Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Computer Analysis of Single-Unit Discharges in the Vestibular Nerve of the Frog JORGE HUERTAS AND RUTH S. CARPENTER Ames Research Center, NASA SUMMARY Description is made of several computer methods lo analyze single-nervous-unit discharges. Their applicability to the study of vestibular units during spontaneous and provoked activity is discussed. A computer window technique that analyzes the frequency, the shortest interval, the longest interval, and standard deviation is described. This technique seems to be particularly suited to describe the changes in activity of vestibular neurons. The results are discussed in lighl of present knowledge of neurophysiology and anatomy. INTRODUCTION During the previous four decades, one of the main activities of the neurophysiologist has been the study of the electrical potentials produced by neural structures. Much advancement in un- derstanding the functions of the brain has been made by correlating electrical activity with the structural arrangement of the nervous system. On the other hand, as methods and technology have improved, understanding the performance at the subcellular level and at the level of the neuron and its neighboring cells has become a challenge. Since the time of Ramon y Cajal, the neuron has been recognized as the functional unit of the nervous system (ref. 1). The neuron receives messages, elaborates messages, produces a response, and also serves as a conductor of the message unit from neural station to neural sta- tion. The neurophysiologist has been intrigued for a long time by the presence of discharges â equal in size and shape but occurring at different intervalsâwhich express neuronal activity. As these discharges in some instances occur at irregular intervals, an analysis of their relation- ship has been difficult, if not impossible, until the advent of the computer. For the study of neuronal activity correlated with the sensory inputs that have a direct effect upon the electrical performance of the neurons, the vestibular nerve of the frog presents several advantages. These advantages arise from the following facts: Direct studies of the nerve can be performed in the intact unanesthetized animal (refs. 2 and 3); exposure of the vestibular nerve for microelectrode recording is accomplished with minimum surgery, which is nondestructive to the nerve, to the blood supply, or to the nervous system; and, most importantly, the nerve can be activated using physiological stimuli. These stimuli can be measured with accuracy by means of accelerometers. There are many descriptions in the literature of computer analysis of neuronal spike patterns (refs. 4 to 7). Some of these methods demon- strated characteristic arrangement in the serial dependence of the intervals that can be described statistically. Most of the papers deal with the "spontaneous activity" resulting from stimulation of the neuron by natural means. The purpose of 137
138 THE ROLE OF THE VESTIBULAR ORGANS IN SPACE EXPLORATION this paper is to present a computer analysis of the performance of the vestibular units during stationary and acceleratory periods. These studies should then demonstrate â 1. How the single unit of the vestibular nerve behaves during periods in which the animal is not submitted to the influences (accelerations) which noticeably alter its performance. For the purpose of this paper, these periods will be referred to as "spontaneous." 2. How the single unit performs when the animal is submitted to acceleratory influences (tilt or rotation). 3. Whether or not there is a constant relation- ship between prestimulus activity and post- stimulus activity. 4. What happens after the animal is returned to the initial prestimulus position. MATERIALS AND METHODS More than 100 single units in the Rana catesbeiana were studied. The frogs were anesthetized in a solution of tricanirie methane- sulfonate (Sandoz MS 222). Immediately there- after, their brachial and lumbar plexuses were dissected and sectioned to prevent voluntary movements (accelerations) from the animal while recording. As the effects of this anesthesia tend to wear off in about 3 hours, this time of elapse was allowed between preliminary prepara- tions and the actual collection of data. By using the technique of Gualtierotti and Gerathewohl (ref. 3), the vestibular nerves were exposed via the roof of the mouth, taking particular care not to damage the endolymphatic organ, the cerebral circulation, the bony labyrinth, or the neural structures. The animal was then placed on a rotating and tilting table. The electrocardiogram (EKG) was monitored constantly. The EKG is a good index of the general condition of the frog. The position of the table in three-dimensional space was monitored by three accelerometers that measured accelerations (positive or negative) in the three spatial vectors x, y, and z (ref. 8). The microelectrodes (tungsten) or micropipets (average tip diameter, 0.5 micron; filled with 4 M NaCl) were driven slowly into the nerve by means of a hydraulic manipulator. The criteria for determination of a stationary vestibular fiber were as follows: