Exp. Brain Res. 23, 407--423 (1975) 9 by Springer-Verlag 1975

Statistical Analysis and Interpretation of the Initial Response of Cochlear Nucleus Neurons to Tone Bursts J. A.M. van Gisbergen, J . L . Grashuis, P . I . N . Johannesma and A . J . H . Vendrik Laboratory of Medical Physics and Biophysics, University of Nijmegen, Nijmcgen (The Netherlands) Received April 7, i975 Summary. 1. Subject of investigation is the initial response of cochlear nucleus neurons and units presumed to be auditory nerve fibres to CF tone burst stimulation. 2. The initial response is characterized by computing the distribution of the latency of the first spike and of the duration of the first interval in the ensemble of responses to a large number of stimuli. 3. I n m a n y of the neurons the properties of both distributions appear to be related. The presumed auditory nerve fibres and spontaneously active cochlear nucleus neurons showing only activation responses to tonal stimuli (A type) exhibit irregularity in both response onset and intervals. Minimum latency and minimum first intervals are short. On the other hand, spontaneously active neurons with both activation and suppression in the response area (AS type) and silent neurons showing only activation (A(S) type) often show a more precisely timed onset of response and narrow interval distributions. I n m a n y neurons this leads to oscillations in the P S T t I (chopping). I n these neurons minimum latency and minimum first interval have higher values. The longer minimum latency cannot be attributed to longer pure time delays in these neurons. 4. The results are interpreted as speaking in favour of temporal integration as an important mechanism in m a n y of the AS and A(S) neurons, particularly those in the DCN. The firing patterns of A neurons are thought to indicate virtual absence of this mechanism. 5. Using pure time delay estimates derived from cross-correlation functions, computed from the responses to stationary noise, an a t t e m p t is made to estimate the integration time in the cochlear and in the cochlear nucleus neurons. Key

words:

Cochlear nucleus units - - Tone responses - - L a t e n c y firing - - Interval distribution

--

Regularity of

Introduction

Statistical analysis of response patterns of cochlear nucleus neurons to tonal and noise stimuli of long duration (Goldberg and Greenwood, 1966) has shown t h a t these response patterns display differences in their interspike interval distributions. Some neurons have interval distributions which resemble exponential distributions with a dead time, having an asymmetrical shape and a small modal

408

J.A.M. van Gisbergen et al.

value. I n a later study Goldberg and Brownell (1973) found these neurons in particular in the large spherical cell area of the ventral cochlear nucleus (VCN). Similar results were obtained from auditory nerve fibres under stimulated and unstimulated conditions (Kiang et al., 1965a; Kiang, 1968). Other units in the cochlear nucleus neuron population studied b y Goldberg and Greenwood (1966) exhibit more regular firing patterns reflected in narrower interval distributions with a rather symmetrical shape. I n the dorsal cochlear nucleus (DCN) most units appear to have such properties (Goldberg and Brownell, 1973). The problem arises of how neurons receiving input from the irregularly firing auditory nerve fibres can display regular firing.

Models Explaining Differences in Regularity o] Firing Two neuron models suggest an explanation of this phenomenon. First, in the Goldberg et al. (1964) neuron model it is assumed t h a t the threshold of the neuron is elevated after the occurrence of a spike and then declines with an exponential time course. I n this model the regularity of the firing pattern is related to the time constant of recovery. Neurons with a long recovery time constant will show only relatively small fluctuations in interval duration. Second, in a so-called Markov process model presented b y Molnar and Pfeiffer (1968) the mechanism causing more regular firing t h a n is present at the inputs is associated not with recovery processes alter spike initiation but with the mechanism of spike initiation itself. I t is assumed t h a t (1) the neuron receives input from a considerable number of irregularly firing inputs; (2) the depolarization caused by an input event is small in comparison with the threshold of the neuron; (3) the depolarization decays towards zero with an exponential time course. As a consequence threshold crossing will result only if the effects of several input events summate. There is a difference between the two neuron models which m a y be used to discriminate between them on the basis of extracellular recordings. The recovery process, the mechanism causing regular firing in the first model, comes into operation alter generation of an action potential, whereas temporal integration, the responsible mechanism in the second model, is effective be]ore the generation of the action potential. These t w o effects can be discriminated if we consider especially the occurrence of first and second spikes in the response to a stimulus evoking strong activity preceded by a period of low neural activity. For this purpose we used tone burst stimuli at the characteristic frequency (CF). Generally, before the first spike occurs in response to a CF tone a relatively long time without spikes has elapsed, and refractory properties do not play a significant role with respect to the generation of this first spike. The second spike in the response to this tone burst has been shortly preceded b y an action potential and as a consequence m a y be influenced b y refractory effects. A statistical analysis was therefore performed on the latency of the first spike and the interval between the first and the second action potential in the responses to a large number of tone bursts. For brevity we will use the terminology of a previous study (van Gisbergen, 1975 a) where the response patterns of cochlear nucleus neurons to tonal stimulation were classified. There appeared to be marked differences in the response patterns observed in the intensity-frequency response area. We distinguished:

Responses of CN Neurons to Tone Bursts

409

activation only (A type), activation and suppression (AS type), aud suppression only (S type). Neurons with low levels of spontaneous activity showed activation responses, but presence or absence of suppression remains an open question. These neurons are therefore indicated as A(S). Four temporal patterns of response were distinguished. The sustained category comprises the primarylike and chopper responses distinguished by Pfeiffer (1966). Build up neurons show long onset latency and gradually increasing firing rates. Transient neurons have quite the opposite response pattern: their response is mainly or o n l y at tone onset. The activation pattern of complex neurons is interrupted shortly after tone onset. I n fact it m a y be thought to consist of a transient and a build up component. Some of the recordings were presumably, but not certainly, obtained from auditory nerve fibres. These units are indicated as primary(?) fibres. The results are compatible with the notion t h a t temporal integration does not play an important role in the spike initiation process of primary(?) fibres and A type neurons. On the other hand, the results obtained from most AS and A(S) neurons, especially those located in the D~N, indicate extensive integration of input signals before a spike is generated. No compelling need is felt to assume a mechanism like t h a t proposed in the Goldberg et al. (1964) model.

