TRANSIENT AND STEADY STATE RESPONSES 0F CAT OPTIC TRACT FIBERS T0 CHANGES IN LUMINANCE AND SIZE OF STIMUU Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY RAY WYATT WINTERS 1969 fir- [— This is to certify that the thesis entitled TRANSIENT AND STEADY STATE RESPONSES OF CAT OPTIC TRACT FIBERS TO CHANGES IN LUMINANCE AND SIZE OF STIMULI presented by Ray Wyatt Winters has been accepted towards fulfillment of the requirements for Mdeqree in Psychology 3&3 Major professor Date August 15, 1969 0.169 '7'" a: a“; f; ..... 3? (swfi‘ "4 h/lILlIIEE‘t.-I :atc Univcxmi y _ 1!’ «gal -.:QJ"f-.r ABSTRACT TRANSIENT AND STEADY STATE RESPONSES OF CAT OPTIC TRACT FIBERS TO CHANGES IN LUMINANCE AND SIZE OF STIMULI BY Ray Wyatt Winters Single unit recordings were made from cat optic tract fibers while varying the luminance and size of spots placed in the center of a cell's receptive field. Both transient and steady state stimulus-response curves could be divided into two segments: a monotonic increas— ing portion for low contrast levels (less than 2.2 log units above threshold) and a non-monotonic decreasing segment for high contrast levels (above 2.2 threshold unitsL‘with a greater amount of high contrast decline for off-center cells. The shape of the stimulus-response functions were not found to vary significantly for dif— ferent spot sizes. Three conclusions were made: (1) that cat reti- nal ganglion cells give a more precise representation of intensity changes in the low contract range; (2) that the intensity code is non—linear for this range; (3) that more information about intensity changes is signalled Ray Wyatt Winters during the transient component, than during the steady state component of the response. The results were interpreted in terms of recep- tive field models presented by Rodieck and Stone (1965) and Rodieck (1967). Approved By: Date: Thesis Committee: Dr. S. Howard Bartley, Chairman Dr. Glenn I. Hatton Dr. John I. Johnson Dr. Robert L. Raisler TRANSIENT AND STEADY STATE RESPONSES OF CAT OPTIC TRACT FIBERS TO CHANGES IN LUMINANCE AND SIZE OF STIMULI BY Ray Wyatt Winters A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1969 ACKNOWLEDGMENTS As is so often the case, it is not practical to name all of the people who have assisted me in this re- search project. Nevertheless, my gratitude is extended to all of them. I am most assuredly indebted to the chairman of my guidance committee, Dr. S. Howard Bartley. I will always remember him as a friend as well as a teacher. I am also grateful to Dr. Richard J. Ball, Dr. Glenn I. Hatton, Dr. John I. Johnson and Dr. Robert L. Raisler for their assistance and guidance during my graduate training. I also wish to thank: Andy Harton for his assist- ance in histology; John Haight for his advise on elec- tronic matters; Stan Cohen for help with the statistics; and Rick Bectal, Terry Hickey, Larry Hutton and Mike Kelley for their assistance in every phase of the research. I cannot overemphasize the importance of Jim Walters' timely assistance and encouragement. I cannot imagine this research being completed without his help. The extreme patience and support shown by my wife are immeasurable. I thank her for being so understanding. This research was supported by predoctoral fellow- ship 5-F1-MH-38, 110-02 (PS), NIH Grants NBOS982 and ii GM10890 and NASA contract NSG-475. It was conducted in the Laboratory of Comparative Neurology, Departments of Biophysics, Psychology, and Zoology, Michigan State University. iii TABLE OF CONTENTS LIST OF TABLES AND FIGURES . . . . . . . . INTRODUCTION 0 O O O I O O O O O O O O O 0 METHOD 0 O O O O O O I O O O O O O O O O 0 Subjects and Apparatus . . . . . . . . Recording System . . . . . . . . . . . Procedure . . . . . . . . . . . . . . . RESULTS 0 O I O O O O O O O O O O O O O O Area-Threshold Studies . . . . . . . . Suprathreshold Studies . . . . . . . . Best Fit AnalYSiS I O O O O O O O O O 0 DISCUSSION 0 O O O O O O O O O O O O O O 0 High Contrast Effects . . . . . . . . . Spatial Summation at Threshold . . . Spatial Summation for Suprathreshold Stimuli Intensity Code . . . . . . . . . . . . Mechanisms . . . . . . . . . . . . . . BIBILOGMPHY O O O O O O O O O O O O O O 0 APPENDIX A: ELECTRODE TRACK TRACINGS . . APPENDIX B: EXPANDED VERSION OF THE METHOD SECTION C O O O O C C C 0 iv Page \J ooqq l3 13 16 41 45 45 45 47 49 55 65 68 75 Table 1. LIST OF TABLES AND FIGURES Summary of best fit analysis for power, logarithmic, and linear functions . . . . . Figure l. 10. 