Exp Brain Res (t992) 88:t5~168

Experimental BrainResearch © Springer-Verlag1992

Neurons in cat primary auditory cortex sensitive to correlates of auditory motion in three-dimensional space E. Stumpf 1, J.M. Toronchuk 1, and M.S. Cynader 1'2 Departments of 1Psychologyand 2Ophthalmology,University of British Columbia, Vancouver, B.C. Canada, V5Z 3N9 Received September 30, 1990 / AcceptedJuly 3, 1991

Summary. Amplitude modulation at the receiver's ears is a characteristic of moving sound sources. When a sound source moves from side to side, stimulus intensity decreases in one ear and increases in the other. When a sound source moves toward or away from the organism, the two ears receive correlated increases or decreases in sound level. We recorded from single cells in the auditory cortex while presenting amplitude modulated pure tones to the two ears which simulated motion either toward or away from the organism, or from side to side. Our results indicate that auditory cortex neurons can be highly sensitive to these correlates of auditory motion in three dimensional space. Three major classes of neurons were encountered. These included 1) neurons sensitive to azimuthal stimulus motion, 2) neurons sensitive to motion directly toward or away from the organism, and 3) monaural-like neurons. More toward-preferring neurons than away-preferring neurons were encountered, and more units preferred contralateral-directed than ipsilateral-directed movement. The different classes of direction-selective neurons were spatially segregated from each other within the cortex and appear to occur in columns. In addition to their selectivity for different directions of simulated sound source motion, auditory cortex neurons could also be highly selective to AM ramp rate and excursion; these are correlates of sound source velocity. Key words: Auditory cortex Depth - Binaural interact i o n - Motion - Amplitude modulation

Introduction An obvious parallel exists in the functioning of the visual and auditory systems: in both systems differences in the This research was supported by MRC of Canada grant no MA-9856 to M.S.C., and a MRC Studentship to E.S. Offprint requests to: M.S. Cynader, Dept. of Ophthalmology

outputs of paired end-organs (the eyes and the ears) are detected and these differences provide cues for determining the locations of stimuli in space. From binocular output, the visual system also extracts information regarding the orientation, surface curvature, and direction and velocity of motion of visual stimuli; in comparison, relatively little is known about the functional significance of binaural auditory inputs beyond the role they play in the localization of stationary sound sources. The aim of this research is to investigate the mechanisms underlying these "higher order" auditory processing tasks, and in particular the neural mechanisms involved in tracking and localizing moving sound sources. The basilar membrane is the receptor organ of the auditory system in higher vertebrates. Because of its tonotopic organization, it codes along its length for the frequency of sound sources but not for their spatial location. Mammals and birds approach the problem of auditory localization by comparing the signals at the two ears. The distance between the two ears and the shadowing effects of the head and pinnae give rise to a series of binaural disparities. Some of these disparities are interaural differences in on-going phase, intensity, transient arrival time and frequency spectrum of auditory stimuli (Gourevitch 1978). These cues play an important role in the coding of sound source location. A particular difference in intensity, phase, spectral content and/or time of arrival of sound at the two ears is correlated with a particular location in space for the sound source. Single neurons in the cat primary auditory cortex (area A1) are sensitive to interaural intensity differences (Reale and Kettner 1986), interaural time differences (Kitzes et al. t980) and to differences in spectral content in the two ears (Mendelson and Cynader 1983). These neurons can be sharply tuned to binaural sound localization cues corresponding to sound sources located in the frontal and contralateral region of space. Using free-field testing, A1 neurons have also been shown to have spatial receptive fields: they respond selectively to the location of free-field auditory stimuli (Eisenman 1974; Middle-

159 brooks and Pettigrew 1981). In addition, ablation studies have shown that integrity o f the p r i m a r y auditory cortex is necessary for accurate sound localization (Jenkins and Masterton 1982; Jenkins and Merzenich 1984). All o f this research has been carried out using stationary sound sources, or correlates o f such sounds. Yet, a conspicuous feature of auditory stimuli in the environment is their motion with respect to the receiver, either as the stimulus moves or as the head is turned. Despite this prominence, m o v e m e n t as a stimulus feature has been largely overlooked by most investigators of auditory neurophysiology. A few investigators have found auditory neurons which responded selectively to moving sound sources (Sovijarvi and Hyvarinen 1974) or to correlates of auditory motion (Altman 1987). These studies are anecdotal in nature; a systematic, quantitative analysis of neuronal responses to moving sound and an analysis of the underlying cortical mechanisms has yet to be performed. One correlate o f moving sound is amplitude modulation at the receiver's ears. W h e n a sound source moves toward or away f r o m the receiver along the midline, the two ears receive correlated (diotic) increases or decreases in sound intensity. When a sound source moves along the horizontal plane, the two ears receive opposite-directed (dichotic) changes in sound intensity: sound intensity increases in one ear and decreases in the other. Stimuli presented dichotically through earphones are perceived as a fused image lateralized within the head; modulating sound intensity at the two ears changes the location o f the fused image and creates an impression of auditory motion. Amplitude modulated sound presented through a sealed sound delivery system can be used to simulate auditory motion in three-dimensional space, but with some limitations. One obvious limitation is the fact that the modifying effects of the head and pinnae on the incoming signal are bypassed altogether. Another one is the signal itself: amplitude modulation as the only auditory m o t i o n cue is an incomplete stimulus. Adding other motion cues (modulations of frequency spectrum, time of arrival and phase) would produce a more accurate simulation of auditory motion and provide additional information to a neural mechanism specialized to detect moving sound sources. These limitations m a y result in perceptual ambiguities: for example, it is impossible for a receiver to distinguish between the amplitude changes associated with sound source m o v e m e n t toward or away from the organism and actual modulation of the amplitude o f a stationary sound source in an anechoic environment (Gardner 1969), although in a natural e n vironment reverberation cues m a y m a k e distance discrimination possible (Mershon and King 1975). With these limitations in mind, we have found that neurons in cat primary auditory cortex (A1) are sensitive to changes in binaural sound source intensity. Cortical neurons respond selectively depending on the relative direction of intensity change in the two ears and also depending on the rate o f intensity change in the two ears. These are correlates of sound source motion and sound source velocity, respectively. In addition, we have found that this selectivity for correlates o f auditory motion is

