Hearing Research, 57 (1991) 45-56 0 1991 Elsevier Science Publishers

HEARES

45 B.V. All rights reserved

03785955/91/$03.50

01658

Maturational

aspects of periodicity coding in cat primary auditory cortex Jos J. Eggermont

Behavioral Neuroscience Research Group, Department of Psychology, The University of Calgary, Calgary, Alberta, Canada (Received

17 April 1991; accepted

22 July 1991)

The click-following responses for single units in the primary auditory cortex of the cat were explored as a function of age. Recordings were obtained in kittens from 9-53 days of age and assembled in four age groups; lo-15 days, 16-21 days, 22-27 days and 30-60 days. Age group means were compared to results obtained in adult cats. The stimulus consisted of one secondlong click trains presented every three seconds with click rates ranging from l-32 clicks per second. The response was characterized by entrainment, rate Modulation Transfer Function (rMTF), vector strength (VS) and temporal Modulation Transfer Function (tMTF). Maturational effects on periodicity coding comprised changes in overall responsiveness as well as click-rate dependent changes. The number of spikes elicited by single stimuli increased on average 3-fold between the second post-natal week and adulthood, probably as a result of more efficient synapses in the central auditory pathway and some improvement in thresholds. Adaptation became less pronounced with age; neurons started to respond to the later clicks in the 8/s and 16/s click trains from the third post natal week on. By the end of the first post-natal month the click following responses resembled the adult ones qualitatively, however, increased firing rates and spontaneous rates together with rebound responses continued to produce quantitative differences between the 30-60 day olds and the adults. Limiting rates for the tMTF (50% of the response at l/s) increased from 6 Hz in the lo-15 day old to 12 Hz in adults. The decrease in the duration of the post-activation suppression coupled with the increased response with age to trains with higher click rates suggested that the maturation of inhibitory processes in the cortex play a major role in this rate dependence. Cat; Auditory

cortex;

Single unit; Maturation;

Click-train

stimuli;

Modulation

Introduction Several response properties in the auditory cortex of altriceous animals like the rat and cat are apparently established very early on in post natal life, before the animal has had extensive auditory experience and even before the cortical circuits have matured structurally. The maturation of these cortical response properties seemed to follow the maturation of the auditory periphery and brainstem with a delay of only a few days (Brugge, 1988; Brugge et al., 1981; Kettner et al., 1985). Among these early established properties are the response threshold, the shape and sharpness of the frequency-tuning curve, the rate-intensity function, and the interaural-intensity and interaural-phase difference responses of individual cortical cells. Nearly all these properties are mature at the level of the brainstem as well as in the cortex at the end of the first post-natal month. Frequency tuning in the inferior colliculus (ICI of the cat for instance is mature by the 26-30th postnatal day (Moore and Irvine, 1979). The shape of the interaural time difference response curve in the audi-

Correspondence to: Jos J. Eggermont, Department The University of Calgary, 2500 University Drive Alberta, Canada, T2N lN4. Fax: (403) 2828249.

of Psychology, N.W., Calgary,

transfer

functions;

Entrainment;

Vector

strength

tory cortex of the cat showed adult form at post-natal day 18 and the interaural intensity difference curve was already adult like at post-natal day 13. Brugge (1988) has called this phenomenon the functional constancy of the auditory system with respect to development and maturation. The neuron properties that show this functional constancy are related to hearing sensitivity, frequency selectivity and sound localization, i.e., to those properties that are well established at the brainstem level. And it appears that the auditory cortex retains these response properties. Auditory cortex specialization, beyond simply representing what is going on at a lower level, most likely reflects the communicative demands posed by the temporal or envelope properties of species specific vocalizations (Ribaupierre et al., 1972; Schreiner and Langner, 1988; Phillips and Farmer, 1990). It seems therefore that the study of periodicityor envelope-coding would reflect upon a specific cortical phenomenon, mainly determined by the pronounced hyperpolarizations (either post firing or by stimulus induced IPSPs) so typical of pyramidal cells (Creutzfeldt et al., 1980). In a previous paper we have described the coding of click trains in the auditory cortex of the adult cat (Eggermont, 1991). Our findings were that click following responses had limiting rates below 24 Hz. However, the limiting rate was very much dependent on the type

4h

of measure used to characterize the data. Synchrony measures such as vector strength typically resulted in a band-pass dependence on click rate with best modulating frequencies (BMF) around 8 Hz. Rate measurements such as entrainment resulted in low-pass functions of click rate with a limiting rate of 8 Hz. These data agreed qualitatively with the entrainment functions obtained for repetitive tone-pip stimulation of Phillips et al. (1989). We argued that most likely our recordings were exclusively from pyramidal cells. The development and maturation of phase-locking has been well studied in auditory nerve and anteroventral cochlear nucleus (Brugge et al., 1978, 1981; Kettner et al., 198.5). Typically the cochlear nucleus lags behind in development for phase locked responses for about a week with respect to the auditory nerve. To my knowledge, studies on the maturation of click-following responses or for that matter of periodicity coding, at any level of the auditory nervous system, have been limited to one study comparing stimulus following for evoked potentials in auditory nerve and inferior colliculus (10 of the mouse (Sanes and Constantine-Paton, 1985). This study showed that click following matured more rapidly in the periphery (mature at post natal day 18) than in the inferior colliculus which became adult-like around day 30. It has been reasonably well established that the limits of phase-locking, where the carrier waveform periodically depolarizes and hyperpolarizes the hair cell, are determined largely by synaptic mechanisms related to the release of transmitter substance (Javel et al., 1988). In case of amplitude modulation or repetitive click stimulation some of the mechanisms responsible are most likely again of synaptic origin and related to transmitter depletion and transmitter-receptor binding kinetics (Eggermont, 1973, 1985a; Smith, 1988). Thus the number and efficacy of synapses is likely to play a major role in the maturation of periodicity coding. No data on synaptic maturation are available for the auditory cortex but in cat visual cortex the number of synapses is still low in the first post-natal week and then increases to a maximum about 6 weeks after birth. By post-natal day 7 in the rat the number of synapses per cortical neuron reaches about 14% of adult levels, with the number of vesicles per synapse only 20% of that in maturity. This makes the overall synaptic capacity at the end of the first post-natal week only 3% of that in adult cortex (Armstrong-Jones and Fox, 1988; Payne et al., 1988) and suggests that available synaptic transmitter stores could be quickly exhausted by synchronous activation because of slow replenishment of transmitter substance. However, in kittens at post-natal day 6, which are relatively more mature than rats at the same age, spike activity from a few somatosensory cortex cells was able to follow repetitive ventrobasal complex stimulation at rates of