11. Schematic delineating how response magnitudes were measured for on-center and off-center cells I O O O I O O O O O O O I O O O I O 0 Approximate retinal locations of receptive fields of optic tract fibers studied . . . Area-threshold curve for a typical on-center cell 0 O I I O O I O O O O O O O O O O O 0 Transient and steady—state stimulus-response functions for an on-center cell . . . . . . Transient component stimulus-response func- tions for an on-center cell . . . . . . . . Steady state component stimulus-response functions for an on-center cell . . . . . . Average response histograms for an on-center cell 0 O O O O O O O O O O O O I O O O O 0 Graph of latency changes for an on-center cell 0 O O O O O O O O I I O O O O I O O O Stimulus-response function for the transient component of an off-center cell . . . . . . Stimulus-response functions for the transient component of an off-center cell. Four spot sizes are shown . . . . . . . . . . . Average response histograms for an off-center cell . . . . . . . . . . . . . Page 43 10 14 17 19 22 24 28 32 34 36 39 Figure 12. 13. 14. 15. A1. A2. A3. Bl. B2. BB. B4. B5. Theoretical area--constant response curves Rodieck and Stone (1965) receptive field mOdel O O O I C O O O O I O , O O O O O O 0 Schematic of mechanisms proposed to account for observed stimulus-response functions Rodieck (1967) receptive field model . . . Electrode track tracings . . . . . . . . . Electrode track tracings . . . . . . . . . Photomicrograph of section showing lesion . Photograph of photic source . . . . . . . . Block diagram of the recording system . . . Photograph of cat in the headholder . . . . Schematic (If the reversible Ophthalmo- scopic technique . . . . . . . . . . . . Block diagram of the data analysis system . vi Page 50 56 6O 63 69 71 73 76 78 82 87 9O I NTRODUCTION The stimulus dimensions of wavelength and inten- sity have, for the most part, been overlooked by investi— gators concerned with cat retinal ganglion cell activity (Bishop, 1967). Extensive examinations of the effects of these stimulus dimensions can be found only in several studies conducted by Granit (1944, 1947, 1955). From Granit's 1947 and 1955 studies it can be concluded that cats have very few retinal ganglion cells which relay information that could be used as a basis for spectral discrimination. Opponent process cells (i.e. color coded cells), which have been found to be predominant in mammals which show Spectral vision (DeValois 1965, Gouras 1968, Michael 1968, Wiesel and Hubel 1966), were rarely encountered with cats. Behav- ioral research (Mello and Peterson 1964, Meyer and Ander- son 1965, Sechzer and Brown 1964) also suggests that cats are relatively deficient in spectral vision. In 1944 Granit studied the effect of variation of intensity of photic stimulation upon the activity of single retinal ganglion cells. To date, this study has been the most extensive investigation of intensity cod- ing in the cat's retina. The stimulus that Granit used was a diffuse, one second photic pulse whose intensity was manipulated over a 6 log unit range. The pulse was shone directly into the cat's eye. For some cells, firing rates were measured during a one second period of time after stimulus onset. For others, firing rates were ob- served at the cessation of stimulation. His results showed irregular relations between intensity and firing rate with very few cells showing the same stimulus- responsel functions. Relationships were only monotonic over limited ranges. In light of the cat's excellent ability to discriminate intensities (Mead, 1942), Granit's results were somewhat puzzling. At best, it can be said that he did not find the code for intensity. The present study was designed to ask the same basic question that was asked in Granit's 1944 investi- gation: how do single retinal ganglion cells signal relative intensity changes? The approach, however, dif- fers from Granit's in two major ways: (1) Granit varied the intensity of large, diffuse, luminous spots. In the present study intensity was manipulated for small focal stimuli placed in the center of a unit's receptive field. (2) Granit measured response magnitude by counting the number of Spikes that occurred during a one second period 1"ReSponse" is used here to refer to Spike activity 0f single retinal ganglion cells. Response magnitude re- fers to the number of Spikes per unit of time. of time. In the present investigation transient and steady state components of a response were analyzed separately. The rationale for the two modifications just mentioned is given below. Although there are several exceptions (Rodieck 1967b, Spinelli 1966, Stone and Fabian 1966), most re- search (BishOp 1967) has shown that cat retinal ganglion cells have receptive fields in which center and surround regions are mutually antagonistic. Two types of cells are usually found: on—center, off-surround cells and off-center, on-surround cells. For an on-center cell, presenting a luminous spot in the center of the receptive field causes an increase in firing at the stimulus onset. A luminous pulse presented in the surround of the receptive field causes an increase in firing at the pulse's termina- tion. The response for center and surround are reversed for off-center cells. Since the