related to other attributes o f auditory cortex neurons that have traditionally not been associated with selectivity for auditory motion or location.

Methods

Animal preparation Acute experiments were performed on 7 healthy adult cats. Surgical anesthesia was induced with sodium thiopental (10 mg/kg i.v. initially); additional doses were administered during surgery to maintain areflexia. To prevent excessive salivation and respiratory difficulties, a single dose of atropine sulfate (0.2 mg, i.v.) was administered. Dexamethasone (0.5 mg, i.m.) was administered to prevent brain edema. A tracheotomy was performed and an endotracheal tube inserted. The animal was supported by a head-holder which left the skull and pinnae free from obstruction. The skull and auditory meatuses were cleared of surrounding tissue. A 5-mm diameter hole was drilled in the skull overlying the left ectosylvian gyrus (area AI), and the intact dura was covered with petroleum jelly to keep it from drying. The auditory meatuses were transected and the pinnae reflected anteriorly to allow insertion of the stimulus delivery system directly into the ear canals. All wounds and pressure points were infiltrated with a long-acting local anesthetic (bupivacaine hydrochloride 2.5%). When surgery was completed, sodium thiopental anesthesia was discontinued. The animal Was paralyzed (gallamine triethiodide 20 rag, i.v.) and artificially respired with a 70:30 mixture of nitrous oxide and oxygen. The animal was maintained on a continuous intravenous infusion of gallamine triethiodide (10 mg/kg/h), sodium pentobarbital (1 mg/kg/h) and 5% lactated dextrose (10 ml/h) in Ringer's. End-tidal CO2, heart rate, blood pressure and EEG were monitored continuously. Experiments (unpublished) in which paralytic is withdrawn confirm that this regime results in anaesthesia and areflexia in experimental subjects. Rectal temperature was maintained at 38°C using a feedbackcontrolled heating blanket; expired CO 2 was maintained at about 4% by adjusting the rate of the respiration pump.

Stimuli Pure tone sinusoids were generated by a Wavetek model 110 function generator. Signals to the speakers were controlled by two digital-to-analog converters of the PDP/ll computer, and fed through an analog multiplier and an amplifier (Technics SU-700 stereo integrated amplifier). The stimulus delivery system consisted of two loudspeakers (Pioneer WXX-172) connected to hollow aluminum acoustic couplers which fitted snugly into the transected auditory meatuses. The stimuli employed were amplitude modulated pure tones which simulated auditory motion in four canonical directions. Likedirected changes in sound intensity at both ears simulated movement toward or away from the head along the midline (Fig. 1A); opposite-directed changes in sound intensity simulated sound moving along the azimuthal plane, toward one ear and away from the other, and vice versa (Fig. 1B). Since the duration of the AM ramp was constant, the rate of rise of the AM ramp in these experiments is a correlate of sound source velocity: increased ramp excursion and speed corresponds to higher speeds of sound source motion. Each of the four simulated directions of motion was presented at four different rates of rise of the AM ramp, giving rise to four different intensity levels which spanned 18 dB (Fig. 1C): The stimuli were shaped with envelope generators to produce rise and fall times of 5 ms; interaural phase was always zero. For azimuthal-simulated motion, stimulus onset was followed by a 295 ms plateau, then a 250 ms AM ramp, another 295 ms plateau and finally an offset (stimulus time course is shown in Fig. 1D). For

160 A

TOWARD

AWAY I+C ..........

~ , ~

I+C~ .........

1 B

l

IPSILATERAL-DIRECTED

CONTRALATERAL-D~ECTED I .........

c

C

~

X

.........