5-20/s for 1 second and thus suggesting a high synaptic security for the thalamo-cortical link at this age (Armstrong-Jones and Fox, 1988). From our results in adult cat cortex (Eggermont, 1991) it became obvious that the limiting rate of clickfollowing was also determined by the duration of the post-activation suppression. This had been suggested previously by Ribaupierre et al. (1972) on the basis of intracellular recordings in AI cells. Intracellular recordings from pyramidal cells in rat sensorimotor cortex (Purpura et al., 1965) suggested that IPSPs were much longer in neonatal cortex (200-600 ms> than in adult cortex (< 100 ms). Evidence that such post activation hyperpolarization limits neural responsiveness in the thalamus and inferior olive has been reviewed by Steriade and Llinas (1988). Mechanisms that might be responsible for limiting the firing rate and click-following capacity include a calcium-activated potassium conductance and a transient outward potassium current (the A-current). If the duration of the post-activation suppression is related to the duration of the inhibitory post synaptic potentials (IPSPs) and this duration is a limiting factor in the click-following capacity of cortical cells one expects that the limiting rate will be lower by up to a factor 6 in neonatal cats than in adult animals. By studying the periodicity following one second duration click trains WC aim at estimating both the effect of the post-activation suppression and that of adaptation. Equating post-activation suppression with IPSPs and membrane mechanisms. and adaptation with synaptic mechanisms one may expect that the study of maturation of periodicity coding can provide insight into the development of pyramidal cell and local circuit mechanisms. It will be of interest to see whether these pyramidal cell and local circuit mechanisms in primary auditory cortex mature at the same time as the functionally constant properties, such as those related to hearing sensitivity, frequency tuning and directional hearing which are basically established at more peripheral levels, or take longer and follow a pattern that is more typical for cortical neurons in general.

Methods Animal preparation Kittens were premeditated with 0.25 ml/kg body weight of a mixture of 0,l ml Acepromazine (0.25 mg/ml) and 0.9 ml of Atropine sulphate (0.5 mg/ml) subcutaneously. After about one-half hour they received an intramuscular injection of 25 mg/kg of Ketamine (100 mg/ml) and 20 mg/kg of Sodium pentobarbital (65 mg/ml) intraperitoneally. Tolerance for these drugs was less in the very young animals t < three

47

weeks of age) and for these animals the ketamine and pentobarbital injections were given in two steps; 75% of the dose was given in the first injection and the remainder about half an hour later. Additional acepromazine/atropine mixture was administered every 2 h, light anesthesia was maintained with intra-muscular injections of 2-5 mg/kg/hour of ketamine. The wound margins were infused every 2 h with durocain and also every 2 h new mineral oil was added if needed. The temperature of the kitten was maintained at 38 o C with a thermostatically controlled electric blanket (Harvard Medical Systems). Acoustic stimulus presentation Stimuli were presented from a speaker (Fostex RM 765) placed with its center 50 cm in front of and oriented perpendicular to an imaginary line through the animals auditory meati. The sound-treated room was made anechoic for frequencies above 625 Hz by

covering walls and ceiling with acoustic wedges (Sonex 3”) and by covering exposed parts of the vibration isolation frame, equipment and floor with wedgematerial as well. Calibration and monitoring of the sound field was done by placing a B and K (type 4134) microphone facing the loudspeaker above the animal’s head. The characteristic frequency (CF) and tuning curve of the individual neurons were determined with random frequency tone-pips presented once per second. After the CF was determined, a click-train was presented every three seconds. Alternately click trains with rates of 1,2,4, 8, 16 and 32 clicks per second were presented in that order. This sequence of six click trains lasting 18 s was repeated 50 times resulting in a total stimulus ensemble duration of 900 s. Fig. lh shows the composition of the stimulus ensemble as used in the majority of our experimental series, in the initial experiments we used a similar stimulus ensemble but without the clicks at the 1 second mark. The

H 32 ,I6 pa pL x2 .u J 1 0:s

timels) Fig. 1. Multi-unit dot displays for the first 1.1 second after the onset of the one-second duration click trains for age 43 days (a), 38 days (b), 27 days Cc), 22 days Cd), 18 days (e), 15 days (f) and 14 days (g) as well as the timing of the click-trains. In (a) a four cell recording (2 on each electrode) is shown for a 43 day old kitten, all units were spontaneously active and showed similar click following properties. (b) shows the dot display for a five cell recording (3 units on one electrode plus 2 on the other) in a 38 day old kitten, all units were spontaneously active. (c) shows a five cell recording (3 units on one electrode plus 2 on the other) in a 27 day old kitten with all units spontaneously active. (d) shows a three unit (2 plus 1) recording in a 22 day old kitten. (e) shows responses for a four unit recording (2 plus 2) in an 18 day old kitten, the short latency unit was lost somewhere halfway the record. (f) shows results for a triplet recording (a single electrode recording) in a 15 day old kitten. (g) shows responses from a 14 day old kitten; in this four cell record (3 units on one electrode plus 1 on the other).