Pulses should appear with varying intervals; the minimum interval should not be shorter than 1.5 msec; the size and shape of the spikes should be very similar; and the discharge pattern of the unit must exhibit stability for at least a 20-second period before recording. Recording was achieved (fig. 1) through a sequence that began as the signal was picked up by an emitter follower with an input impedance of 40 megohms (G. Deboo, personal communica- tion). Then the output of the emitter follower was amplified, displayed on a cathode-ray oscilloscope, and also monitored by means of a loudspeaker. The signals were recorded on magnetic tape for further analysis only when we were reasonably convinced that we were dealing with one fiber. Ancillary display systems con- sisted of an inkwriting oscillograph coupled with a ramp generator with an automatic reset, where the interspike interval is then represented by a sequence of saw teeth, the shorter ones corre- sponding to short intervals and the longer ones to long intervals. An Ampex analog magnetic tape recorded the activity of the unit under con- sideration, a fiducial mark that determined the beginning of a sequence, a time mark in real time, the frog's electrocardiogram, and the values expressed on the accelerometers. The computer processing was divided into three steps: digitation of analog data, which for EXPERIMENTAL ARRANGEMENT IN VESTIBULAR LABORATORY [M TILTHS AND ROTATIK TAKE - CLIPPED 1â EMITTER FOLLOIER AMPLIFIER ^ EVEHT MARKER r ACttLEMIETERS i Ft ^ 1 AUDIO â¢* MAMETK TAPE L NOMI1M 1 RECORDER f 1 4 OSCILLOSCOPE 4 I CLIPPER â * tKPLIFIEI m PROCESS j ] T INK t- â¢,. OSCILLOGRAPH Â«- GENERATOR RAMP 0Â»Â£ SHOT -i â¢ BIAS WIT FIGURE I.-Flow diagram demonstrating how the single nenv fiber potentials were processed for computer analysis.
SINGLE-UNIT DISCHARGES IN THE VESTIBULAR NERVE 139 the present purpose is the measurement of the interval between two spikes; statistical analysis of such intervals; and display of the results. This display is done either using the cathode-ray oscilloscope screen or an x-y plotter. This report is concerned with the findings in 90 single fibers. RESULTS Spontaneous Activity Computer analysis revealed a great deal of variability in the range of the firing frequencies from unit to unit. Expressed in terms of the mean firing frequency, values ranged from under 1 per second to 46 per second. This analysis is limited to the slower firing units. Figure 2 is representative of an interval histo- gram of spontaneous activity as described above. The histogram at the upper left shows a fast rise time and an exponential-like decay of frequency of intervals. It shows also that 60 percent of all intervals are contained in the 0- to 250-msec class. The histogram on the upper right represents the same data but with different bounds. It still shows the same pattern; how- ever, as the bounds have been changed, the decay does not look so steep as it was in the first instance. The histogram in the lower left shows the distribution of intervals between 0 and 500 msec. This histogram contains essentially the same information as the first two classes of the first histogram. Finally, the histogram in the lower right shows the interval frequencies con- tributing to the fast rise of the first histogram. This illustration shows very clearly how the same data plotted within different bounds have a seemingly different appearance. This must be kept in mind for statistical analysis of the data. The histograms for all fibers studied were uni- modal and were skewed to the right. The mean interval in these histograms changed with each unit; however, we noticed that the shortest interval, the longest interval, and the mean inter- val maintained a dynamic relationship. Other authors have found unit-specific stability (ref. 6), and our findings confirm this observation. Provoked Activity Analysis of the short-term provoked activity of the neurons of the vestibular system present special problems resulting from the physical characteristics of the system. Here the stimu- lating force is an acceleration that by definition involves a rate of change. This is in contraposi- tion to other forms of stimuli acting upon other receptors in which the stimulus control can be maintained within set limits. When a neuron becemes activated-by an-atajeleration, the firing pattern changes suddenly. Figure 3 shows how angular acceleration alters the pattern of inter- spike intervals. If the acceleratory force is applied in one direction, the intervals become shorter (facilitation); if the force acts in the opposite direction, the intervals become longer (inhibition). It has been found that, in the hear- ing organ, the same stimulus applied a second 100 30 60 I20 Â£I00 Â§*> E 60 40 i ao wo TO iwo mo eco irÂ» woo ii IE M in son cs ix in TO o 20 0 EVENT MARKER â¢ a at K m vi M i u n i H o n ii UPPER LIMITS OF SPIKE INTERVALS, msec FIGURE 2.âInterval histograms of spontaneous activity of a frog vestibular unit using four different bounds. INTERVAL, . I sec -5f 0L EKfl ACCEL Y â¢ TIME CODE Un^illli FIGURE 3. â Responses of a vestibular nerve fiber to horizontal angulnr acceleration.