Prediction o/ Latency on Basis o/ Cross-Correlation Function (CCF) Properties I n this paper an a t t e m p t is made to separate components contributing to the latency of cochlear nucleus neurons. Cross-correlation functions (CCFs) of neurons present interesting possibilities in this regard. As described in the previous paper (van Gisbergen et al., 1975b) cross-correlating the input signal consisting of stationary noise and the resulting sequence of action potentials of cochlear nucleus units results in a CCF with a shape as shown as Fig. 1. The CCF is obtained by averaging a large number of pre-spike noise stimuli. A necessary condition for obtaining a CCF which is clearly distinguishable from the noise is t h a t there is a phase lock of t h e action potentials with respect to the input signal. The CCFs, R(~), deviate significantly from zero for ~ > a until a point in time (v = 5) is reached where the amplitude differs negligibly from zero (Fig. 1). I n the preceding paper (van Gisbergen et al., 1975 b) a is defined as a parameter, representing a time delay, of a mathematical expression which approximates the experimentally obtained CCF, l~(v), a is determined by fitting the CCF, R(~), by this mathematical expression. 5 is operationally defined as the point in time such that the area under the envelope A(T) of t~(~) for ~ < 5 equals ~ 99% of the total area under A(T). The time period a < ~ < 5 is interpreted as the mean time period during which the signal is relevant for the generation of a discharge at ~ = 0. Starting from this interpretation it is feasible to predict a range within which the latency to supra-threshold tone bursts m a y vary provided t h a t two assumptions are made. First, R(~) is conceived of as the impulse response of a linear filter which accounts for the frequency selectivity of the neuron as well as for the various pure time delays present in the system. Second, spikes are generated as soon as the amplitude of the output signal of this filter crosses the threshold level of a pulse generating element following the filter. The first assumption determines the time course of the amplitude of the output signal of the linear filter when a tone (with a rise time of 2.5 msec) is presented at

410

J.A.M. van Gisbergen et al.

2.5 m s e c

/

time envelope of tone burst stimulus

cross-correlation function amplitude time course of output signal I i I I

I I I i

ct

{~

5+2.5

time (msec )

Fig. 1. Expected range of minimum latency (Lm) to tone burst stimulation of model outlined in text. When the threshold is greater than zero the latency to tone burst stimuli of given frequency will depend on the properties of the cross-correlation function (CCF) (see text), the threshold of the neuron and the intensity of the stimulus. For supra-threshold stimuli it is expected that a < Lm < 6 -t- 2.5 msec

t h e i n p u t . A f t e r t h e t i m e d e l a y a t h e a m p l i t u d e of t h e o u t p u t signal of t h e filter will increase m o n o t o n i c a l l y u n t i l after t > 6 q- 2.5 msec no f u r t h e r a p p r e c i a b l e increase in a m p l i t u d e is t o be e x p e c t e d (Fig. 1). T h e l a t e n c y of response of t h e neuron, in this model, will d e p e n d u p o n t h e p r o p e r t i e s of i~(z), u p o n t h e t h r e s h o l d o f t h e n e u r o n as well as u p o n t h e i n t e n s i t y of t h e s t i m u l u s (of given frequency). As s u p r a - t h r e s h o l d s t i m u l i were used i t w o u l d be e x p e c t e d from t h e m o d e l t h a t t h e s h o r t e s t l a t e n c y of response to t o n e b u r s t s (Lm, see m e t h o d s ) w o u l d be a < Lm < fi -1- 2.5 msec. a is i n t e r p r e t e d in t h e p r e c e d i n g p a p e r (van Gisbergen et al., 1975b) as r e p r e s e n t i n g t h e t i m e d e l a y s in t h e s y s t e m u p to t h e l o e a t i o n of t h e microelectrode, a is c o m p o s e d m a i n l y of t h e acoustic d e l a y from t h e s o u n d source to t h e t y m p a n i c m e m b r a n e , t h e t r a v e l t i m e of t h e w a v e in t h e cochlea, t h e s y n a p tic d e l a y in t h e cochlea, t h e c o n d u c t i o n t i m e of t h e spikes along t h e a u d i t o r y n e r v e fibre a n d t h e s y n a p t i c d e l a y in t h e cochlear nucleus (see Fig. 7). T h e p r e d i c t i o n a < Lm < 6 q- 2.5 msec has b e e n t e s t e d for a n u m b e r of neurons. I t a p p e a r s t h a t t h e m i n i m u m l a t e n c y , Lm, t o t o n e b u r s t s exceeds a. A f u r t h e r r e s u l t is t h a t n e u r o n s w i t h i r r e g u l a r firing p a t t e r n s g e n e r a l l y h a v e lower L m - - a v a l u e s t h a n n e u r o n s w i t h a m o r e r e g u l a r firing p a t t e r n . I n some DON n e u r o n s t h e l a t e n c y to t o n e b u r s t s exceeds t h e p r e d i c t e d u p p e r l i m i t 6 q- 2.5 msec. I n t h e results section a m o d e l is p r e s e n t e d to e x p l a i n these findings. Methods The results were obtained from eats under nembutal anaesthesia. The methods used for the preparation of the animal, recording of single unit activity, generation and application of stimuli, data collection and histological control of the anatomical location of neurons are described in the preceding paper (van Gisbergen et al., 1975a). A large number of neurons have been stimulated with CF tone bursts having an intensity level usually 20--30 dB above threshold. Only neurons showing an activation response will be considered. The data from experiments, which has been stored on digital magnetic tapes, were used for two computations.