individual clicks were 0.1 ms rectangular electric pulses; the maximum click, and tone-pip stimulus, level was 105 dB p.e. SPL measured at the animals head. Recording procedure Two tungsten micro electrodes (Micro Probe Inc) with impedances between 1.5-2.5 MR were independently advanced perpendicular to the AI surface using remotely controlled motorized hydraulic microdrives (Trent-Wells Mark III). Tip separation of the microelectrodes at the surface was within 300-500 pm. The electrode signals were amplified using extracellular preamplifiers (Dagan 2400), filtered, to remove evoked field potentials, between 100 Hz (Kemo VBFS, highpass, 24 dB/oct) and 3 kHz (6 dB/oct, Dagan roll-off). The signals were sampled through 12 bit A/D convertors (Data Translation, DT 2752) into a PDP 11/53 micro computer, together with a timing signal from two Schmitt-triggers. In general the recorded signal on each electrode contained activity of more than one

neural unit. The PDP was programmed to separate these multi-unit spike-trains into single-unit spiketrains using a maximum variance algorithm (Eggermont, 1990). The unit code plus the time of the spike occurrence were sent to the MicroVax II which presented on a Vectrix graphics processor an on-line color coded multi-unit dot display organized per frequency or click rate, depending on the stimulus. The boundaries of the primary auditory cortex were explored by taking a series of evoked potential (EP) and multi-unit measures (with the high-pass filter set at 30 Hz with 6 dB/oct) from caudal to rostra1 and assuring that there was a gradual increase in CF, which after a region with no responses to tones (because of our limited high frequency range) reversed in direction. These boundaries were indicated on a drawing of the cortical surface showing the location of the blood vessels and sulci. From this map we estimated the desired CF location for our recordings and inserted the electrodes in that location.

C 32

16

!““‘_!““‘_1”“‘~““‘_ --ccm-:

i:

: : :

w’!

click

number

-

:::::

: : : : :

: : : :

“‘j’

:::::

clrck

::

:::

:::::

:::

““’

:::::

::

:-4+-H-+----

number

“‘A

:::::

-

Fig. 2. PST-histograms for single units in a 43 day old (a), 22 day old tb), 18 day old Cc) and 15 day old Cd). The PST histograms are calculated over 6.5 ms and are arranged in the same order as the responses in the dot displays: the top row represents responses for the 32/s click trains, the second row for the 16/s train etc. The histograms in each particular row represent the responses to subsequent clicks in the click trains. Extreme values for the counts in the histograms are indicated, all histograms for a particular single-unit are scaled to that extreme.

49

Further details about preparation, stimulation and recording procedures, identical to those described previously, can be found in Eggermont (1990, 1991). Data analysis

A rapid overview of the click following properties of the neurons involved was obtained from the multi-unit and single-unit dot displays (e.g., Figs. 1 and 9). Subsequently for each of the clicks in a train a PST histogram was made for a duration of 65 ms post clickonset using 1 ms bins, in case of the 32 Hz click stimulus the histogram contained the response to two clicks. For each histogram a distinction was made between first spikes to a click (the latency histogram) and all spikes to the click (the common PST histogram). For all PST- and latency-histograms the number of spikes, the modal (peak) latency and the entrainment (the ratio of the number of first spikes in the histogram to the number of clicks presented) were calculated. The single-unit PST- and latency-histograms, at least up to the first eight or nine in a series (depending on the particular click stimulus ensemble used), were displayed in a way to facilitate comparison with the dot displays (e.g., Fig. 2). The click-following capacity of the neurons was evaluated 1) from the significance and magnitude of the vector-strength, VS, assuming a sinusoidal stimulation with period equal to the repetition period of the clicks and calculating the vector-strength in the usual way (Goldberg and Brown, 1968; Epping and Eggermont, 1986, Eggermont, 1991). 2) From the temporal modulation transfer function, tMTF, defined as the coefficient of the fundamental of the Fourier transform of the period histogram as a function of click rate. 3) On basis of the entrainment-rate function (entrainment is defined as the percentage of clicks that produces at least one spike) and 4) from the mean firing rate as a function of click rate (the rate modulation transfer function, rMTF). Consequently, the tMTF was obtained by multiplying the rMTF with the VS-rate function. MTFs were calculated both as number of spikes per click train and per click, this latter normalization is justifiable on basis that the sound energy in the trains is proportional to the number of clicks (Eggermont, 1991). The significance of the vectorstrength was calculated using the Raleigh-test (Mardia, 1972). Only data that had a VS significantly different from zero at the 0.001 level were included in the analysis of the VS behavior. Other statistical tests were based on regression analysis, t-tests and one-factor ANOVA, using the STATVIEW II software package for the Macintosh II. Results

Recordings were made from the primary auditory cortex in 17 kittens ranging in age from 9 days to 53

days. Reliable responses to click stimuli could be obtained in 15 kittens of 14 days and older. For statistical purposes the kittens were divided into four age groups; second post-natal week (8-15 days, 4 kittens, 11 neurons, 19 files), third post-natal week (16-21 days, 5 kittens, 26 neurons, 32 files), fourth post-natal week (22-29 days, 5 kittens, 96 neurons, 148 files) and second post-natal month (30-60 days, 3 kittens, 53 neurons, 71 files). The number of neurons responding to the click stimuli already indicates that the nervous activity in auditory cortex greatly increased after the first 3 post-natal weeks. All neurons had characteristic frequencies in the range of 5-20 kHz. Individual units