D

C

S

onset

AM ramp

offset

I

I

l

150

450

700

toward and away motion the 295 ms plateau was preceded or followed respectively by a 250 ms AM ramp. Total data collection time was 1300ms, with a 200ms wait time between stimulus presentations. There were 40 presentations for each of the 16 conditions, for a total of 640 stimulus presentations (4 directions x 4 ramp excursions x 40 presentations) for each unit. The initial level of the plateau portion was presented approximately 20 dB above each cell's threshold at CF. Increasing the steepness of the ramp gave rise to plateau levels which were 6, 12, and 18 dB louder (see Fig. 1C). This range of intensities represents an eight-fold change in distance for toward and away motion, and a 180 ° change in location for each of the azimuthal conditions.

Calibration A probe microphone (IVIE 1300) was inserted in the acoustic couplers for in situ measurement of sound pressure levels near the tympanic membrane. A waveform analyzer (DataPrecision Data 6000) produced fast Fourier transforms of the sound spectra. The output of a Bruel and Kjaer pistonphone (124 dB at 250 Hz) was used to convert relative intensity measurements into sound pressure level (dB SPL re 20 gN/m2). Variations in sound intensity at different frequencies were corrected with reference to the output of the waveform analyzer. Sound levels were calibrated at the beginning of each experiment and monitored on-line throughout the experiment.

Data collection The animal was located in a sound-attenuating chamber (IAC Controlled Acoustic Environments). Except for the sound delivery system, microdrive, the stereotaxic apparatus, table and animal, all

I

i

Fig. 1A-D. Representation of stimuli used to simulate 'auditory motion. Pure tones are presented dichotically via a sealed system. A Correlated increases or decreases in sound level at the two ears simulate auditory motion in depth. B Opposite-directed changes in sound level at the two ears simulate auditory motion in azimuth. C Ramp rate and excursion simulates sound source velocity. D Stimulus time course

1000 msec

of the equipment used (infusion pump, respirator, computer, etc.) was located outside the chamber. The responses of single neurons were recorded extracellularly using glass-coated platinum-iridium microelectrodes with impedances of 1-1.5 Mf~ at 1 kHz. Electrodes were advanced perpendicularly through the dura and cortex using a remote controlled microdrive. Signals from the electrode were bandpass t~ltered (500 Hz to 12 kHz), amplified (1000 x ), discriminated with a window discriminator and monitored on an oscilloscope and an audio monitor (Grass AM8). Stimulus presentation and on-line data collection and display were controlled by the PDP/11 computer via an IBM PC serving as an intelligent terminal. White noise and tone bursts were used as search sitmuli. When a unit was encountered, its characteristic frequency (CF) was determined. Stimulus presentation was done at CF, although additional frequencies were sometimes used. The 16 stimulus conditions delineated above were individually interleaved and presented in a randomized order. This process was repeated for a total of 40 trials at which point data collection for an individual unit terminated. The total length of an experiment was ~ 3 days; recording time was 12-36 h. When recording was completed, the animal was killed with an intravenous injection of sodium pentobarbital (100 mg/kg). The skull was opened to verify electrode placement using anatomical landmarks and previous maps of the auditory cortex surface (Schreiner and Cynader 1984).

Data analysis The computer generated post-stimulus histograms and spike counts for all conditions. Aural dominance and binaural interaction category were determined by comparing responses to stimulation of the contralateral ear alone, stimulation of the ipsilateral ear alone and stinmlation of both ears during onset of the plateau portion of the stimulus. Responses to sound onset and offset were examined in a similar way.

161 Responses to correlates of the four different directions of motion were determined by analyzing responses during the AM ramps. Preferred direction of movement in depth was determined by comparing responses to simulated sound source movement toward and away from the receiver (correlated binaural increases and decreases in sound levels, respectively); similarly, preferred direction of movement in azimuth was determined by comparing responses to ipsilateral- and contraIateral-directed simulated sound source movement (increase in sound level in one ear and simultaneous decrease in sound level in the other ear). Units which gave weak responses to AM ramps (less than one spike per sweep above spontaneous activity) were classified as insuMciently responsive and not evaluated further. A 2 : 1 ratio between responses to opposed directions of motion along one dimension (depth or azimuth) was taken as the criterion for directional selectivity. For all analyses, average spontaneous activity during the first 150 ms was calculated for comparison by averaging across all 640 (16 x 40) presentations. The latency of the onset response to the plateau portion was calculated for each cell and the ramp analysis period was delayed in order to take this latency period into account.

Results D a t a were o b t a i n e d for a t o t a l o f 80 n e u r o n s . N i n e t e e n n e u r o n s d i d n o t r e s p o n d well to the A M r a m p s a n d will n o t be d e s c r i b e d further. T h e C F s o f the r e m a i n i n g 61 units r a n g e d f r o m 2 to 44 k H z . S h a r p s t i m u l u s onsets (5 m s e c rise times) t y p i c a l l y p r o d u c e d s t r o n g t r a n s i e n t r e s p o n s e s ; the d i s c h a r g e elicited b y t h e A M r a m p s w a s more sustained.