Responses of CN Neurons to Tone Bursts

411

Latency Distribution o/First Spike in the Response The distribution of the latency of the first spike occurring in the time window extending between Lm and D -k Lm was computed from the responses to a large number of tone burst stimuli (usually 500) with duration D. In this notation Lm represents the point in time following stimulus onset where the firing probability begins to deviate significantly from the level of base line activity or, in other words, the minimum latency (as opposed to mean latency). Lm was determined by visual inspection of high resolution PSTHs. Stimuli which did not evoke an action potential within the indicated time window have been ignored. This has been observed for instance for some neurons with a build up temporal pattern of activation. From the distribution the mean values (M), the standard deviation (S) as well as the coefficient of variation (C.V. --~ S/M) were computed.

First Interval Distribution The distribution of the intervals between the first and second spike after Lm in the response to CF tone burst stimuli was computed. Stimuli which did not evoke a first and/or a second spike in the time window indicated above were left out of consideration. From the first interval distributions the mean value and the standard deviation as well as the coefficient of variation have been computed. The minimum first interval observed in the distribution will be indicated by Ira.

Results

Characteristics o / F i r s t Spike Latency and First Interval Distributions Shapes oI Distributions T h e initial response of cochlear nucleus n e u r o n s a n d p r i m a r y ( ? ) fibres was a n a l y z e d b y c o m p u t i n g t h e first spike l a t e n c y a n d t h e first i n t e r v a l d i s t r i b u t i o n s . T h e shape of t h e d i s t r i b u t i o n s for cochlear nucleus neurons shows v a r i a t i o n s which are r e l a t e d to t h e gross t e m p o r a l p a t t e r n of response as r e v e a l e d in t h e P S T H (Fig. 2). Most of t h e A neurons h a v e e x p o n e n t i a l a s y m m e t r i c d i s t r i b u t i o n s w i t h a m o d a l v a l u e which is small c o m p a r e d w i t h those f o u n d in A S a n d A(S) n e u r o n s (Fig. 2A). These d i s t r i b u t i o n s were also f o u n d in p r i m a r y ( ? ) fibres. Generally, in low-CF u n i t s of t h e A t y p e as well as p r i m a r y ( ? ) fibres p h a s e locking was reflected in m u l t i m o d a l first spike l a t e n c y a n d first i n t e r v a l d i s t r i b u t i o n s . S i m i l a r i n t e r v a l d i s t r i b u t i o n s h a v e been described for a u d i t o r y nerve fibres w i t h a low C F u n d e r t o n a l s t i m u l a t i o n (Rose et al., 1967 ; H i n d et al., 1967). I n contrast, m o s t of t h e AS a n d A(S) s u s t a i n e d cochlear nucleus n e u r o n s d i s p l a y first spike l a t e n c y a n d first i n t e r v a l d i s t r i b u t i o n s d e v i a t i n g from those of t h e p r i m a r y ( ? ) fibres b y t h e i r m o r e n a r r o w a n d s y m m e t r i c a l shapes (Fig. 2B). Some neurons h a v i n g a c o m p l e x t e m p o r a l p a t t e r n o f a c t i v a t i o n d i s p l a y e d b i m o d a l first spike l a t e n c y d i s t r i b u t i o n s (Fig. 2C). This response p a t t e r n presuma b l y reflects s t r o n g i n h i b i t i o n a t t o n e onset w i t h a slightly longer l a t e n c y t h a n t h e e x c i t a t i o n (van Gisbergen et al., 1975a). I t should be n o t e d in passing t h a t , obviously, t h e b i m o d a l first spike l a t e n c y d i s t r i b u t i o n s o b s e r v e d c a n n o t be exp l a i n e d in t e r m s oi' a r e c u r r e n t i n h i b i t i o n originating from t h e n e u r o n itself. S t r o n g i n h i b i t i o n a t t o n e onset is also held responsible for t h e long first spike latencies o b s e r v e d in n e u r o n s w i t h a b u i l d u p t e m p o r a l p a t t e r n of a c t i v a t i o n (Fig. 2D; cf. v a n Gisbergen, 1974).

J.A.M. van Gisbergen et al.