Dot displays are shown in Fig. 1 a-g for kittens from 14 days to 43 days of age together with a stimulus sequence on the same time scale (Fig. lh) to facilitate locating the response. We will describe the click-following properties first and thereafter the findings for suppression of spontaneous activity. In part a a multiunit dot-displayfrom a four cell recording (2 on each electrode) is shown for a 43 day old kitten, all units were spontaneously active and had identical click-following behavior. Clear click-following can be discerned for the l/s, 2/s, 4/s and 8/s click trains and, with skipping responses to even numbered clicks, may be also for the 16/s train. For the 32/s click train one observes some time-locked responses in the uppermost part of the graph close to the 300 and 500 ms markers. Part b shows the dot display for a five cell recording (3 units on one electrode plus 2 on the other) in a 38 day old kitten, all units were spontaneously active. There is clear click following for the l/s up to the 8/s click trains and some response to the third and fifth click for the 16/s train. The response to the S/s click train becomes less reliable toward the end of the train. Part c shows results for a five cell recording (3 units on one electrode plus 2 on the other) in a 27 day old kitten with all units spontaneously active. Two cells from the triple unit recording contributed the majority of click following responses while the rebound response was the result of the third unit, the recording on the other electrode also showed a clear onset. Clear following can be seen for the l/s up to the 8/s click trains, albeit for the last one with clear adaptation. Some responses to the 16/s train may be discerned for the lst, 3rd, 4th and 5th click. Part d represents a three unit (2 plus 1) recording in a 22 day old kitten. This kitten was a clear exception in that it showed quite clear responses even for the 32/s click train. This was due to one unit on one electrode only, the two units on the other electrode showed only click following up to 8/s (cf. Fig. 9 where two single unit dot displays are shown). Part e shows responses for a four unit recording (2 plus 2) in an 18 day old kitten. The shorter

latency clusters are from one electrode and the somewhat longer latency responses are from the other electrode. The short latency units were lost somewhere halfway the record. Click-following is evident for the I/s up to X/s trains and also somewhat for the first 4 clicks in the 16/s rain. Part f shows results for a triplet recording (one electrode) in a 15 day old kitten. Click following responses are only evident for the l/s up to the 4/s click train. Part g shows responses from a 14 day old kitten; in this four cell record (3 plus 1) responses are absent for the first clicks in the series and there is clearly evidence for temporal integration resulting in long latency responses to the 2nd, 3rd and 4th click in the 4/s train. In part h the timing of the clicks in the various trains is shown for easy reference. Let us turn now to the suppression phenomena. In Fig. la there is a clear suppression of spontaneous activity that lasts about 135 ms. After the initial suppression one notices a chopper-type rebound with an average period somewhat shorter than the suppression duration. Fig. lb shows suppression after the response to the first clicks in each train that iasts about 175 ms. In Fig. lc the suppression has a duration of about 100 ms. Fig. Id shows that for these units there is no clear sign of post-activation suppression for the lowest click rates, which is entirely the result of firings of the unit that showed click following up to 32/s. The other units showed suppression of about 125 ms in duration. Part f shows units for which spontaneous activity is suppressed for at least 1 second after the onset response to the 8, 16 and 32/s train. Note that the units represented in Fig. lg show no suppression of spontaneous activity at the onset of the click trains and before spikes are actually elicited by the clicks, whereas suppression is evident after click-evoked spikes. Representative single-unit PST histograms are shown in Fig. 2a-d. Fig. 2a shows results for a unit recorded in a 43 day old kitten (one of the units shown in Fig. la). Modal response Iatencies are 18-22 ms for the responses to the first clicks in the train. The 8/s click series evoked stronger responses to the first 7 clicks in the train than to any of the other clicks in any other series and showed an increase in modal latency up to 28 ms. For the 16/s train there are no responses to the second and fourth click while clusters of spikes are observed at about the correct latency in response to the 3rd and 5th click. Fig. 2b shows the series of PST histograms from a unit in a 22 day old kitten (corresponding to Fig. Id). The response latency to first clicks is about 33 ms. Click-following is clear for the l/s up to the 8/s series and still obvious for the 16/s train although adaptation is pronounced. Even for the 32/s train clusters of spikes with about the right latency can be found; this is most clear from the response to the last clicks in the trains. Finally examples are given for an 18 day old kitten in Fig. 2c and for

--wft

698_3 (IS days)

491-2 (18 days) 761LY (22 days) 761-Z (22 days) Xd”h meP”

10

20

30

click rdte (Hz)

Fig. 3. Click-rate

dependence

(h) for selected individual

of entrainment

(a) and vector-strength

units. For more details see text.

a 1.5 day old in Fig. 2d. One observes the increased response latency and width of the PSTH and the decreasing ability to respond to the 8/s click train. In the 15 day old even the responses to the 4/s stimulus are strongly adapting. The entrainment for the responses to the second click as a function of chck rate for selected units is shown on double logarithmic coordinates in Fig. 3a; the largest response was found for one unit in the 22 day old kitten; maximum entrainment was 0.64 and reached its 50% point at a click rate of 10/s. For the other unit in the 22 day old and for the units from younger animals the entrainment was considerably smaller. The genera1 shape of the entrainment-rate functions is similar for the adult-mean and those for the 22 day olds, which suggests that from this age on only the overall firing rate is affected by age. The entrainment for the 15 day old kitten decreased more rapidly with rate than for the older animals. The VS is shown in Fig. 3b for 8 units: the results for the 43 day old kitten resembled those for the adult cats (cf. Fig. 6); the VS peaked sharply at a click rate of 8/s. For the younger kittens the VS. surprisingly, increased for the lower click rates so that the VS click rate function became low-pass like. There was also an increase in VS

51 I.. .

“I.

‘1

’

,

‘,

”

I..

.

7

I.

-

A

. .

;. ., i . . :I: .

.

.. : !. .

I’

11

. * * * ‘.I ;; :;,;

.

1

*. I. .

.

10

.

8

t

. .

.

20

i . i. .

I

.

.

I . . . :

. . .

.

.

.

.

..I

30

.

.

..I

.

40

.

50

age (weeks)

...

10

I

.

.

.

’

.