Directional selectivity T h r e e - q u a r t e r s (61/80) o f the units e n c o u n t e r e d res p o n d e d to c o r r e l a t e s o f a u d i t o r y m o t i o n . T h e m a j o r i t y

IPSILATERAL-DIRECTED ,--M

200-1

away 3 ipsilater[~ateral

monaural-Ftke (ipsi)

[ | !

"~ "~ 16 monaural-like (contra)

20 toward Fig, 2. Summary of directional preferences for all direction-selective units encountered (n = 54). The perspective is from directly above the head of the animal. The vertical axis corresponds to depth; the horizontal axis corresponds to azimuth. Recordings are assumed to be made from left hemisphere. The length of each arrow is proportional to the number of units preferring a particular direction of simulated sound som'ce motion; the numeral next to each arrow indicates the number of such units that were encountered. Oblique arrows refer to monaural-like units; these units showed equal preference for increases in sound level in both ears and in one ear alone. Units not included in this figure were those which did not show any directional preference (n= 7), and those which did not respond to correlates of sound source motion (n = 19) o f these A M - s e n s i t i v e units r e s p o n d e d selectively to the d i r e c t i o n o f a u d i t o r y m o t i o n a n d were classified a c c o r d ing to p r e f e r r e d d i r e c t i o n o f s i m u l a t e d s o u n d source m o t i o n . T h r e e b r o a d c a t e g o r i e s o f d i r e c t i o n a l selectivity were o b s e r v e d : t) selectivity for m o t i o n in d e p t h ( t o w a r d o r a w a y f r o m the receiver), 2) selectivity for m o t i o n in

TOWARD 200]

It

100 I

I

I

I

I

I

I

I

I

I

I

I

I

' 2;0' g 0 ' 600' 8(~0 '10})0' 1200''

200 400 600 800 1000 1200

CONTRALATERAL-DIRECTED



_k I

I

I

I

!

I

!

I

I

I

!

!

AWAY

100 I

200 400 600 800 1000 1200 time (msec)

AU22.047 13 kHz

300

' 2(]0 '400 '600' '800' '1000'lif00' time (msec)

Fig. 3. Post-stimulus time histograms showing the responses of a single neuron in relation to the time course of the stimuli, for the four direction of auditory motion. Responses are collapsed across the four values of ramp excursion. Bin width is 10 ms. The bracketed portions are the responses during the AM ramp; these responses are subject to further analysis. This unit is a toward-preferring unit because its response during correlated AM ramps in the two ears (top right panel) is the strongest

162 azimuth (ipsilateral- or contralateral-directed), and 3) selectivity for motion directed either toward the receiver or in azimuth (monaural-like responses; see Fig. 7). A small number of units (n = 7) responded to AM ramps but did not show any directional selectivity. The distribution of units falling into each of the major groups is summarized in Fig. 2.

A

B

8

Motion in depth Units preferring motion in depth tended to show facilitatory binaural interactions (EE cells). Many more o f these units responded selectively to increases than to decreases in sound level (i.e., more toward- than away-preferring units were found; see Fig. 2). Toward-preferring units gave transient responses to stimulus onset and sustained responses during A M ramps in both ears. Figure 3 shows post-stimulus time histograms depicting the timing of a toward-preferring neuron's discharge in relation to the time course of the stimulus, for the four directions of simulated sound source motion. In the top left panel of Fig. 3, note the strong transient response to monaural stimulus onset in the contralateral ear at 150 ms. There is virtually no response above background during the period between 450 and 700 ms during which the stimulus increases in amplitude in the ipsilateral ear and decreases in the contralateral ear. The bottom left panel shows a weaker transient response with stimulus onset in the ipsilateral ear at 150 ms, and a weak response during the period between 450 and 700 ms marked by oppositedirected A M ramps in the two ears. By contrast, the response obtained during correlated increases in amplitude in the two ears illustrated in the top right panel of Fig. 3 is quite vigorous, and contrasts markedly with the lack o f response observed with correlated decreases in amplitude in the two ears in the bottom right'panel of Fig. 3. Panel A of Fig. 4 illustrates the responses of the unit described in Fig. 3 in the form of a polar plot. In this panel the perspective is from above and the receiver is positioned at the bottom of the open circle. The length of each arrow is proportional to the number of spikes evoked during the A M portion of the 40 trials. F o r the unit in Fig. 4A, 260 spikes above spontaneous activity (represented by the shaded circle) were evoked during correlated increases in sound level in the two ears and correlated decreases in sound level in the two ears produced inhibition below spontaneous levels. Inhibition or weak responses were evoked when sound level increased in one ear and decreased in the other ear (corresponding to azimuthal motion directed toward the ipsilateral or contralateral ear). The breadth of tuning for this correlate o f direction o f sound source motion could vary from neuron to neuron. Panels B and C in Fig. 4 show the responses o f two other units, both of which responded most vigorously during correlated increases in stimulus amplitude in the two ears. In some neurons like that illustrated in panel C, responses to the other directions of motion tested (aside from motion directly toward the receiver) were less than spontaneous activity, indicating inhibition in the non-preferred directions. The