412

first spike latency distribution

PSTH ON

OFF

1

c~

L

Ms 2.50 $ : 1.78

o otl ,o

!2~1

c

1

M : 5.54 S : 1.70

o!

first interval distribution

M : 6 64 S : 0.60

A

2'0

S =0.42

B

oo

2b

9

o

2b

o

'a

9 20] 26.8

KI-Iz

u n i t 4 0 - 1 DCN AS c o m p l e x

M :17.97 S : 9.13

c

2o

M : 12.99 = .

o

C

I i

0

5 0

o

0

21i so

loo

time often tone onset (reset)

o

l,

a_

5b

latency

50

2ot,

M = z! 1 0 Q S= 5.64

"

0

,a~,l ~b (reset)

0 ,

o

M = 9.60 S = 1.91

'

D 5b

interval

(msec)

Fig. 2. First spike latency and first interval distributions of 4 units computed from responses to CF tone burst. The PSTHs (left column) and the distributions were computed from the responses to 500 tone bursts. The number of bins is 400 in the PSTHs and 200 in the first spike latency and first interval histograms. M and S are mean and standard deviation respectively of the distribution. Note short minimum first spike latency and minimum first interval in the A sustained neuron. This neuron has typical asymmetrical first spike latency and first interval distributions. These are also characteristic o~ primary(?) fibres. Note differences in time scales

Mean, Standard Deviation and Coe~ieient o/ Variation These p a r a m e t e r s will be used to characterize first spike l a t e n c y a n d first i n t e r v a l d i s t r i b u t i o n s of various types of neurons. Most of t h e d a t a presented refer to u n i m o d a l distributions. F o r the sake of completeness the p a r a m e t e r s from b i m o d a l d i s t r i b u t i o n s (e. g. Fig. 2C), which are more difficult to characterize, have been included. Plots of the s t a n d a r d d e v i a t i o n against the m e a n of t h e first spike l a t e n c y d i s t r i b u t i o n for u n i t s h a v i n g different a n a t o m i c a l locations (Fig. 3A) a n d for t h e

Responses o~ CN Neurons to Tone Bursts

413

standard deviation (msec) []

Primary (?) fibre 10.

VCN 9 DCN []

D

o n m

z, D

m

9 m

i&-~, D , ~ .

.

.

.

9 i

.

.

.

9 .

99

"Xt

10:

.

i

.

.

.

j

.

.

.

.

o

o

B

o

(?) fibre

S CN ~ A(S)J

o

o .o

o

9

A o

o

9 o o

dA" ~ 9

9 ~

9 oA

~

~ ~

a o

o9 o

o o

o

5

10 15 9mean firstspike latency (msec)

20

Fig. 3. Plots of standard deviation against mean of the first spike latency distribution. (A) For primary(?) fibres and cochlear nucleus neurons with an indication of anatomical locations (181 units). (B) For primary(?) fibres and cochlear nucleus neurons with an indication of t y p e of response area (148 units). The data from 35 neurons (most of them in the DCN) with mean values exceeding 20 msee were not plotted

s a m e units w i t h a n i n d i c a t i o n of t h e t y p e of response areas (Fig. 3B) show interesting correlations. C o m p a r i n g neurons h a v i n g a p p r o x i m a t e l y t h e s a m e m e a n first spike l a t e n c y d e m o n s t r a t e s t h a t p r i m a r y ( ? ) fibres as well as A t y p e cochlear nucleus neurons g e n e r a l l y e x h i b i t larger s t a n d a r d d e v i a t i o n s t h a n cochlear nucleus A S a n d A(S) t y p e neurons. T h e d a t a from AS neurons in Fig. 3B i n d i c a t e t h a t t h e presence of i n h i b i t o r y i n p u t s is a c c o m p a n i e d w i t h a m o r e precisely t i m e d onset of response (cf. Fig. 2A a n d B). T h e s a m e tendencies can be o b s e r v e d s o m e w h a t m o r e clearly in Fig. 4 where similar d a t a referring to first i n t e r v a l d i s t r i b u t i o n s are presented. These d a t a allow a crude s e p a r a t i o n of t h e entire p o p u l a t i o n into two groups differing in t h e unif o r m i t y of first intervals. G e n e r a l l y t h e ensemble of first i n t e r v a l s of a given n e u r o n d i s p l a y s less r e l a t i v e s p r e a d in i n t e r v a l d u r a t i o n in t h e VCN a n d D C N n e u r o n s o f 29

~Exp.

Brain

l~es.

Vol.

23

414

J.A.M. van Gisbergen et al. standard deviation (m sec) A 10 84

Primary {?) fibre VCN 9 DCN

c~ []

[:

I

a m[J

n ii i

D z, a

~%%

9

" =

, m

l|

9

9 9 ";

,

9 9

~i I

,

i9 I

I

i

k

9

i

i

m

m

I I

o

Primary (?) fibre

B

10 ~ A(S} J a

o

o

o o

o 9

A 9

o

r

5

o

~o

9

10

15

20

m e a n first interval (msec~

Fig. 4. Plots of standard deviation against mean o~ the first interval distribution. (A) For primary(?) fibres and cochlear nucleus neurons with indication of anatomical location (192 units). (B) For primary(?) fibres and cochlear nucleus neurons with indication of type of response area (165 units). The data of fifteen neurons with mean values exceeding 20 msec were not plotted

the AS and A(S) type than in primary(?) fibres and VCN neurons of the A type. There is a correlation, for some groups of neurons, between the coefficient of variation (C.V.) values from the first spike latency and the first interval distributions. This means t h a t the spread in first intervals is correlated with the precision of timing of the first action potential in the response (Fig. 5A and B). This correlation is most clear for VCN neurons and primary(?) fibres. The VCN neurons with small C.V. for both distributions are generally of the AS or A(S) type. Neurons with larger C.V. values for both distributions are most often VCN A type neurons or primary(?) fibres. A group of neurons which tends to obscure the overall positive correlation between C.V. values derived from first interval and first spike latency distributions is formed by those displaying a complex temporal pattern of activation, most often found in the DCN. These neurons have a C.V., derived from the first spike