140 THE ROLE OF THE VESTIBULAR ORGANS IN SPACE EXPLORATION time does not necessarily produce the same in- terval response (ref. 9). In the vestibular organ. it is far more difficult to repeat the same stimulus several times, and we are still in the process of making these comparative studies. The statistical sample necessary to obtain an interval histogram from a single unit must be rather large. Yet, during the period in which the unit is being excited, it is undergoing a con- tinuous change in performance. Therefore, it is difficult to obtain a large enough sample that can be compared statistically with the prestimu- lus period. We are dealing here with two seem- ingly different time series: the spontaneous activ- ity time series and the stimulated activity time series. It has been suggested that the time pat- terns in spike discharges are influenced by both stimulus and refractory properties of the neurons. Therefore, the stimulus histograms should re- flect the continued influence of these two factors, the "recovered probabilities" (ref. 10). The question of how long a sample must be during a stimulus remains unanswered. Stim- ulation during 2 seconds of angular acceleration produces a response that is related to the vec- tor of the stimulus. During the 2 seconds in which the acceleration is applied, the unit responds either in a facilitatory or an inhibitory pattern. During facilitation the intervals are very short and compressed in time. During inhibition the long intervals provide a very meager sample. Moreover, in the examples under consideration the acceleratory rate is con- tinuously increased. No two spikes occur during the same acceleratory value. There are two procedural routes that can be used for the analysis of intervals during short periods of stimulation. A set number of spikes may be taken as an index of activity and to es- tablish a relationship between the number of spikes and the time during which they occur (fig. 4). The computer was programed to count a predetermined number of spikes and clock the time during which they occur. In this case, the spikes are grouped in sets of 25. The abscissa represents elapsed time. If the intervals are long, the unit of time in which they will occur is long; if the intervals are short, the predetermined number of spikes will occur in a short period of time. If the effect of stimulation is facilitatory, the time lapse becomes short; if the effect is inhibitory, the time lapse becomes longer. The arithmetical mean for each group of 25 spikes is represented by the heavy line and the maximum and minimum intervals are repre- sented in figure 4 by dots and dashes above and below the mean, respectively. The value of the mean is changed noticeably when the neuron is activated, a fact noted previously by Nakahama (ref. 11). Yet, regression analysis does not prove that there is a direct relationship between the mean values and a given stimulus. The second procedure to be considered here involves the use of an arbitrary time window and observation of the behavior of the spike intervals during these time limits. As the angular ac- celeratory forces that were used in our experi- ments lasted for about 2 seconds, 2-second windows were selected to study neuronal activity. Figure 5 demonstrates such a window. The duration in real time for each event displayed here is 2 seconds. The number of intervals is expressed in percentage, and the bounds for the histograms have been established between 0 and 500 msec, subdivided into 50 classes of 10 msec each. During the spontaneous firing event represented in the first histogram, 13 spikes occurred in 2 seconds. The next 2 seconds were 500 r 400 E300 200 100 â¢ MAXIMUM â MEAN - MINIMUM 50 100 150 TIME, sec 200 250 FIGURE 4. - Responses of a vestibular unit of the frog to hori- zontal angular acceleration. Each point represents the average of 25 intervals.