.

tim: constant = 13

20

30

40

50

60

age (weeks) Fig. 4. Age dependence of spontaneous rate (a) and modal (b) for individual neurons.

latency

for click rates above 8/s, most likely because the response is limited to the first click in the train only and because of the nearly complete absence of spontaneous activity. Results by age group

Spontaneous activity (in spikes/second) depended on age in a significant way (regression analysis, slope different from zero, P < 0.0001). At day 15 the mean spontaneous rate was 0.2 spikes/s, at day 53 it was 0.75/s. Highest rates found were about 3/s and occurred in the 27, 37 and 53 day old kittens (Fig. 4a). Modal latency (in ms, including 1.5 ms acoustic travel time) for the first click in the train showed a significant decrease with age (regression analysis, slope different from zero, P < O.OOOl),the mean value at 15 days was 35 ms and at 37 and 53 days about 20 ms which was not significantly different from adult values. An exponential curve fit indicated a decrease in latency toward the

adult mean value of 20 ms with a time constant of 13 days (Fig. 4b). The effect of age on the entrainment, vector strength and modulation transfer functions was further studied by grouping the data. Fig. 5a-d shows the changes in mean entrainment as a function of click number by age group and click rate. Part a shows the results for the train with click rate of 2/s. First of all one observes that the entrainment for the first click in the train tends to increase with age, however, this was not statistically significant. Only the adult data show an enhancement in the response to later clicks (two-tailed t-test, paired data, P < 0.001) while the younger age groups either show a small non-significant decrease with click number or show no change. In part b results are shown for the 4/s click train; younger age groups tend to show a decrease in the entrainment with click number, adults showed a tendency for an increase. Neither of these changes was statistically significant at the P < 0.01 level. Part c shows results for the 8/s click trains; the notch in the functions for the second click is present in the older age groups only. For the lo-15 day old group there is no response except to the first click. Significant age related differences (P < 0.001) were found for click responses 2-7 in the 8/s train. Finally, part d shows results for the 16/s click trains. Again the lo-15 day old group does not show a response beyond the first click. For the adults and the two oldest kitten groups one observes the notch in the response function for the second click. Statistically significant differences with age were found for the responses to clicks 3-9 in the 16/s train. The behavior of the vector strength with age is, as we have seen above for some individual units, rather unexpected (Fig. 6). While the adult group and the 30-60 day old group show qualitatively the same result in a band-pass type behavior with a maximum VS at a rate of 8/s, the younger age groups tend to show a low-pass behavior. This suggests that despite a large reduction in the entrainment for the youngest age group the relative precision of the spike timing is superior to that in the more adult animals. Most likely the reduced spontaneous activity in the younger animals combined with the smaller or absent rebound response accounts for this behavior. The rate MTF is shown on double logarithmic coordinates in Fig. 7a,b normalized per train as well as per click. Except for an overall lower response the rMTF/train for the 30-60 day olds has the same shape as that for the adult group, this is less clear for the younger age groups. The impression is that basically all curves are shifted downward with respect to the adult group. This suggest an overall firing rate effect and not a specific click rate effect. When the rMTF is normalized for the number of clicks in the train, a similar decreasing function of click rate is obtained for all age

0.4k’ ‘1. ‘. . A rate 2/s

”

* ”

” ---b

”

I-c

-R-

I

1

I

1

2

3

I

I

I

4 5 6 click number

I

1045days

. :

22-21 days 30-60 days adult

_

16-21days

-

0.3 - m__+=

I’-

i

I

I

8

9

.-

-W-

-

1

0.4

2

3

4 5 6 click number

7

8

9

“.

*

D rate 16/s

lo-15 days

0.3 5 ‘a 2

I.

IO-ISdays : 16-21 days . 22-27 days 30-60 days adult

-d-C-

.

7

“, *

--B--

lo-15 davs

-

22-27 days

-I-

adult

0.2

2a, 0.1

1

2

3456789

1

2

click number Fig. 5. Dependence

of age group means of ent~dinment

3

4

5

6

7

8

9

click number for click trains with rates of 2/s. 4/s,

groups. This suggests that the limiting rate (at the 50% point relative to maximum) is the same in all age groups. The tMTFs are also shown on double logarithmic coordinates in Fig. 8a,b; the tMTF/train shows a weak band-pass response with a tendency for the BMF to shift from about 4 Hz in the youngest age groups to 8 Hz in the 30-60 day olds and the adults. Normaliza-

8/s and 16/s on click number.

tion on the number of clicks in the train indicates that in young age the tMTF/click is low-pass, showing an increasing responsiveness with age until at 30-60 days the number of synchronized spikes/click for the l/s train is the same as in the adult cats. The main change taking place between 60 days and adulthood is the increase in the firing rate for higher click rates. In contrast to the rMTFs where most of the change can be accounted for by an overall drop in responsiveness, the tMTFs show age dependent synchrony changes with click rate as well. Discussion

Fig. h. Dependence

of VS (age group means) on click rate.

This report is to my knowledge the first to describe the maturational aspects of periodicity coding in the auditory cortex, therefore a comparison with other studies cannot be made. Where appropriate I have made references to results obtained in cortex of other sensory modalities. This study is also the first to describe maturational properties of single units in the auditory system to periodic click trains. A study by

Sanes and Constantine-Paton (1985) used click evoked potentials in both auditory nerve and inferior colliculus (IC) in the mouse. In the inferior colliculus for low frequency regions (CF = 3-9 kHz) the amplitude click-rate function’s 50% point shifted from around 6 Hz at day 14 to about 30 Hz at day 30 which is comparable to our findings. For high frequency regions (CF = 8-17 kHz) the initial 50% point was again at 6 Hz, however from day 30 on the amplitude rate function was nearly all-pass up to 75 Hz (highest click rate tested). In our data we did not see any differential effect of CF on the click-rate dependence, however, such an effect might have been obscured by the fact that nearly 75% of our fibers had CFs above 10 kHz. While both the IC and auditory nerve data showed maturity for day 30, at day 18 the IC still clearly lagged behind the auditory nerve. Our data, although in a different species and in a more central structure, show

i . :

f I?