l~esponses of CN Neurons to Tone Bursts

415

C.V.first spike tatency distribution

1,0. ~ Primary (?) fibre oVCN 9 DCN

~ D aD

0,5

~a~

D

o

1,0 9 Primary (?) fibr~ .~AAS l CN

~

~ 9

0,5

A(S) J

9

o

9

~

9 / :

o

o

'AA O 9 9

0,5

9 A

9149 OA , ~

~

9 9 9

1,'0 C.V.first interval distribution

I,'5

2,0

Fig. 5. Plots of the coefficient of variation (C.V.) of the first spike latency distribution against the C.V. computed from the first interval distribution. (A) For primary(?) fibres and cochlear nucleus neurons with indication of anatomical location (215 units). (B) For primary(?) fibres and cochlear nucleus neurons with indication of type of response area (176 units)

latency distribution, which may be large compared with the C.V. computed from the first interval distribution (Fig. 20). Other DCN neurons have smaller C.V. values for both distributions which overlap with those of AS and A(S) VCN neurons. The correlation observed between the C.V. values computed from both distributions is an important finding since it indicates that the same processes may be operative during the generation of the first as well as the second action potential in the response. The generally rather uniform first intervals in VCN neurons of the AS and A(S) type, reflected in an oscillating PSTH (Fig. 2B), may be explained by the Goldberg et al. (1964) model, which assumes threshold elevation after the generation of an action potential. Alternatively these effects may be explained by the Markov process model of Molnar and Pfeiffer (1968) where temporal integration of the irregular firing inputs takes place. 29*

416

J.A.M. van Gisbergen et aI.

However, the generally rather precise timing of the first action potential in the response, observed in the neurons displaying uniform first intervals, could be anticipated in terms of the temporal integration model (Molnar and Pfeiffer, 1968) but cannot be explained by the recovery properties of the neuron (Goldberg et al., 1964), as has been stated in the introduction. Therefore, we think t h a t temporal integration underlies the generally more narrow and symmetrical first spike latency and first interval distributions of m a n y cochlear nucleus neurons in comparison with the primary(?) fibres. This does not mean, however, t h a t a t e m p o r a r y threshold elevation after spike initiation is excluded.

Comparison o/Experimental and Predicted Lm Values I f the CCF, I~(T), computed from the responses to stationary noise stimulation, is interpreted in terms of a simple model (see Introduction) a relation m a y be expected between the time period during which R(r) deviates significantly from zero (a < ~ < 3) and the minimum latency, Lm, of the same neuron to stimulation with supra-threshold tone bursts. When using tone bursts with a 2.5 msec rise time it was predicted t h a t a < Lm < ~ ~- 2.5 msec. This prediction gives a relation between temporal aspects of the stimulus-response transfer characteristics under stationary noise and non stationary tonal stimuli. The prediction is tested for primary(?) fibres and cochlear nucleus neurons with various response patterns to tonal stimuli (Fig. 6A, B). Several interesting conclusions m a y be drawn from the results : _First, all neurons appear to have latencies which exceed a (positive L m - - a values). I f our interpretation of a as a pure time delay is correct it means t h a t the latency to tone burst stimulation is also determined b y other processes which can lengthen the latency greatly. I n particular, the large spread in L m - - a values is intriguing. I n one neuron we found an L m - - a value of 80.3 msee. I t should be noted t h a t the spread in a values is small (Fig. 6). I t is also striking to observe t h a t units with a primarylike response pattern (which are usually identical with A type neurons) and primary(?) fibres generally have lower L m - - a values t h a n neurons with different CF tone burst P S T H shapes such as chopper neurons. This is exactly what is to be expected when temporal integration is assumed to be the underlying mechanism of chopping in AS and A(S) type neurons whereas this mechanism is believed to be virtually absent in A type neurons. We are all the more inclined to relate these differences in L m - - a values between primarylike and chopper neurons to temporal integration in chopper neurons because the only obviously deviating primarylike unit in Fig. 6A deviates also from the other primarylike units in t h a t it exhibits first spike latency and first interval distributions which are more similar to those of chopper neurons. Second, most of the neurons have latencies Lm < ~ -J- 2.5 msec. As these neurons have in addition positive L m - - a values they obey the prediction t h a t a < Lm % 5 -J- 2.5 msec. Three A(S) neurons, one with a chopper and two with a build up temporal pattern, had latencies Lra > ~ @ 2.5 msec. These neurons were found in the DCN. Model to Account/or Di~erenees in Lm--a Values D a t a p r e s e n t e d in the previous section (Fig. 6A and B) indicate t h a t in none of the neurons the latency of response to tone bursts can be attributed solely to time delays in the system. I n primary(?) fibres and cochlear nucleus primarylike neurons the L m - - a value is smaller t h a n in neurons with a chopper or build up

Responses of CN Neurons to Tone Bursts

417

LM{msec) A

LM-,Z.= 10 8 6 /~ 2

0 msec

15-

/

, oOpri~t:trrYm{~fi bir~/

10- /

/

/

201

B

LM-~:= 10 8 6 4

2

Omsec

15-

10-

i~y(?.) fibre

4 / 1(]

15

2'0 ~. (msec)