SINGLE-UNIT DISCHARGES IN THE VESTIBULAR NERVE 141 SPONTANEOUS ou 40 20 H 13 0 HI? I Â° â¢ â¢III . ,i.l|..l.. 1 â¢ 1 ^â¢H,.',. l.,,I. ..I l.|,...I,. .1 g60 o. ROTAT ION -PLUS >40 L N 112 . i Â»59 o â¢ z ao UJ Â§ 0 UJ J. 1 , I', ii'...I 'â¢L i. - , i Â£60 40 20 n ROTATION-MINUS H10 - HI? il III II ,I ,|,,,,I ,| ii iii i .L..i.l 0 0 100 ZOO 300 400 500 0 100 200 300 400 500 TIME, msec FIGURE 5.â Interval histograms of consecutive 2-second samples of a vestibular unit that responds to rotation. populated by 17 spikes. In both histograms there is a relatively good spread of intervals among the different classes. These histograms can be described in terms of the arithmetic mean, standard deviation, largest interval, and shortest interval, together with the number of spikes occurring during each event. The second row of histograms in figure 5 shows the same windows during facilitatory stimulation. Here the number of spikes has been increased by a factor of 8.6, and all the intervals have shifted toward the shorter periods. When deceleration occurs, the neural unit becomes inhibited. This is ex- pressed by longer intervals and a slight decrease in the number of spikes. The third row shows the distribution of intervals when the animal is rotated in the opposite direction. Here inhibi- tion occurs, and the number of intervals diminishes. The method that was selected and devel- oped for the present study stresses the time sample, but it also takes into consideration the number of spikes. The program written for this purpose produces the following data: It divides the time events into consecutive 2-sec- ond epochs, counts the number of intervals within this period, and finds the longest and the shortest interval. It also calculates the mean, standard deviation, frequency distribution, and percentage frequency. Figure 6 shows data similar to those displayed in figure 5, but with values obtained by the window technique plotted on semilogarithmic paper. The abscissa represents real time, and the number of intervals counted for each 2 seconds are registered at the top. The values for each window, maximum interval, mean in- terval, standard deviation, and minimum inter- val, are plotted for each time period. The plot for spontaneous activity obtained in this manner shows how these four values remain within a limited range, and fluctuation of maxi- mum and minimum intervals occurs propor- tionately without disorderly dispersion or conver- sion. The mean and the standard deviation remain within very close limits, and in a few instances we have seen them expressed by the same value. When the unit is stimulated (in this case by rotation), all four values shift in the same direc- tion. If there is facilitation, both the maximum and minimum intervals become shorter con- comitantly. The parallelism observed during spontaneous activity is shifted, and the log- arithmic differences between the maximum and minimum intervals remain seemingly unaltered. In the case of inhibition, there is a shift in minimum and maximum to longer intervals, which are clearly defined by the increase in value of the mean interval. The window technique allows the physiologist to observe four statis- tical parameters at the same time. This is in marked contrast to other displays such as SKttssnn inn s/urus FIGURE 6. â Firing patterns of a vestibular unit at rest and during horizontal rotation. The numbers along the top represent the number of intervals that occurred during each 2-second period.
142 THE ROLE OF THE VESTIBULAR ORGANS IN SPACE EXPLORATION histograms in which only one parameter is ob- served and time dependencies are lost. The window technique also allows a comparison of the number of intervals that occur during a given period of time and serves as an index as to whether the vestibular unit is firing spontaneously or is being influenced by a stimulation. HW 77 7J U I I2 33 S5 8 PI ?Â« ?3 28 0 57 J7 2I SUCOSSIff TIME SiMPIES FIGURE 7. â Firing patterns of a vesfibular unit at rest and during horizontal rotation. The numbers along the top represent the number of intervals that occurred during each 2-second period. Figure 7 illustrates the behavior of another unit that is essentially the same as the one dis- played in figure 6; however, the responses are more marked. It also exemplifies how the firing pattern returns to the prestimulus level. Figure 10,000 iOOil i | I 100 55 55 53 51 52 54 53 51 52 5Â« 54 54 58 58 55 57 53 61 51 40 45 5T 57 SPOMTAMEOUS TILT LEfT TUT - RIGHT - MAXIMUM - M[AN - STANDARD DEVIATION -- HIIIMUM SUCCESSIVE TIME SAMPLES FIGURE 8. â Vestibular nerve unit that is not sensitive to tilt. Note how the values remain within the prestimulation limits during tilt. EVENT MARKER VESTIBULAR UNIT EKG Z AXIS Y AXIS TIME CODEââ EVENT MARKER VESTIBULAR UNIT EKG Z AXIS Y AXIS TIME CODE FIGURE 9. â Response of a vestibular nerve unit to linear acceleration.