central regions are more sensitive than the surround regions (Kuffler 1953, Rodieck and Stone 1965), the more effective stimuli are ones which primarily activate the center of the receptive field, i.e., are not large enough to invade the surround. Using dif- fuse Spots as stimuli as Granit did, would maximize spatial antagonism and thus lead to minimal activation of the cell. His use of other than Optimal stimuli may have been the reason that he found irregular stimulus-reponse relation- ships. In any case, it seems more reasonable to use stimuli which maximally trigger the cell under scrutiny. Therefore, intensity coding mechanisms were investigated in the present study with small luminous spots placed in the center of a unit's receptive field. Rodieck and Stone (1965) have shown reliable time course changes in the response of single ganglion cells to luminous pulses. For an on-center cell, for example, there is an initial burst of firing followed by an ex— ponential decay to a firing rate of about 1/5 the magni- tude of the initial phase of the response. For a one second luminous pulse, the firing rate remains at this level until the illumination terminates. The initial phase, which lasts about 75 msec. is called the transient component of the response. Firing rates can reach as high as 700 spikes/second during this phase. For a one second luminous pulse Rodieck and Stone defined the activity during the last 300 msec. of the pulse as the steady state component. Firing rates for this component reach as high as 175 spikes/second. Off—center cells show a similar sequence at stimulus termination. The transient firing rate, however, returns to the maintained (sponta- neous) level rather than to an intermediary magnitude. According to Rodieck and Stone the transient phase, for an on-center cell, for the most part, represents the ef- fects of excitatory inputs to the cell while the steady state component reflects the sum of both excitatory and inhibitory inputs. For an off-center cell the transient component is thought to be the result of the release of inhibitory influences. Granit measured response magni- tude by summing spikes over a one second period. Differ- ences in the signals of intensity changes coming during the transient and steady state components could have been concealed by this measure. Thus, in the present study transient and steady state components of the re- sponse were analyzed separately. The value of the approach advocated here has been demonstrated by a recent study by Stone and Fabian (1968). In varying the intensity of a small luminous pulse placed in the center of a unit's receptive field, they found a monotonic relationship between stimulus magnitude and the maximum firing rate (of the transient component). The shapes of the stimulus-response func- tions under these conditions—~in contrast to those found by Granit--were the same for all retinal ganglion cells studied. Their primary concern was not intensity coding mechanisms so they only examined the effects of low con- trast stimuli. They did not examine changes in the re- sponse of the steady state component nor did they observe the effects of more than one size of spot. The present investigation can be considered as an expansion of the Stone and Fabian study. It was in- tended as a step toward understanding intensity coding mechanisms in the cat's retina. The mechanisms were studied under the following conditions: (1) small lumi- nous pulses (from .1° to 7.0°) were placed in the center of a unit's receptive field, (2) pulses were 920 msec. in duration, (3) ambient background levels were 0 or 1 log candle/m2, (4) photic intensities ranged up to 4.2 log units above threshold. METHOD Subjects and Apparatus Single unit recordings were made from 81 (44 on- center cells and 37 off-center cells) optic tract fibers of normal, adult cats. The cats faced a diffusely lit (either 0 or 1 log candle/m2) 1-1/2 meter translucent tangent screen. The photic source, a modified 750 watt slide projector was placed behind the screen. The Size of the luminous spots flashed on the screen was controlled by placing metal slides in the projector. A metal neutral density filter holder and a variable speed shutter were attached to the movable lens of the projector. Recording System Glass insulated tungsten microelectrodes were used in a conventional capacitance coupled recording system. After being amplified and filtered for low fre— quencies, Spike activity was passed through a Schmitt trigger circuit. The Schmitt trigger output was fed to an audio monitor and to one channel of a tape recorder. The second channel of the tape recorder was used to re- cord pulses that were synchronized with the stimulus. Procedure A11 surgery including cannulation of the femoral vein, a bilateral cervical sympathectomy (to minimize eye movements, Rodieck §t_21. 