--c

-8-

30-60 22-27 16-21 lo-15 adult

days days days days

I . . . . I . . . . I..

10

20

30

click rate (Hz)

y .P “0 2 ‘&

s

IO0 10-1

A e” ;x 10-2 -a-c-

B 5 -a-a-A+

30-60 22-27 16-21 lo-15 adult

---A-

1o-3

days days davs days

-)_

30-60 22-27 16-21 IO-15 adult

days days davs da$s

I . . . .I.*..1 10

20

30

click rate (Hz) 10

20

30

click rate (Hz)

Fig. 8. Mean temporal modulation transfer functions per age group. In (a) the normalization is in spikes per train, in (b) in spikes per click. For the 30-60 day olds and the adult group 95% confidence limits for the means are indicated with error bars.

that maturation is still continuing beyond the 30-60 day old age group. This suggests that there could be a maturation gradient for click-following responses from the auditory nerve to the auditory cortex.

lo’-

Regular spiking units and fast spiking units

-A-

lo-15 days

10

20

30

click rate (Hz) Fig. 7. Mean rate-modulation-transfer-functions per age group. In (a) the normalization is in spikes per train, in (b) in spikes per click. For the 30-60 day olds and the adult group 95% confidence limits for the means are indicated with error bars.

In neocortex two classes of neurons, which in extra cellular recordings respectively show regular spikes or fast spikes and which correspond to fast and slow adapting cells have been described (e.g., Ribaupierre et al., 1972; Simons, 1978). In intracellular recordings with subsequent labeling these regular and fast spike neurons have been identified with respectively pyramidal cells and aspiny (smooth) stellate cells (McCormick et al,, 1985). We have found in a 22 day old kitten (cf. Fig. ld) responses from a cell that were clearly ‘too good’ for that age and also better than anything we have so far found in adult cats (cf. Eggermont, 1991). In Fig. 9 we show two single unit dot displays from that

.J8

B 1i

‘. ”

‘.. ; :

‘.‘:;:

‘:

: i : :.. ,I,’

_’

,’

/

Fig. 9. Dot displays for single units in the 22 day old kitten. (a) shows a slowly adapting neuron with good click following up to 32/s. (b) shows a neuron simultaneously recorded on the second electrode that shows poorer, but adequate for the age group, click following.

multi-unit recording; one representing the ‘slowly adapting cell’ and the other a normal (for that age) fast adapting cell. The ‘normal’ cell shows click following for the entire 4/s click train, and only to the first 4 or so clicks of the 8/s train. The slowly adapting cell showed click following up to the highest rate employed. The spike waveforms for both cells, shown as inserts and recorded on different electrodes, were virtually the same in duration and size. Thus on that basis we did not have a criterion for distinguishing them. Therefore the slowly adapting unit was included in the calculation of the mean values reported in this paper for the 22-27 day old age group. The onset of click following responses

At day 7 thresholds at the characteristic frequency for auditory nerve fibers in the cat are around 130 dB SPL. They improve by about 10 dB/day between day 7 and day 20 and at day 20 are around 20 dB SPL which is still somewhat short of adult values (Walsh and McGee, 1986). The first responses to tone-pips in the present study (maximum p.e. SPL of 105 dB) could be obtained at day 10; thresholds were 95 dB SPL which is in the same range as thresholds in the auditory nerve at that age. Our click levels were also limited to 105 dB p.e. SPL and the first time-locked responses to click trains were obtained at day 14 for clicks of 105 dB. This suggests that single-unit thresholds in AI are following those in the auditory periphery with at most a very short delay. The click responses obtained in the 14 day old kitten showed temporal integration; there were no responses to the first click in a train but for all click trains a

response was obtained to the second click. Because this applies to the 2/s click-train as well as to higher click rates this points to an integration time of the order of 500 ms. Response latencies for the 14 day old kitten, calculated from the second click on, were of the order of 70 ms, click following was clear for the later clicks in the 4/s train but could barely be discerned at 8/s. Modal response latency for kittens from day 15 on decreased exponentially with age to a mean adult value of 20 ms (including 1.5 ms acoustic delay) with a time constant (T) of 13 days. The mean adult value was obtained at day 53 (7-8 weeks). This exponential decay in latency is comparable to those obtained for the Pl component of the cortical evoked potentials (T = 20 days) and wave V of the auditory brainstem response latency (T = 11-12 days) from the studies by Walsh and McGee (1986) in cats. Evoked potential latencies in general take longer to mature than those for single units; a possible reason is that the synchronization between the firings of individual units which make up the EP has a longer maturation time than the latency determining mechanism (see Eggermont, 1985b for a review). Myelination of the auditory nerve was reported to take 3-6 months to reach adult levels (Walsh et al., 1986) and it is expected that myelination in more central structures will follow the same time course. Myelination has frequently been cited as the dominant factor in latency changes during development. Single unit latencies in visual cortex in response to electrical stimulation of the thalamus were reported not to reach adult values until 8-9 weeks after birth, this was attributed to a decreasing thalamo-cortical synaptic delay from 6.4 ms (SD = 1.5 ms) at 2 weeks to 0.9 ms (SD = 0.3 ms) in adult animals (Armstrong-Jones and Fox, 1988). Thus the main limiting factors to account for the about 20 ms decrease in latency between day 15 and day 53 could be both myelination and synaptic delays. Entrainment and other rate measures

