship between rate-intensity type and threshold. To examine this relationship m o r e closely we have normalised thresholds at CF relative to the CAP threshold at the frequency nearest a fibre's CF. For example, if a fibre had a CF of 19.5 kHz and threshold at CF of 20 dB SPL and the CAP threshold was 25 dB SPL at 20 kHz, the relative threshold for that fibre would be - 5 dB (i.e. 5 dB more sensitive than the CAP threshold nearest it's CF). This normalising procedure reduces the variability in threshold caused by pooling data across animals and across different CF regions. Fig. 3 shows the relationship between spontaneous rate, relative threshold normalised according to the above procedure, and rate-intensity type. All fibres with straight rate-intensity functions had thresholds at CF greater than the corresponding CAP threshold, in contrast to fibres having saturating rate-intensity functions, which had thresholds at CF below their corresponding CAP thresholds. Fibres with sloping-saturation 20-
-o 0 t-
1510
~
•
+
¢-
4-
0-