AU22.047 6 kHz

J'L24.013 20 kHz

C

SE13.004 44 kHz

Fig. 4A-C. Plots of spike counts during AM ramps simulating four directions of auditory motion. Each plot refers to a different unit. The perspective is from directly above the head of the animal; the filled inside circle indicates spontaneous activity. The vertical axis represents depth (toward-away), and the horizontal axis represents azimuth (ipsilateral-contralaterat). The ipsilaterat ear is on the left side in all cases. The length of each arrow is proportional to the spike count during the AM ramp for each direction of motion; the accompanying numeral refers to the number of spikes evoked during the same period (minus spontaneous activity). These three toward-preferring units have different breadths of tuning for simulated motion toward the receiver. A Polar plot for the unit shown in Fig. 3. B-C Toward-preferring units with different breadth of tuning for motion toward the receiver. The added response to sideways motion becomes more apparent in more broadly tuned units (B) neuron in panel B showed responses that were much more broadly tuned than those of the cells of panels A and C but responses to other directions of motion were still clearly weaker than those evoked by simulated motion toward the receiver. We accepted neurons into the category of "toward units" only if responses to simulated motion toward the receiver were at least twice as vigorous as those to the opposite direction, and t.5 times stronger than to simulated motion in azimuth. Figure 5 shows polar plots representing the responses of two neurons that responded most vigorously to simulated motion directed away from the receiver. As illustrated in Fig. 2, neurons preferring simulated motion away from the animal were relatively less c o m m o n than those preferring motion toward the animal. In both cases, the degree of selectivity appeared to vary in strength from neuron to neuron. The neuron in panel B

163

A

B

A

B

64

lO 2 8 10

AU22.024 6 kHz

AU22.067 13 knz

Fig. 5A, B. Conventions are the same as for Fig. 4. Polar plots for two away-preferring units

AU22.075 2 kHz

C

D

SEI3.003 16ldtz

showed almost no response to motion toward the animal while the neuron in panel A was more broadly tuned for motion in depth. Motion in azimuth

Figure 6 shows polar plots illustrating the responses of four neurons which responded most vigorously to simulated motion in the direction contralateral to the hemisphere under study (panels A and B) or in the ipsilateral direction (panels C and D). Units selective for azimuthal motion appeared relatively more rare than units preferring motion toward the animal (see Fig. 2) and tended to be characterized by inhibitory binaural interactions (EI or IE cells) and strong excitatory responses via only one of the two ears (Toronchuk et al. 1991). Azimuthal selective neurons gave transient responses to sound onset in the dominant ear, sometimes accompanied by transient responses to sound offset in the non-dominant ear. Figure 6 shows that the degree of selectivity among azimuth preferring units could vary greatly. In some cases, responses to the opposite direction of motion could be either non-existent or less than spontaneous activity. In other cases, responses to the opposite direction of motion could be fairly vigorous. Monaural~tike units

It is important to understand the response o f a neuron with no binaural interaction under our stimulus conditions. If a neuron prefers stimulus onset in the contralateral ear (and therefore responds to increasing stimulus amplitude via that ear) and has no input at all from the ipsilateral ear, then that neuron will respond well in two of the four conditions employed. These include the conditions in which stimuli increase in amplitude in both ears (because of the increase in amplitude in the contralateral ear) and the azimuthal motion condition in which the stimulus increases in amplitude in the contralateral ear while decreasing in the ipsilateral ear. Figure 7A is a polar plot representation of a monaural-like unit which

AU22.027 8 kHz

AU22.028 10kHz

Fig. 6A-B. Conventions are the same as for Fig. 4. Polar plots for two neurons preferring contralateral-directed simulated azimuthal sound source motion. C-D Polar plots for two neurons preferring simulated ipsilateral-directed azimuthal sound source motion

A

B

43 @

296=,,,,,1~

AU2.018 5 kHz

JL24.007 2 kHz

Fig. 7A, B. Conventions are the same as for Fig. 4. A A contralateral-dominated monaural-like unit. Note that the responses to simulated sound source motion toward the observer and azimuthal contralateral-directed motion are equivalent. B An ipsilateral-dominated monaural-like unit. Again, the responses to simulated sound source motion toward the observer and azimuthal ipsilateral-directed motion are similar. This unit had almost no spontaneous activity

is contralateral-ear-dominated. This neuron responds best to increases in sound level in the contralateral ear, regardless of the input to the ipsilateral ear. Hence, the responses to contralateral-directed motion and simulated auditory motion toward the receiver are essentially the

A

AU08.002 5.5kHz JL24.013 20kHz

AU08.003 5.5kHz

@

AU08.004 5.5kHz

U

~

"

II..-

@

YL24,017 25kHz

@

31224.022 25kHz

AU08.005 4.5kHz JL24,025 20kHz

AU08,006 5kttz

C

D

JL24,003 2kHz

AU22.026 10kHz IL24.006 2kHz

AU22.027 8kHz

JL24.008 2kHz

AU22.028 10kHz

JL24.010 2kHz

7,00~ I

i

200g



JI.24.012 2kHz

165 same. An ipsilateral-ear-dominated monaural-like unit is shown in Fig. 7B; the response to ipsilateral-directed azimuthal motion is essentially the same as that directed directly toward the receiver.