:Fig. 6. Plots of m i n i m u m latency (Lm) to tone burst stimulation against pure time delay estimate (a) derived from CCF. (A) For primary(?) fibres and cochlear nucleus neurons with different temporal patterns of response to tone bursts. One unit could not be subdivided into one of the categories (27 units). (B) For primary(?) fibres and cochlear nucleus neurons with different types of response area (24 units). One unit (A(S) build up type) having an Lm value of 85 msec a n d an a value of 4.7 msec has not been plotted

response pattern. We propose here a model (Fig. 7) to account for the observed differences. I t is assumed t h a t the latency of response to tone bursts is composed of time delays plus integration times. The various time delays in the system are indicated b y thin lines in Fig. 7, integration times by thick lines. I t is further assumed t h a t cochlear nucleus neurons have a time delay which is one synaptic delay longer than in the auditory nerve fibres. The time delay caused by the travel time of the travelling wave in the cochlea m a y be dependent on the frequency of the tone. I n this model (Fig. 7), however, Lm and a both comprise the time delays in an identical way, so Lra--a only contains the integration times. Therefore a possible dependence of the travel time on the frequency of the tone is eliminated from Lm--a.

418

J. A.M. van Gisbergen et al.

components contributing t o latency of response to tone bursts ill

2

134~

5

nature of inputs

_~

post stimulus t~me h i s t o g r a m s

CCF

E

y~

E E§

~.16 ~time

7

E+]

after tone onset ~

~

time after tone onset--

t y p e of L M -(x=

~time

neuron

4

auditory nerve fibre

4

sustained p r i m a r y like

4+7

sustained chopper

4+7

build up

before spike

]Pig. 7. Model to account for differences in latency of response to tone bursts and time delay estimations derived from CCF. Components contributing to latency: (A) Time delays (indicated by thin lines) : 1. Acoustic delay from sound source to tympanic membrane. 2. Delay caused by travelling wave in the cochlea. 3. Synaptie delay in the cochlea. 5. Conduction time of spikes along auditory nerve fibre. 6. Synaptic delay in synapses on cochlear nucleus neurons. (B) Integration time periods (indicated by black bands) : 4. Integration time in cochlea (To). 7. Minimum integration time in cochlear nucleus neuron (Ten) I t is assumed t h a t some temporal integration is necessary to reach threshold in a u d i t o r y nerve fibres. I t is further proposed t h a t the firing patterns of A t y p e neurons with primarylike response p a t t e r n can be accounted for b y Moinar and Pfeiffer's (1968) superposition model. Thus, in these neurons no further appreciable temporal integration is assumed to occur. On the other hand, the typically higher L m - - a values of AS and A(S) neurons with a chopper response p a t t e r n are attributed to a mechanism of temporal integration. The v e r y large L m - - a yMues obtained in some of the build up t y p e neurons are a t t r i b u t e d to interaction of excitatory and inhibitory inputs with slightly different time constants of d e c a y (cf. model in v a n Gisbergen et al., 1975a). Because the inhibition is p r e s u m a b l y m u c h weaker in AS chopper neurons threshold is reached m u c h quicker in these neurons. Further A n a l y s i s o / I n t e g r a t i o n T i m e

The model in l~ig. 7 enables us to envisage the components contributing to L m - - a values of neurons with a primarylike and a chopper response pattern. As

indicated in Fig. 7 the L m - - a values of p r i m a r y ( ? ) fibres and primarylike cochlear nucleus neurons would be expected to be identical. This is based on the assumption t h a t no appreciable temporal integration occurs in the cochlear nucleus primarylike neuron. I n other words, Ten, the m i n i m u m integration time in the cochlear nucleus, equals zero f o r these neurons. I f this assumption is correct the L m - - a values of p r i m a r y ( ? ) fibres and primarylike neurons can be used to estimate the hypothesized integration time T c in the cochlea. W h e n the one deviating primarylike n e u r o n is left out of consideration we arrive, b y averaging the L m - - a values of primary(?) fibres and primarylike neurons, at a m e a n of 1.7 msee as the estimate for the integration time in the cochlea under our stimulus conditions. According to the model the L m - - a value of chopper neurons is composed of the sum of two integration time periods. Lm--a = Tc-bTcn or taking T c ~ 1.7 reset: Lm--a ~ 1.7~-Tcn

Responses of CN Neurons to Tone Bursts

419

LM- .c (msec) 10 t

LM - ~; =1,7+ ]M

o o [] []

9

, primary (?) fibre o primary like} [] chopper CN

oo

minimum first interval (msec)

Fig. 8. Relation between minimum first interval and Lm--a value for primary(?) fibres and for primarylike and chopper cochlear nucleus neurons for 23 units. One unit with a minimum first interval of 12 msec and an Lm--a value of 8.5 msec was not plotted

Assuming that the same process underlies the generation of the first and the second spike the shortest interval observed in the first interval distribution (Ira) is an estimate of Ten. Accordingly it would be expected that L m - - a ---- 1.7 -F Ira. This prediction was tested for the 8 chopper neurons where sufficient data were available. I t can be observed (Fig. 8) that a positive correlation exists between L m - - a and the minimum first interval Ira. Primary(?) fibres and primarylike neurons generally have short minimum first intervals and do not show such a correlation. I t is interesting to note, however, that the primarylike unit with an exceptionally large Lm--a value also has a longer minimum first interval. I n highfrequency neurons a values were not available. Cochlear nucleus neurons with a CF > 3 kHz presumably have rather similar pure time delays (van Gisbergen et al., 1975b). Accordingly, differences in latency among these neurons will predominantly reflect differences in integration time. I n Fig. 9 minimum latency of response to tone bursts has been plotted against the minimum first interval for units with different anatomical locations and for units with different types of response area, in all cases with CFs > 3 kHz. The data in Fig. 9A indicate that extensive integration of input signals occurs in almost all of the DCN neurons. I n the VCN the extent of integration is generally less than in the DCN. Comparison with the primary(?) fibre data suggests that in the VCN neurons of the A type evidence for temporal integration is barely