The mean spike entrainment to first clicks in a train was a factor 2-3 lower in the youngest animals than in adults, and the general effect of age on the entrainment as a function of click number for trains with click rates of 2/s and 4/s was an overall increase in neural activity. However, for trains with higher click rates there were also qualitative differences, the most obvious one was that in the youngest age group there were no responses except for the first click. Thus in addition to a general decrease in responsiveness there was also an age dependent adaptation effect. This relates to findings of McCormick and Prince (1987) that trains of action potentials (generated by a constant depolarizing current) in both immature and mature neurones (rat sensorimotor cortex in vitro) displayed substantial adaptation (instantaneous firing rate decreased with

55

every new spike). Adaptation in general appeared less steep in the mature neurons although both mature and immature neurons could display similar degrees of this time dependent reduction in firing rate. They also found that more mature neurons were capable of maintaining a steady and constant discharge rate (to the current) while immature neurons often adapted after 1-5s to such an extent that action potential generation practically ceased. The in vitro data also showed that cortical pyramidal neurons shortly after birth were capable of engaging in short periods of sustained firing with firing rates of up to 100 Hz. An important limiting factor to the ‘circuit’ behavior of immature cortex was suggested to be present at the synaptic level. It appeared that the neurons in newborn cerebral cortex were constrained to a large extent in the frequency of their discharges by the presence of immature and poorly developed synaptic connections (McCormick and Prince, 1987). A decrement in the suppression duration with age will also tend to increase the entrainment. We did not see the profound fatiguelike effect in the firing rate of single units, showing decrements in response over minutes of stimulation, that has been reported by Sanes and Constantine-Paton (1985) for IC units with characteristic frequencies below 9 kHz. The rate MTFs for the various age groups, both normalized per train and per click, when plotted on double logarithmic coordinates are essentially parallel. This suggests that the functions for the various age groups are to a large extent scaled versions of each other and that the major developmental factor is a change in overall firing rate. We concluded differently for the entrainment measure where in addition to a scaling effect a specific rate dependence may have been present for the youngest age group in the 8/s and 16/s trains. This may have had its origin in the method of calculation: entrainment is only concerned with first spikes to a click in a fixed time window, rMTF calculations incorporate all spikes generated per period or per train. In the latter case suppression effects, post inhibitory rebound and spontaneous activity become important in determining its size. Effect of age on vector strength

The VS showed an unexpected dependence on age and became larger in younger age kittens rather than smaller. As we have discussed previously (Eggermont, 1991) the VS is extremely sensitive to the level of spontaneous activity because of its normalization on the total number of spikes. The rebound firings at the end of the suppression period which for long modulation periods tend to fall in the ‘wrong’ part of the period histogram also have a detrimental effect on the magnitude of the VS for certain click rates. In the younger animals the spontaneous activity is low and

rebound responses are not as pronounced, hence the VS in younger animals is not artificially depressed for low click rates as in older kittens (30-60 days) and adult cats. In addition when spontaneous rates are high, an overall, rate independent, depressing effect on the value of the VS can be seen from a comparison of the oldest kitten group and the adult cats. We have argued before (Eggermont, 1991a) that the VS is not a good measure for the characterization of periodicity coding in auditory cortex, the findings in the present paper underscore this once more. Temporal modulation transfer functions

Calculations of the tMTFs are based on a multiplication of the respective rMTFs with the VS-rate function. Alternately one can say that the VS is the ratio of the tMTF/train (the coefficients of the fundamental of the Fourier transform of the period histogram) and the rMTF/train (the number of spikes per period). The tMTF describes therefore the number of synchronized spikes per period. Because the rMTFs did not show a click rate dependent age effect but just an overall scaling and the VS showed a considerable age and click rate effect one expects an interaction of age and click rate for the tMTFs. In effect one observes a gradual increase in the limiting rate (50% point based on the value at l/s) for the tMTF/click from 6 Hz (lo-15 day olds) to 12 Hz in the adult cats. In addition there is an enhancement in the low-pass characteristic for the 4/s and 8/s values so that the function becomes slightly band-pass in the adult cats. This enhancement for the 8/s clicks may have been caused by the addition of the rebound response for click number N to the onset response for click number N + 1 (Eggermont, 1991).

Conclusions

Maturational effects on periodicity coding comprise changes in overall responsiveness as well as rate-dependent changes. The responsiveness or the number of spikes elicited by single stimuli increases on average 3-fold between the second post-natal week and adulthood, probably as a result of more efficient synapses in the central auditory pathway. Adaptation becomes less pronounced with age; neurons start to respond to the later clicks in the 8/s and 16/s click trains from the third post-natal week on. By the end of the first postnatal month the click following responses resemble the adult ones qualitatively, however, increased firing rates and spontaneous rates together with rebound responses continue to produce quantitative differences between the 30-60 day olds and the adults. The maturation of periodic&y coding in AI therefore takes at least a factor 2 longer than that of the threshold and the binaural response properties. The decrease in the

duration of the post-activation suppression coupled with the increased response to trains with higher click rates with age suggests that post-activation hyperpolarization plays a major role in this rate dependence. A comparison with findings in other sensory modalities suggests that the maturation of cortical cellular properties is the limiting property for this rate dependence.

Acknowledgements

This investigation was supported by grants from the Alberta Heritage Foundation for Medical Research and the Natural Sciences and Engineering Research Council of Canada. Geoff Smith wrote the stimulus generation software, data acquisition software, and some of the data analysis software.