Distribution of unit types A total of 21 penetrations, roughly normal to the cortical surface, were made. Neurons found along m o s t electrode penetrations normal to the surface of the cortex showed similar directional preferences, which indicates that units with constant responses to simulated sound source m o tion m a y occur in columns within the cortex. Examples of schematic reconstructed penetrations are shown in Fig. 8. Each polar plot represents the responses to simulated sound source motion o f one unit, as in previous

figures. Neurons encountered along penetrations A and B responded best to simulated sound source motion toward the receiver; some units also responded to sideways motion as well, to varying degrees. Neurons encountered along penetration C responded best to simulated ipsilateral-directed sound source motion. Neurons encountered along penetration D preferred simulated sound source motion toward the receiver and ipsilateraldirected, again with some variability in breadth of tuning.

Velocity selectivity As previously mentioned, each stimulus condition was presented at four different rates of rise o f the A M r a m p ; the first intensity was always approximately 20 dB above

A SE06.015

16 kHz

18

24

30

36

29

35

41

47

50

56

B AU22.026 10 kHzK

C AU22.052 5kHz

-12 0~~

Ib61

-7 -9~ 1 ~ 4 4

87

1

79

38

44

ramp excursion(dB/250msec)

Fig. 9A-C. Plots of spike counts during AM ramps for the four velocity conditions. Directional conventions are the same as in figure 4. The four polar plots refer to the response of one unit at the four different velocity conditions. The four values of ramp excursion (in dB per 250 msec) are printed below each plot. A Monotonic speed-dependent unit: increases in velocity (rate of rise of the AM ramp) produce stronger responses in the preferred

direction. B Non-monotonic speed-dependent unit. A particular velocity (in this case the 3rd of the 4 conditions, corresponding to 41 dB/250 ms) produces the strongest response in the preferred direction; further increases in velocity lead to a diminished response. C Velocity-independent unit: changes in velocity (ramp excursion) have no influence on the strength of response or the directional preference of the unit

Fig. 8A-D. Schematic reconstruction of 3 electrode penetrations normal to the cortex in A1, showing units encountered and their directional preferences. Each polar plot corresponds to one unit; directional conventions are the same as in previous figures

166 the threshold, and subsequent levels increased in steps of 6 dB. The majority of units encountered responded differently depending on the speed and amplitude of the ramp chosen. Differing ramp speeds and excursions are correlates of sound source velocity and range of motion. Of the 61 units which responded to AM ramps, 41 units gave monotonic responses to increased ramp excursion: a faster rate of rise of the AM ramp produced a stronger response in the preferred direction of motion (Fig. 9A). Ten units responded in a non-monotonic, speed-specific way: a particular rate of rise of the AM ramp produced the strongest response; further increases in ramp excursion reduced responding. Again there was no change in preferred direction of motion (Fig. 9B). The remaining ten neurons either showed little speed selectivity over the ranges we employed, giving similar responses regardless of ramp excursion (Fig. 9C), or gave idiosyncratic responses to increased ramp excursion, sometimes accompanied with changes in preferred direction. Discussion

Considerable, highly specific information is available to the auditory system concerning the trajectory of a moving sound source. When a sound source moves along the horizontal plane, alterations in relative interaural intensity, in interaural time differences, and in interaural spectral content result. Obviously, there is a survival advantage for an organism to detect motion in its environment as quickly and efficiently as possible. Aside from their obvious utility for avoiding collisions, motion detection mechanisms may be important for extracting information about the identity and form of a sound source. In principle, one can imagine two different methods to infer source motion from sound intensity in the two ears. One method would be to simply use the same system which is used to localize stationary sound sources, to register the position of the sources at different times, and then compare them to extract direction and velocity of motion. In this case, specialized systems for detecting motion are not needed or employed. On the other hand, psychophysical and neurophysiological evidence (includ, ing the data presented in this paper) point to the existence of specialized motion selective mechanisms in the auditory system. For example, Perrott and Marlborough (1989) found that with moving broadband stimuli presented in a free-field environment, sound localization performance is superior to that which would be predicted if the subject were simply comparing the location of the sound source at signal onset and offset. They found that minimum audible movement thresholds obtained when the sound source was moving were significantly lower than when the end-points of the arc of travel were marked. Although this finding does not necessarily confirm the existence of specialized motion detecting mechanisms, it indicates that such mechanisms must involve more than simple position comparators. In addition, since range (or distance) cannot be calculated directly (in anechoic space) from any of the interaural difference cues described above, motion of a