1967), tracheotomy, crainotomy, and the mounting of the cat's head in a head holder was performed under Pento Short (a short acting bartituate given intravenously). Upon completion of surgery the animals were elec- trically decerebrated by a technique Similar to the one described by Martin and Branch (1958). In order to im- mobilize the eyes an intravenous injection of 80 mg of Flaxedil was given. A mixture of this drug (28 mg/hour) and physiological saline (6 cc/hour) was continously in- fused through the femoral vein for the duration of the experiment. Respiration was maintained by a Harvard Apparatus respirator pump (stroke volume 45 cc; rate, 20/minute). Body temperature was held at 38° C by a heating blanket connected in a 12 volt D.C. circuit. Corneal contact lens were fitted to prevent corneal clouding. A ten per cent solution of neosynephrine was used to keep the pupils dilated. Retinoscopy was per- formed in the usual manner with spherical and cylindrical lens appropriately placed in front of the eyes to correct refractive errors. The methods used for locating the Optic tract were the same as those outlined by BishOp et a1., 1962a. Once a fiber was isolated its receptive field was mapped from the rear of the tangent screen with .1°, luminous spots. Threshold Studies.--Thresholds for various sized spots--placed in the most sensitive portion of the recep— tive field--were determined by varying their intensity in .1 log unit steps while listening to changes in activity over the audio monitor, i.e. the signal which had been filtered and passed through the Schmitt trigger circuit. For eight of the 21 cells studied, thresholds were also determined by analyzing changes in maintained (spontan- eous) firing rates with a computer of average transients (C.A.T.). The threshold values from the two techniques were never more than .1 log unit apart. Suprathreshold StudieS.--The luminance of various spots, ranging from .1° to 7°, was varied in .3 log unit increments over a 4.2 log unit range. Because most dif- fusing surfaces are directional, luminances have been ex- pressed in threshold units i.e. log units above threshold intensity or in terms of neutral density filter values (where smaller values represent greater luminance levels.) The duration of stimulus pulses was 920 msec.; their rise time was 5 msec. Each stimulus was repeated 15 times at a rate of .2 cycles/second and was presented in the most sensitive portion of the receptive field. Figure 1 shows how response magnitudes were measured. 10 Figure l.--Schematic delineating how response magnitudes were measured for on-center and off-center cells. Two response measures were used for on-center cells; (1) the average firing rate (for 15 stimulus cycles) during the first 75 msec. of the response, i.e., the transient com- ponent; (2) the average firing rate during the last 300 msec. of the response--the steady state component. Be- cause the steady state component for off-center cells these cells. nituis Two I 9 t“ ADV ; 300 Be- ll 1’ transient 5 / steady state U On .2 / Center W I I a I ' :es ' ' e i ‘ I {—75—t: l4——300—+ msec. msec. stimulusr 6” j t 53.9. I 1; transient : 3 / on 2‘"; Center 3 . J I . o ":1 : malnquned m , actlwty M—75-ti msec stimulusl JP Figure 1 12 Upon completion of recording from a unit, its receptive field location, with respect to the area cen- tralis, was determined by the reversible opthalmoscopic technique (Bishop et al., 1962b). After completing work in a puncture a microlesion was placed at the recording site. Recording locations were histologically confirmed. RESULTS Confirming the reports of other investigators (Hubel 1960, Kuffler 1953, Wiesel 1960, Rodieck and Stone 1965) the optic tract fibers that were studied showed concentrically arranged receptive fields where center and surround regions were mutually antagonistic. The less common types of receptive fields reported by Rodieck 1967b, Spinelli 1966, and Stone and Fabian 1966, were not encountered. The central regions of the receptive fields ranged from .6° to 7° in diameter (at the widest margins) with smaller receptive fields, in general, being found nearer the area centralis (Figure 2). Mapping experiments dem- onstrated that the central portion of the fields were more sensitive than surround regions (Kuffler 1953, Rodieck Thus, as a rule, units could be cate- and Stone 1965). gorized by their response to ambient illumination: on— center, off-surround cells gave an on response and off- center, on-surround cells gave an off response. Agea-Threshold Studies Area-threshold curves were determined for 12 on- center cells and 9 off-center cells. Most of the functions 13 14 Figure 2.