SINGLE-UNIT DISCHARGES IN THE VESTIBULAR NERVE 143 8 is an example of a unit that is not sensitive to tilt. Notice how the values remain within the same range during stimulation. In the case of vestibular units that respond to tilt, theoretically the situation should be different because tilt can be maintained at the same value for a long period of time. Therefore, a compari- son of a spontaneous interval activity and an equivalently long sample obtained during tilt should be possible. Figure 9 shows the perform- ance of a unit that responds to tilt. It is obvious that this is not a steady state and that with time there is a trend toward adaptation. Therefore, an interval histogram is not representative of the active process that is taking place. Because of this fact, the window technique is also better suited for the analysis of the behavior of these units that respond to tilt. Figure 10 exemplifies the sequence of events that characterize the per- formance of such a unit. By use of this method, the trend toward adaptation in time can be ob- served with accuracy during sustained tilt. In one instance a unit was followed with this trend for 80 seconds, and the unit did not return to the baseline values of spontaneous activity. At the present time we are in the process of studying more of these long-term shifts. CONCLUSION The vestibular units that have been found to date can be classified as follows: (1) units that respond directionally to rotation only, (2) units that respond to tilt only, (3) units that respond to tilt and rotation, and (4) units that do not respond to the acceleratory stimuli applied. The spon- taneous activity for these types of units varies markedly in range. The different frequencies were represented in all four types; however, the faster frequencies were found in those units that responded to rotation. In spite of this variability, each fiber seems to conform to a typical firing pat- tern, and the values of the longest and the shortest intervals fluctuate about a characteristic mean. Despite this, the spontaneous discharges do not predict the degree of change during stimulation. The ratio of change can be greater in some of the slowest firing units. For other units, the pattern of the prolonged response to tilt is characterized _B it B Â« H Â» Â« â¢ JI>B S 8 J4 M Â» Â» a Â« 8 Â» Â» B â¢ 8 a Â» UittIt) mi : un - FIGURE 10. â Comparison of the spontaneous firing patterns of a vestibular unit and the pattern obtained during tilt. The numbers along the top represent the number of intervals that occurred during each 2-second period. by increasingly longer intervals proportional to the length of time during which the unit remains constantly stimulated. The shifts that we found in the values for the maximum interval, minimum interval, and stand- ard deviation suggest that peripheral and/or central biasing mechanisms must exist, of which the dynamic range of the spontaneous discharge is an expression. This mechanism must involve a regulatory system that inhibits or releases the first-order neuron from its in- fluence according to a pattern of information. Perhaps this originates at other sensory terminals of the same organ, as is the case in the retina, or at the level of the central nervous system, as is the case in the muscular sensory system. The long periods of inhibition during linear ac- celeration suggest the importance of inhibition for carrying information. Our findings are in agreement with what has been found in single- unit work of other sensory systems, particularly the visual and auditory systems. To understand the performance of the first- order neuron of the vestibular system, a great amount of research is still needed. This re- search must be carried out at the cellular level using electron-microscopy techniques, at the end-organ level using histochemistry procedures, and at the level of the central nerv- ous system using neurophysiological techniques. The computer analysis of these data will help in the progress of such an endeavor.