source moving directly toward or away from the receiver cannot be inferred by a position comparison system, particularly over the range where differences in air absorption for various frequencies would be small. Hence, mechanisms specialized for detecting motion would be the major ones available to an organism in this case, although spectral and reverberatory cues would be available under some conditions. Previous neurophysiological studies have provided some support for the notion that auditory neurons respond selectively to sound source motion or correlates of such sounds. Sovijarvi and Hyvarinen (1974) have found auditory cortical neurons sensitive to the direction of azimuthal sound source movement in a free-field situation using continuous tones presented with a hand-held speaker. Altman (1987) has studied the responses of auditory neurons to modulation of interaural phase disparity in a sealed system, and has found A1 neurons which responded selectively to the direction of phase change, a correlate of auditory motion in azimuth. These reports are valuable as far as they go, but in neither report was the degree of directional selectivity quantified or further analyzed. In addition, relatively few units were analyzed. The results of the experiments reported here indicate that single neurons in the primary auditory cortex of the cat can be highly sensitive to at least one correlate of auditory stimulus motion, namely relative amplitude modulation at the two ears of the receiver. These neurons are quite common, constituting three-fourths of our sample. They respond selectively to the direction of simulated sound source motion and can be classified according to their preferred direction of motion. Motion-in-depth units respond best to correlated increases or decreases in sound source intensity at the two ears; azimuth-motionselective units prefer opposite-directed changes in sound source intensity at the two ears. Monaural-like units respond equally well to changes in sound intensity in one ear alone or in both ears. These direction-selective neurons are spatially segregated from each other and may occur in columns within the cortex. These units also respond selectively to AM ramps of different speeds and excursions, a correlate of sound source velocity and range of movement. It is important to recognize the limitations of the conditions that we have used as simulations of auditory source movement. In the free-field case, changes in interaural time differences and in spectral content would accompany the alterations in stimulus amplitude that we have used in this study. In addition, the pattern of interaural intensity changes that we have presented are not entirely unique with regard to particular directions of motion. A full account of the relationship between alterations in relative aural amplitude and the direction of motion of a sound source of constant strength in threedimensional space is beyond the scope of this paper, but we have addressed this issue in another publication from this laboratory (Zakarauskas and Cynader 1991), in which a mathematical analysis of the set of amplitude modulations associated with motion of a constant intensity sound source moving in three dimensions has been

167 carried out. This analysis has revealed that certain changes in sound level at the two ears are not necessarily associated with one specific sound source trajectory but may also result from a variety of other stimulus conditions. It is possible, for instance, that correlated increases in sound level are associated with stimuli moving tangentially to the receiver as well as directly toward him, resulting in some ambiguities. We are also aware that the stimuli used in this experiment do not represent the full set of binaural intensity modulations produced by sound sources moving in azimuth or in depth. The analysis cited above demonstrated that unlike the AM ramps we used to simulate auditory motion, the rates of intensity change associated with sound source movement are not necessarily linear, nor do intensity changes occur at the same rate at the two ears (in the case of auditory motion in azimuth). In spite of this, we used stimuli with constant rates of change and with similar binaural time courses because they produce an appropriate (albeit simplified) simulation of sound source movement that can be easily implemented in a sealed system. We are well aware of the limitations and ambiguities inherent in efforts to simulate motion of real world sound sources without actually providing that motion. In this paper, we deal only with binaural sound intensity modulation, recognizing that it is by no means the only cue available to a system detecting moving sound sources. On the other hand, it should also be noted that interaural onset time differences do not always represent a single position along the azimuth (Roth et al. 1980). A striking result that can be derived from the acoustic analysis described above (Zakarauskas and Cynader 1991) is that the rate of change of monaural intensity function is found to be directly proportional to the velocity scaled by the distance of a sound source for omnidirectional sources of constant intensities. This emphasizes the importance of a system that uses amplitude modulation as a component of a specialized motion detection system. This contention is supported by our neurophysiological evidence indicating that in addition to responding selectively to the correlates of direction of motion, auditory neurons also show selectivity to correlates of stimulus velocity. In our neuronal population, we found neurons that seem to be sensitive to different rates of intensity change in the two ears. Some neurons preferred low rates of change, while others preferred higher rates o f AM. There was a distinct preponderance of neurons that responded best to the highest rates of A M that we employed. This is in agreement with psychophysical evidence that shows that low rates of change of amplitude are poorly resolved by human receivers (Small 1977).