--Approximate retinal locations of receptive fields of units recorded from the left optic tract, as inferred from the reversible ophthalmoscopic tachnique. All points in the 1st and 4th quadrants represent re- ceptive fields from the nasal hemiretina of the right eye. Points at greater distances to the right of the origin are more nasal to the area centralis of the right eye. All points in the Second and third quadrants rep- resent fields in the temporal hemiretina of the left eye. Points at greater distances to the left of the origin are more temporal to the area centralis of the left eye. The origin of the graph represents the area centralis for both the left and right eyes. The two dots which lie on the vertical axis represent right eye receptive fields. Dots symbolize on-center cells; tri- angles symbolize off-center cells, with sizes proportional to receptive field center size. Three nasal units from the right eye are not plotted. Maps were not made for five units. >tive t,fi nique gm th 15 superior left Eye 45A A Right Eye - . . . 30 .A : o ‘ A o . ~" ‘. . 15 " A . .1 a ‘. oU ":’. apls. ‘ I. t I n J ‘ i 4 'vela 4 ‘ ‘ empora 30 IS. ‘ offil‘S ‘30. 45 60nasal ‘ A ‘.O‘ . s '15‘ ‘ ' ‘ . ‘ . ‘ 30- A 45- inferior Figure 2 16 were similar in shape to the one shown by the cell in Figure 3. Starting with the smallest spot, increasing spot Size resulted in a drop in threshold until the edge of the central region--as determined by the mapping technique——was reached. Beyond this point the curves either became horizontal or a small increase--no greater than 1/3 of a log unit--in threshold occurred. The mean value of the slope of a least squares straight line fitted to the decreasing portion of the curve was determined to be -1.48 (S.D. = .54) for the 12 on-center cells and -1.43 (S.D. = .41) for the 9 off-center cells. A "t" test comparing these two means showed them not signifi- cantly different (t = .12, df = 19, p < .20). Slopes in the range of -l.45 fall between those predicted by Ricco's and Piper's laws. (Figure 3) Suprathreshold Studies: High vs Low Contrast Stimuli On-Center Cells.-—As Figure 4 shows, intensity variation had similar effects upon the shape of the transient and steady state component stimulus-response curves. The functions shown in the figure can be divided into two segments: an increasing portion for low con- trast levels (up to about 2.2 log units above threshold) and a decreasing segment for high contrast levels (beyond 2.2 10g units above threshold). The low contrast portion Of the curve was generally monotonic for both components 17 Figure 3.--Area-threshold curve for typical on-center cell plotted on log-log coordinates. The SIOpe for a least squares straight line is -1.45. The lepe predicted by Ricco's law (area x intensity = constant) is -2. The slope for Piper's law (Varea x intensity = -1. constant) 15 Background intensity = 0 log candles/m2 log Threshold .l 18 J .2 .3 .6 1.0 2.0 4.0 Spot Diameter (Degrees) Figure 3 l9 ’ - e Figure 4.--Transient and steady state stimulus respons on functions for an on-center cell. The data are plotted ' - ' ordi- semilog coordinates in the lower graph, on linear co nates for upper left graph (for low contrast levels only); on log-log coordinates for upper right graph. Values on the abscissa are expressed in threshold units. Bars around the points in the lower graph are 99 per cent confidence intervals. Spot size = 0.8°, in right eye receptive field with 1.2° center, located at a position 16° nasal and 7° superior to the area centralis. Back- ground intensity = 0 log candles/m2 esponse .ottei ;: T C0333.“ '20 420} .2 'g _ 3 2.6 - 3300- 3 . o \ on - a 2.2 - s .3 J. '80 i —-= .a f5 3% ‘3") 60 l L L L _1 4L3 [’lleJJJILLLLk 20 60 100 I40 .3 1.2 2.1 3.0 3.9 Intensity log Intensity 400 F ‘5 H transient C " Hsteady state 8 300 — d) m .. w a, 200 - -.-“ .. O. W 100 — i I L I I I L I l I I J L I .3 .9 1.5 2.] 2.7 3.3 3.9 Log Intensity Figure 4 21 with a mean length of 2.1 log units (S.D. = .67) for the transient phase and 2.3 log units (S.D. = .79) for the steady state phase. In most cases non-monotonic decreas- ing relationships were found for the high contrast portion of the curve. Three cells had functions for the transient component which came to an asymptote at about 2.1 log units above threshold; two other cells showed a Similar effect for the steady state component. The peak firing rate of the transient component was usually about 3 times greater than the peak firing rate for the steady state component. The slopes (on a semilog plot) of the rising portion of the curves were analyzed and found, in general, to be about four times greater for the transient component. Table l, which is Shown in a latter part of the Results section, gives a more detailed account of the Slope analysis. Spot size variation had the effect of changing the threshold intensity without significantly altering the shapes of the suprathreshold stimulus-response func- This was found to be true for both the transient tions. component curves (Figure 5) and the steady state component curves (Figure 6). Following the pattern of the Spots shown in the two figures, the stimulus-response functions for most spots showed the previously mentioned monotonic increas- ing portion for low contrast stimuli and the non-monotonic 'h T"_I.