420

J.A.M. van Gisbergen et al. i L M (msec) 15-

iii

9

o o

10

9

F

Es ~ u

o

,

[3 U 9 1 4 9

I19

9

u ~D 9

o u9

o

o9

"~" (3{3 i9

o

9

o

o 9

~

o ee

Oo

~ o ~162~

u

ii 9149 9149

9

~

o

{31

9

9

Oe

#;o D

Primary (?) fibre VCN

o 9 DCN

o

9 Primary (?) fibre

9'~" ~,o

9 AS o A(S)

.

5

~

,~~ '~'

10

.

.

.

,

.

.

CN

.

.

,

5 t0 minimum first interval (msec)

Fig. 9. Plots of minimum latency against minimum first interval. (A) For primary(?) fibres and cochlear nucleus neurons with indication of anatomical location (163 units). (B) For primary(?) fibres and cochlear nucleus neurons with indication of type of response area (135 units). Neurons with a CF below 3.0 kHz were left out of consideration. Neurons with minimum latencies exceeding 15 msec (13 units) or with minimum first intervals exceeding 11 msec (3 Units) were not plotted

demonstrable (Fig. 9B). The VCN neurons of the AS type have longer minimum latency and a longer minimum first interval which indicates temporal integration. Figure 9B strongly suggests that extensive temporal integration is associated with the presence of inhibitory inputs. The positive correlation between minimum latency and minimum first interval observed for the AS neurons (Fig. 9B) corroborates the temporal integration hypothesis. The large spread is caused by neurons with a pause or a build up temporal pattern of response. Discussion

I n this paper and in a number of previous studies of cochlear nucleus neurons such aspects of neuronal function as the frequency/intensity dependence of the response pattern, the temporal pattern of the response, the statistical distribution of interspike intervals and the quality of phase locking were studied (Greenwood and Maruyama, 1965; Evans and Nelson, 1973; Pfeiffer, 1966; van Gisbergen, 1974; Goldberg and Greenwood, 1966 ; Goldberg and Brownell, 1973 ; Lavine, 1971). I t is striking to observe that out of many combinations of properties only a few are repeatedly observed experimentally. That is, certain combinations of properties are often found to occur jointly, others are rare.

Responses of CN Neurons to Tone Bursts

421

Two of these combinations of properties may be illustrated by comparing characteristics of A and AS type neurons. 1. The temporal pattern of response of A neurons is always sustained whereas in the AS neurons more complicated temporal patterns of response, probably due to interaction of overlapping excitatory and inhibitory inputs (van Gisbergen et al., 1975a), may be observed. 2. The response latency is appreciably shorter in A neurons than in AS neurons (Figs. 6B and 9B). This difference cannot be accounted for by differences in time delays (Fig. 6B). 3. Whereas A type neurons exhibit irregular firing patterns (Goldberg and Brownell, 1973; this paper), the firing pattern of AS neurons tends to be more regular. This regular firing pattern is reflected in many AS neurons with a sustained temporal pattern of response in regularly spaced peaks in the P S T H computed from sound-burst responses. A positive correlation was found in these neurons between the minimum latency and the minimum first interval. 4. The quality of phase locking of auditory nerve fibres is largely retained in the cochlear nucleus A type neurons but is generally considerably impaired in AS neurons (van Gisbergen et al., 1975b; Goldberg and Brownell, 1973). The retention of auditory nerve fibre properties in A type neurons, also noticed by Goldberg and Brownell (1973), probably reflects absence of integration of synaptic inputs. On the other hand, the initial response characteristics of many of the AS neurons are interpreted by us as largely corroborating the hypothesis that the response depends on temporal integration of a relatively large number of inputs. Integration of a large number of inputs may also explain the deterioration of phase locking in these neurons. Neu.rons with the highest maximum firing rates found in the cochlear nuclei (up to 600 spikes/sec, measured in 10--15 msec intervals) appear to have the chopper response pattern. I f the temporal integration hypothesis is correct, it is imperative to assume that strong convergence of excitatory inputs takes place in these neurons. Although electron microscopic studies are needed to reach a more definite conclusion it seems that many auditory nerve fibre endings terminate on multipolar cells, octopus cells (Osen, 1970), giant and pyramidal cells (Cohen et al., 1972) but not on large and small spherical cells (Osen, 1970). In the VCN region where multipolar cells are found we have often encountered the chopper response pattern (cf. Godfrey, 1972). But chopping is almost certainly not exclusively associated with one single cell type. I t is also encountered in other parts of the cochlear nuclei and has also been observed in the superior olivary complex (Boudreau and Tsuehitani, 1968 ; Clark and Dunlop, 1969 ; Guinan et al., 1972a, b). In view of the often rather narrow and symmetrical first spike latency and first interval distributions we are inclined to think that temporal integration also takes place in DCN neurons. The data in Fig. 9 lead us to believe that this process occurs here to an even larger extent than in the VCN. Direct evidence that temporal integration may take place in cochlear nucleus neurons is provided by intraeellular recordings (Gerstein et al., 1968). One of their recordings (see their Fig. 7) was obtained from a neuron with a regular firing

422

J.A.M. van Gisbergen et al.

p a t t e r n u n d e r t o n e b u r s t s t i m u l a t i o n . T h e t h r e s h o l d was r e a c h e d w i t h a g r a d u a l l y d e v e l o p i n g m e m b r a n e d e p o l a r i z a t i o n which does n o t clearly r e v e a l i n d i v i d u a l E P S P s . T h e r a t e o f m e m b r a n e d e p o l a r i z a t i o n is fast near t h e CF a n d m a r k e d l y slower at less effective frequencies.