Analogies to vision This paper has provided evidence for specialized systems dealing with auditory motion in three dimensional space. It is interesting to note that similar mechanisms appear to exist in vision and that like the auditory system, both systems appear to operate by dynamically comparing parameters of the stimuli in the paired end organs. Abun-

dant psychophysical evidence indicates that the human visual system contains specialized channels for dealing with three-dimensional motion and that these channels are distinct from those that process static depth (Richards and Regan 1973; Beverley and Regan 1973; Regan et al. 1979). In addition, neurophysiological studies (Cynader and Regan 1978, 1982) have shown that single cells in cat visual cortex can be highly sensitive to the direction of stimulus motion in three dimensional space. In the visual system, the relative velocity of the two retinal images has been shown to be a sensitive cue to the direction of stimulus motion in three dimensional space (Beverley and Regan 1973; Regan et al. 1979) and visual cortex neurons (Cynader and Regan 1978) show differential sensitivity depending on relative stimulus velocity. This provides a striking parallel with the response properties described here, in which single cells in the auditory cortex appear sensitive to the relative rate of change of intensity in the two ears. It appears that both sensory systems use similar principles to extract three dimensional motion information. The neurophysiological parallels between the two systems are supported by psychophysical evidence indicating that the resolving capacities of the visual and auditory systems for stimulus motion are strikingly similar (Waugh et al. 1979). Perception of stimulus motion in the visual and auditory modalities thus appear to depend on similar central mechanisms of motion detection. The following paper (Toronchuk et al. 1991) comprises an investigation of the mechanisms underlying direction and velocity selectivity. The results show that the selectivity for sound source motion reported in this paper can be related to particular combinations of monaural and binaural response properties that have been studied extensively by other workers.

References

Altman JA (1987) Information processing concerning moving sound sources in the auditory centers and its utilization by brain integrative and motor structures. In: Syka J, Masterton RB (eds) Auditory pathway: structure and function. Plenum Press, New York, pp 34%354 Beverley KI, Regan D (1973) Evidence for the existence of neural mechanisms selectively sensitive to the direction of movement in space. J Physiol 235:17-29 Cynader M, Regan D (1978) Neurons in cat parastriate cortex sensitive to the direction of motion in three-dimensional space. J Physiol 274:549 569 Eisenman LM (1974) Neural encoding of sound location: an electrophysiological study in auditory cortex of the cat using free field stimuli. Brain Res 75 : 203-214 Gardner M (1968) Distance estimation of 0° or apparent 0°-oriented speech signals in anechoic space. J Acoust Soc Am 45 : 47-53 Gourevitch G (1978) Directional hearing in mammals. In: Popper AN, Fay RR (eds) Comparative studies of hearing in vertebrates. Springer-Verlag, New York, pp 352373 Jenkins WM, Masterton RB (1982) Sound localization: effects of unilateral lesions in central auditory system. J Neurophysiol 47:987-1016 Jenkins WM, Merzenich MM (1984) Role of cat primary auditory cortex for sound-localization behavior. J Neurophysiol 52:819-847

168 Kitzes LM, Wrege KS, Cassady JM (1980) Patterns of responses of cortical cells to binaural stimulation. J Comp Neurol t92:455-472 Mendelson JR, Cynader MS (1983) Frequency disparity selectivity in cat primary auditory cortex (A1). Soc Neurosci Abs V9:955 Mershon D, King L (1975) Intensity and reverberation as factors in the auditory perception of egocentric distance. Perception and Psychophysics 18: 409-415 Middlebrooks JC, Pettigrew JD (1981) Functional classes of neurons in primary auditory cortex of the cat distinguished by sensitivity to sound source location. J Neurosci 1 : 107-120 Perrott DR, Marlborough K (1989) Minimum audible movement angle: marking the end points of the path travelled by a moving sound source. J Acoust Soc Am 85:1773-1775 Reale RA, Kettner RE (1986) Topography of binaural organization in primary auditory cortex of the cat: effects of changing interaural intensity. J Neurophysiol 56:663-682 Regan D, Beverley KI, Cynader M (1979) Stereoscopic subsystems for position in depth and for motion in depth. Proc R Soc B (London) 204:485-501

Richards W, Regan D (1973) A stereo field map with implications for disparity processing. Investig Ophth 12: 904-909 Roth GL, Kochhar R, Hind JE (1980) Interaural time differences: implications regarding the neurophysiology of sound localization. J Acoust Soc Am 68:1643-51 Schreiner CE, Cynader MS (1984) Basic functional organization of second audito~ field (AII) of the cat. J Neurophysiol 51 : 1284-1305 Small AM (1977) Loudness perception of signals of monotonically changing sound pressure. J Acoust Soc Am 61 : 1293-1297 Sovijarvi ARA, Hyvarinen J (1974) Auditory cortical neurons in the cat sensitive to direction of sound source movement. Brain Res 73:455-471 Toronchuk JM, Stumpf E, Cynader MS (1992) Auditory neurons sensitive to correlates of auditory motion: underlying mechanisms. Exp Brain Res 88:16%t80 Waugh W, Strybel TZ, Perrott DR (1979) Perception of moving sounds: velocity discrimination. J Aud Res 19:103-110 Zakarauskas P, Cynader MS (1991) Aural intensity for a moving source. Hearing Res 52:233-244

Neurons in cat primary auditory cortex sensitive to correlates of auditory motion in three-dimensional space.

Amplitude modulation at the receiver's ears is a characteristic of moving sound sources. When a sound source moves from side to side, stimulus intensi...
2MB Sizes 0 Downloads 0 Views