-._-— ___ll 22 I o o S Figure 5.--TranSient component stimulus-response funCtMMI for an on-center cell. Numbers on the abscissa are neu- tral density filter values (higher densities represent lower intensities).' The first point, i.e., the one closest to the vertical axis, on each function corresponds to threshold intensity. Right eye unit whose receptive field was located 10° nasal and 4° superior to the area centralis; receptive field center size = 3.2°; background light = 1 log candle/m2 23 L2 .6 0 l. 3.6 3.0 2.4 _. .w . A .s. r. n I . I A‘ . . . I my nu nu nu Av IL DAY/.3 Qua‘nunu o J i _ — — — _ — _ _ b _ _ .1 0 0 O O 0 8 O 4 w M w. M 8 éuomkaiamv 20¢ DE...”— F 51...: IUAL.s- re 3588: IE mm: area I log Intensity Figure 5 24 Figure 6.--Steady state component stimulus-response functions for an on-center cell. Numbers on the abscissa are neutral density filter values. The first point on each function corresponds to threshold intensity. Right eye unit Whose receptive field was located 11° nasal and o 0 0 3 inferior to area centralis; receptive field center Size = 4.5°, 25 flit. It: will Iliihlll . C O C I I O O O 0 JJ 7.2 4300 ihi — F — h P h n — b n I- sl. 1! .ll 1' pecuomwovzam 1.2 1.8 3.6 3.0 2.4 Log Intensity Figure 6 26 decreasing portion for the high contrast stimuli, with a change in tonicity occurring at about 2.2 threshold units. Although the peak firing rates for various spot sizes differed by as much as 70 Spikes per second for the tran- sient component curves and 30 spikes per second for the steady state component curves, the differences did not follow any systematic pattern. For some cells the small spots showed the highest firing rate; for others the larger spots showed higher peak responses. In most cases the differences in peak firing rates were minimal. An F test for homogeneity of regression2 (Winer 1962) was per- formed for the monotonic, i.e., low contrast portion, for the transient component curves (8 cells) and for the steady state curves (8 cells) of 16 on—center cells. In all 16 cases non-significant F values (p < .10) were obtained, thus indicating that the slopes of the least squares straight lines could not be distinguished statistically. Systematic differences in the amount of high contrast de- cline were also difficult to find. Thus, in general, statements made regarding the differences between the transient and steady state functions, i.e., slape, peak firing rate, and point of tonicity change, can be made 2This is typically used in an analysis of covari- ance to test the null hypothesis that the population co- efficients of linear regression are equal. Thus, Signifi- cant F_va1ues would indicate differences in the slopes of the least squares lines. 27 for any sized Spot (used in this study) which primarily activated the central region of the receptive field. Although their effects were not systematically examined, spots which were large relative to the size of the receptive field center, e.g., a 7° spot covering a receptive field with a 2° center, resulted in irregular stimulus—response relationships with very limited ranges of monotonicity. The slopes of the low contrast portion of such curves were usually more gradual than those found for smaller spots. The average response histograms in Figure 7 dis- close other differences in the effects of low and high contrast stimuli. For low contrast levels the histograms were Shaped like the ones first reported by Rodieck and Stone (1965). The major effect of increasing luminance was to increase the magnitude of the on response. The time constants of decay for the transient component, which remained relatively stable for low contrast stimuli, were usually between 35 and 45 msec. At the termination of the stimulus the firing rate dropped to zero before re- turning to the maintained level. The histograms for high contrast stimuli differed markedly from those for low contrast stimuli. The magni- tude of the on response was found to decrease, rather than increase, with increases in luminance. The decay from the peak of the transient component became 28 ' -center Figure 7.