Acknowledgements. This study was supported by the Netherlands Organization for the Advancement of Pure Research (Z.W.O.). Thanks are due to H.J. Krijt and J.W.C. Braks for skilled technical assistance.

References Boudreau, J.C., Tsuchitani, C. : Binaural interaction in the cat superior olive S segment. J. Neurophysiol. 31, 442~454 (1968) Clark, G.M., Dunlop, C.W.: Poststimulus-timc response patterns in the nuclei of the cat superior olivary complex. Exp. Neurol. 23, 266--290 (1969) Cohen, E.S., Brawer, J . g . , Morest, D.K. : Projections of the cochlea to the dorsal cochlear nucleus in the cat. Exp. Neurol. 35, 470-479 (1972) Evans, E.F., Nelson, P. G. : The responses of single neurons in the cochlear nucleus of the eat as a function of their location and the anaesthetic state. Exp. Brain Res. 17, 402--427 (1973) Gerstein, G.L., Butler, R.A., Erulkar, S. D. : Excitation and inhibition in cochlear nucleus. I. Tone-burst stimulation. J. Neurophysiol. 31, 526--536 (1968) Gisbergen, J.A.M. van: Characterization of responses to tone and noise stimuli of neurons in the cat's cochlear nuclei. Ph.D. Thesis, Lab. of Medical Physics and Biophysics, Univ. of Nijmegen, Nijmegen 1974 Gisbergen, J.A.M. van, Grashuis, J.L., Johannesma, P.I.M., Vendrik, A .J.H .: Spectral and temporal characteristics of activation and suppression of units in the cochlear nuclei of the anaesthetized cat. Exp. Brain ges. 23, 367--386 (1975a) Gisbergen, J.A.M. van, Grashuis, J.L., Johannesma, P.I.)/[., Vendrik, A.J.H. : Neurons in the cochlear nucleus investigated with tone and noise stimuli. Exp. Brain Res. 23,387--4:06 (1975b) Godfrey, D.A. : Localization of single units in the cochlear nucleus of the cat: An attempt to correlate neuronal structure and function. Doctoral Dissertation, Department of Physiology, Harvard University, )/[.A. 1972 Goldberg, J.M., Greenwood, D.D.: Response of neurons of the dorsal and posteroventral cochlear nuclei of the cat to acoustic stimuli of long duration. J. Neurophysiol. 29, 72--93 (1966) Goldberg, J.M., Brownell, W. E. : Discharge characteristics of neurons in anteroventral and dorsal cochlear nuclei of cat. Brain Res. 64, 35--54 (1973) Greenwood, D.D., Maruyama, N. : Excitatory and inhibitory response areas of auditory neurons in the cochlear nucleus. J. Neurophysiol. 28, 863--892 (1965) Guinan, J.J., Jr., Guinan, S.S., Norris, B.E. : Single auditory units in the superior olivary complex. I. Response to sounds and classifications based on physiological properties. Int. J. Neurosci. 4, 101--120 (1972a) Guinan, J.J., Jr., Norris, B.E., Guinan, S.S. : Single auditory units in the superior olivary complex. II. Locations of unit categories and tonotopic organization. Int. J. Neurosci. 4, 147--166 (1972b) Hind, J.E., Anderson, D.J., Brugge, J.F., Rose, J.E. : Coding of information pertaining to paired low-frequency tones in single auditory nerve fibers of the squirrel monkey. J. Neurophysiol. 30, 794--816 (1967) Kiang, N., Watanabe, T., Thomas, E. C., Clark, L. F. : Discharge patterns of single fibers in the cat's auditory nerve. Cambridge, Mass. : M.I.T. Press 1965 a Kiang, N. : A survey of recent developments in the study of auditory physiology. Ann. otorhino-laryng. 77, 656--676 (1968) Lavine, R.A. : Phase-locking in response of single neurons in cochlear nuclear complex of the cat to low-frequency tonal stimuli. J. Neurophysiol. 34, 467--483 (1971)

Responses of CN Neurons to Tone Bursts

423

Molnar, Ch. E., Pfeiffer, R.R. : Interpretation of spontaneous spike discharge patterns of neurons in the cochlear nucleus. Proc. I.E.E.E. 56, 993--1004 (1968) Osen, K . K . : Course and termination of the primary afferents in the cochlear nuclei of the cat. An experimental anatomical study. Arch. ital. Biol. 108, 21--51 (1970) Pfeiffer, R. R. : Classification of response patterns of spike discharges for units in the cochlear nucleus: tone-burst stimulation. Exp. Brain Res. l, 220--235 (1966) Rose, J. E., Brugge, J. F., Anderson, D.J., Hind, J. E. : Phase-locked response to low-frequency tones in single auditory nerve fibers of the squirrel monkey. J. Neurophysiol. 80, 769--793 (1967) Dr. J.A.M. van Gisbergen Laboratory of Medical Physics and Biophysics University of Nijmegen Geert Grooteplein Noord 21 Nijmegen The Netherlands