144 THE ROLE OF THE VESTIBULAR ORGANS IN SPACE EXPLORATION REFERENCES 1. RAMON Y CAJAL, S.: Ses Nouvelles Idees sur la Struc- ture du Systeme Nerveux chez I'Homme et chez les Vertebres. Deuxieme tirage. Traduction de 1'Es- pagnol par le Dr. L. Azoulay. C. Reinwald, Paris, 1895. 2. Ross, M.: Electrical Studies of the Frog Labyrinth. J. Physiol., vol. 86, 1936, pp. 117-146. 3. GuALTIEROTTl, T.; AND GERATHEWOHL, S. J.: Spon- taneous Firing and Responses to Linear Acceleration of Single Otolith Units of the Frog During Short Pe- riods of Weightlessness and During Parabolic Flight. The Role of the Vestibular Organs in the Exploration of Space, NASA SP-77, 1965, pp. 221-228. 4. KuFFLER, S. W.; FITZHUGH, R.; AND BARLOW, H. B.: Maintained Activity in the Cat's Retina in Light and Darkness. J. Gen. Physiol., vol. 40, 1957, pp. 683-702. 5. WERNER, G.; AND MOUNTCASTLE, V. B.: The Variability of Central Neural Activity in a Sensory System, and Its Implications for the Central Reflection of Sensory Events. J. Neurophysiol., vol. 26, 1963, pp. 958-977. 6. RODIECK, R. W.; KIANG, N. Y.; AND GERSTEIN, G. L.: Some Quantitative Methods for the Study of Sponta- neous Activity of Single Neurons. Biophys. J., vol. 2, July 1962, pp. 351-368. 7. MOORE, G. P.; PERKEL, D. H.; AND SEGUNDO, J. P.: DISCUSSION Loweiistein: How much hysteresis did you find on re- peated stimulation with the same magnitude of stimulus, say in utricular preparations? Huertas: Hysteresis was always found. It varied from case to case. I have no measurements on hand, but hys- teresis is present. Pompeiano: In your experiments was the vestibular nerve contralateral to the recorded side cut or not? Huertas: No; I have not recorded that. It is very hard to get equivalent fibers. In my attempt to do so I kept getting two differently acting fibers all the time. To establish a mathematical order with one fiber is hard enough. Precht: You mentioned that you got quite different re- sponses when you tested the same unit more than one time with the same acceleration. The illustration makes the reason for this difference in response quite obvious; the stimulus was not the same in each test. Huertas: No. Precht: The second point is. Did you try to correlate the shortest spike intervals measured in a single unit during constant stimuli of various magnitude with the stimulus intensity; that is, did you try to get a stimulus-response relationship? Huertas: Do you mean did I correlate the stimulus with the minimum interval only? Precht: Yes. Statistical Analysis and Functional Interpretation of Neuronal Spike Data. Ann. Rev. Physiol., vol. 28. 1966, pp. 493-522. 8. ALLEN, WILLIAM H. (ED.): Dictionary of Technical Terms for Aerospace Use. NASA SP-7, 1965, table XI. 9. SlEBERT, W. M.: Some Implications of the Stochastic Behavior of the Primary Auditory Neurons. Kyber- netik. vol. 2, June 1965, pp. 206-215. 10. GRAY. P. R.; KIANG. N. Y. S.; AND SHIPLEY, S. W.: Prob abilities Associated With Spike Discharges in Auditor) Nerve Fibers. Proc. Am. Physiol. Soc., 17th Meeting, Aug. 1965. 11. NAKAHAMA, H.: Relation of Mean Impulse Frequency to Statistical Dependency Between Intervals in Neuronal Impulse Sequences. J. Neurophysiol., vol. 29, 1960, pp. 935-941. Huertas: Yes, we measured automatically the minimum interval, and the minimum interval was different; it varied within a few milliseconds to a similar stimulus. Precht: Is there any kind of a mathematical relation between the stimulus and the response, for instance, linear or logarithmic? Huertas: Not if one uses only one parameter. We are attempting to find a definite answer to your question in the near future. Graybiel: My comment is in regard to the behavioral aspect. If you tilt a person and measure ocular counter- rolling, the roll will remain about the same value for a period of hours. If you expose a person under conditions wherein he observes a change in direction of a line of light in response to a chaqge in direction of the resultant vector with respect to the observer, there is a dynamic phase, followed by a static phase; there is no decay over a long period of time, at least up to 'H hours. Money: Did the height of the spikes vary with acceleration or did the height of the spikes remain constant? Huertas: It remained constant. But one of the beliefs in neurophysiology is that the spikes of the same neuron are exactly alike. They are not. Therefore, other criteria for stationary are needed. Let me answer your question this way: They remain within the same patterns as before ac- celeration.