--Average response histograms for an on cell. Histograms have been traced from oscilloscope " . IAOTO photographs taken from the "y aXlS output of the C (bin width = 5 msec.). Smoothing has concealed small changes--less than 5 per cent-—in firing rate. Values on the "z" axis are in log units above threshold; the ' h pulse length is 920 msec. For lurninance levels up throng 2.1 threshold units the firing rate drops to zero at the termination of the stimulus pulse; it does not droP to zero for levels above 2.1 threshold units. Peak firing rates, during transient phase are: .9 - 397, 1.5 - 596, 2.1 - 710, 2.7 - 653, 3.3 - 590, 3.9 - 482. Spot size = 0.6° for right eye unit with receptive field center Slze of 1.2°, located 5° nasal and 12° superior to area centralis. Spikes/Sec. 720- Q 0 O L 29 o . Figure 7 30 progressively more accelerated with time constants of de- cay reaching as low as 10 msec. This rapid decay was followed by a second rise in response magnitude.3 The maximum firing rate during this phase, which was as high as 250 spikes per second, occurred 35-55 msec. after stimulus onset. Its magnitude decreased in a non-monotonic fashion with increases in luminance. Although the firing rate decreased immediately following the termination of the stimulus, it rarely dropped to zero. Off responses were usually present at high contrast levels with their firing rates reaching as high as those during the steady state component of the on response. The strength of the off response increased non-monotonically with increases in luminance. Latency changes also accompanied increases in intensity. For a stimulus .3 log units above threshold, the highest firing rate, i.e., the peak of the transient component, occurred about 50 msec. after the onset of the stimulus. AS intensity was increased the latency of the response became progressively shorter until, at 3The term "transient component" has been used in this paper to refer to the portion of the response in which the rate of change of firing rate is high. In a similar way the term “steady state component" has been used to refer to the phase where the rate of change of firing approaches zero. The second rise in firing rate observed for high contrast stimuli, therefore, was con- sidered as a part of the transient component and was included in the measurements of its magnitude. firm—flu“, ’ c—A‘. .v-v 1o. “ ' A ‘ swam-J ‘L-‘PE... ‘ fi'.—__'_AL . 31 about 2.4 threshold units, the peak firing rate occurred at about 5 msec. after stimulus onset. The latency re- mained at this level for high contrast stimuli. Figure 8 shows this effect in a more detailed form. Off-Center Cells.--Figure 9 reveals that there were both similarities and differences between stimulus— response relationships for on-center and off-center cells. As in the case of on-center cells, the stimulus-response functions could be divided into low and high contrast segments. The low contrast component was usually mono- tonic increasing with a mean length of 2.2 threshold units (S.D. = .61) and looked very similar to the corres- ponding portion for on-center cells. High contrast stimuli were found to be more suppressive for off—center cells than for on-center cells. It was not uncommon to observe firing rates for this intensity range which were as low as the maintained firing level. None of the on- center cells studied showed this amount of decline in response magnitude. The variability of the response mag- nitude (as measured by confidence intervals) for high contrast stimuli was found to be greater for off-center cells than for on-center cells. Manipulation of spot size had minimal effects upon the shapes of the stimulus-response functions of off-center cells (Figure 10). Marked high contrast sup- pression was observed for most spots which primarily 32 Figure 8.--Latency changes for an on-center cell. Num- bers on the abscissa are expressed as log units above threshold. Ordinate values refer to the latency, in msec., of the peak of the transient component. Spot Size = l.8° for left eye unit with receptive field cen- ter of 2.2°, located 6° temporal and 11° inferior to the area centralis. Background intensity = 0 109 candles/m2. 55 A 45 U 0 bye 0’ . E 35 m \—v at >~ ' U 1%. ‘: :25; a t O -J .—l U! 33 1 l l J l .L, l, L 1.5 2.1 2.7 3.3 3.9 Log Intensity Figure 8 34 Figure 9.--Stimulus-response function for transient component of an off-center cell. Numbers on abscissa are in log units above threshold. Bars around points are 99 per cent confidence intervals. l.0° Spot size = , for right eye unit with receptive field center size of 2.5°, located 5° nasal and 2° superior to the area centralis. Background intensity = 0 109 candles/m2. 35 -9 3 J -3 - 3 7V. lot 2..“ - n -.l..m 2n 39m - Lu -9 -3 L_.__-.._..____ s s s n s u 2 5 4 3 2 1 «deflmoiamv 20¢ act: nnnnnn lent sciss )0 r 36 Figure 10.