References Armstrong-Jones, M. and Fox, K. (1988) The physiology of developing cortical neurons. In: A. Peters and E.G. Jones (Eds.), Cerebral Cortex, Volume 9, Development and maturation of cerebra1 cortex, Plenum Press, New York, London, pp. 237-272. Brugge, J.F. (1988) Stimulus coding in the developing auditory system. In: G.M. Edelman, W.E. Gall and W.M. Cowan (Eds.), Auditory function. Neurobiological bases of hearing, John Wiley and Sons, New York, pp. 113-136. Brugge. J.F., Javel, E. and Kitzes, L.M. (1978) Signs of functional maturation of peripheral auditory system in discharge patterns of neurons in anteroventral cochlear nucleus of kitten J. Neurophysiol. 41, 1557-1579. Brugge. J.F., Kitzes, L.M. and Javel, E. (1981) Postnatal development of frequency and intensity sensitivity of neurons in the anteroventral cochlear nucleus of kittens. Hear. Res. 5, 217-229. Creutzfeldt, O.D., Hellweg, F.-C. and Schreiner, Chr. (1980) Thalamocortical transformation of responses to complex auditory stimuli. Exp. Brain Res. 39, 87-104. Eggermont, J.J. (1973) Analog modeling of cochlear adaptation. Kybernetik 14, 117-126 Eggermont, J.J. (1985a) Peripheral auditory adaptation and fatigue: a model oriented review. Hear. Res. 18, 57-71. Eggermont, J.J. (1985b) Physiology of the developing auditory system. In: SE. Trehub and B. Schneider (Eds.), Auditory development in infancy. Plenum, New York, pp. 21-45. Eggermont, J.J. (1990) Temporal modulation transfer functions for single neurons in the auditory midbrain of the leopard frog. Intensity and carrier-frequency dependence. Hear. Res. 43, 181198. Eggermont, J.J. (1991) Rate and synchrony measures of periodicity coding in cat primary auditory cortex. Hear. Res. 56, 153-167. Epping, W.J.M. and Eggermont, J.J. (1985) Single-unit characteristics in the auditory midbrain of the immobilized grassfrog. Hear. Res. 18, 223-243. Epping, W.J.M. and Eggermont, J.J. (1986) Sensitivity of neurons in the auditory midbrain of the grassfrog to temporal characteristics of sound. I. Stimulation with acoustic clicks. Hear. Res. 24. 37-54. Goldberg, J.M. and Brown, P.B. (1969) Responses of binaural neu-

rons of dog superior olivary complex to dichotic tonal stimuli: some physiological mechanisms of sound localization. J. Neurophysiol. 32, 613-636. Javel, E.. McGee, J.A., Horst, J.W. and Farley, G.R. (1988) Temporal mechanisms in auditory stimulus coding. In: G.M. Edelman, W.E. Gall and W.M. Cowan (Eds.), Auditory function. Neurobiological bases of hearing, John Wiley and Sons, New York, pp. 515-558. Kettner, R.E., Feng, J.-Z and Brugge, J.F. (1985) Postnatal development of the phase-locked response to low frequency tones of auditory nerve fibers in the cat. J. Neurosci. 5, 275-283. Steriade, M. and Llinas, R. (1988) The functional states of the thalamus and the associated neuronal interplay. Physiolog. Rev. 68, 649-742. Mardia, K.V.(1972) Statistics of directional data. Academic Press, London. McCormick, D.A., Connors, B.W., Lighthall, J.W. and Prince, D.A. (1985) Comparative electrophysiology of pyramidal and sparsely spiny stellate neurons of the neocortex. J. Neurophysiol. 54, 782-806. McCormick, D.A. and Prince, D.A. (1987) Post-natal development of electrophysiological properties of rat cerebral cortical pyramidal neurones. J. Physiology Lond. 393, 743-762. Moore, D.R. and Irvine, D.R.F. (1979) The development of some peripheral and central auditory responses in the neonatal cat. Brain Res. 163, 49-59. Payne, B., Pearson, H. and Cornwell, P. (1988) Development of visual and auditory cortical connections in the cat. In: A. Peters and E.G. Jones (Eds.), Cerebral Cortex, Volume 9, Development and maturation of cerebral cortex. Plenum Press, New York and London, pp. 309-389. Phillips, D.P., Hall, S.E. and Hollett, J.L. (1989) Repetition rate and signal level effects on neuronal responses to brief tone pulses in cat auditory cortex. J. Acoust. Sot. Am. 85, 2537-2549. Phillips, D.P. and Farmer, M.E. (1990) Acquired word deafness, and the temporal grain of sound representation in the primary auditory cortex. Behavioural Brain Res. 40, 85-94. Purpura, D.P., Shofer, R.J. and Scarff, T. (1965) Properties of synaptic activities and spike potentials of neurons in immature neocortex. J. Neurophysiol. 28, 925-942. Ribaupierre, F. de, Goldstein, M.H. Jr. and Yeni-komshian, G. (1972) Intracellular study of the cat’s primary auditory cortex. Brain Res. 48, 185-204. Sanes, D.H. and Constantine-Paton, M. (1985) The development of stimulus following in the cochlear nerve and inferior colliculus in the mouse. Dev. Brain Res. 22, 255-267. Schreiner, C.E. and Langner, G. (1988) Coding of temporal patterns in the central auditory nervous system. In: G.M. Edelman, W.E. Gall and W.M. Cowan (Eds.), Auditory function. Neurobiological bases of hearing. J. Wiley and Sons, New York, pp. 337-361. Simons, D.J. (1978) Response properties of vibrissa units in rat SI somatosensory neocortex. J. Neurophysiol. 41, 798-820. Smith, R.L.(1988) Encoding of sound intensity by auditory neurons. In: G.M. Edelman, W.E. Gall and W.M. Cowan (Eds.), Auditory function. Neurobiological bases of hearing, John Wiley and Sons, New York, pp. 243-274. Walsh, E.J. and McGee, J. (1986) The development of function in the auditory periphery. In: R.A. Altschuler, D.W. Hoffman and R.P. Bobbin (Eds.), Neurobiology of Hearing, Raven Press, New York, pp. 247-269. Walsh, E.J., McGee, J. and Javel, E. (1986) Development of auditory-evoked potentials in the cat. II. Wave latencies. J. Acoust. Sot. Am. 79. 725-744.