--Stimu1us-response functions for transient component of an off-center cell. Numbers on the abscissa are neutral density filter values. Reading from left u: right, the first point on each function represents threshold intensity. Right eye unit with receptive field center of 3.5°, located 1° nasal and 14° inferior to the area centralis. Background intensity = l 109 candle/m2 37 1... — _ _ . _ O 0 0 w 9 3 “beauomFevzamv 20¢ 05:“. art Es MR: 5 m En m .1 .7 3.7 3.1 2.5 1.9 1.3 Intensity Log Figure 10 38 activated the central region of the receptive field, i.e., that were not large relative to the Size of the receptive field center. An F test for homogeneity of regression for the monotonic portion of the transient curves was made for eight off-center cells. The non-significant §_values, (p < .10) found in all eight cases indicate, as in the case of on-center cells, that the shapes of the rising portions of the curves were not, from a statistical stand— point, discriminable. In general, the conclusions made for on-center cells regarding the effect of Spot size variation could also be made for off-center cells. The average response histograms shown in Figure 11 illustrate the effect of luminance variation upon the time course of the response of a typical off-center cell. For low contrast levels the shape of the off response re- mained relatively constant for the various luminance levels. As was the case for the on response for on-center cells, the magnitude of the off response increased in proportion to luminance. Many of the high contrast ef- fects that were found for on-center cells were also ob- served for off-center cells: the peak of the transient component showed an accelerated decay; a second rise in firing rate occurred; and the overall magnitude of the ' ' ' ' . Althou h reSponse decreased With increases in luminance g off responses were quite common for on-center cells, on responses (which would be analogous) were rarely observed 39 Figure ll.--Average response histograms for an off-cenunr cell. Histograms were traced from oscilloscope photo- graphs. Smoothing has concealed small (less than 5 per cent) changes in firing rate. Numbers on "2" axis are log units above threshold. Peak firing rates are: .9 - 240, 1.5 - 408, 2.1 - 600, 2.7 - 432, 3.3 - 288, 3.9 - 192. On responses were not observed for this units SP0t size = 1° for left eye unit whose receptive field center was 2° in diameter, located 5° temporal, and 2° inferior to the area centralis. Background intensity = l 109 candle/m2. 40 d 3J9 Sp i kos/So con 0) O i R On Figure 11 41 for off-center cells. Decreases in latency were small—- usually less than 20 msec--in comparison to on-center cells. Best Fit Analysis Using a least squares criterion, logarithmic (R = a loglo 8), power (R = bSn) and linear (R = cS) functions were fitted to the low contrast data4 for ten on—center and ten off-center cells. A medium sized spot, e.g., a 1° spot in a field whose center was 2° in diam- eter, was examined for each of the twenty cells. Stimulus magnitudes were expressed in threshold units; response magnitudes as the difference between observed and main- tained (spontaneous) firing rate. A linear regression model was employed in the analysis of each of the three functions. A straight line was fitted to the data plotted on linear coordinates to test the accuracy of the linear prediction rule; to test the log function it was fitted to the data plotted on semilog coordinates; and for the power function, to the log-log data. Employing the theory of linear regression enabled the use of the Pearson product-moment correlation coefficient as a measure of 4From a statistical standpoint a linear regres- sion model may not be applicable to the high contrast data because the curves were generally non-monotinic, i.e., there may be non-linear components. 42 the goodness of fit. A "t" test for zero linear regres- sion (Walker and Lev 1953) was also made. As can be seen from Table l, the values of the Pearson product-moment correlation coefficients were usually lower for the linear function than for either of the negatively accelerated functions. Of the 20 cells studied, only one showed a statistically significant value of £_for a linear function whereas, highly Sig- nificant correlation coefficients were found in all but two cases for the log and power relations. The mean r values for the 20 cells were above .90 for both log and power relationships, with individual cells often showing values as high as .98. Although these data indicate that the shapes of the transient and steady state com- ponent curves were very similar, it also shows that their slapes differed markedly. On the average, the slope, for the semilog data, was about four times greater for the transient component curves. This difference in rate of rise is also reflected in the difference in exponents for the power function. 43 “II -..14H..fltl. oI-l_. - _i r. . .tizi.L be. we. mm mm. was mm. mm. om. cs. as. name mm. as. we ism. Hma .mm. mm. «am. am. sum. OH Hm. mm. wv «om. mmH smm. Hm. smm. Hm. *mm. m as. an. om em. ems .nm. em. sma. mm. sew. m mm. on. we «om. ANN «smm. vm. seem. ow. «ewm. h on. Hm. mv sHm. hmH sehm. om. tom. mw. «smm. 0 Nb. om. mm «ewm. mmH «swm. mm. emm. 5v. emm. m Hm. oh. 5v «emm. omH smm. em. *mw. Nm. seem. v m5.. mm. mm «Hm. OHN ram. ow. *«mm. Hm. smm. m mh. Nb. mv *«mm. mom ssmm. om. «mm. mm. sum. N am. mm. mm «rem. HmH seam. Hm. *evm. Nv. «sum. H H n macaw H omon H c H a H -II chasm ucofimcmue oumnm ucoflmcmua oumuwII :0 -II ...... ...... ...... . ...... A.mnq.. moum m oases bum ammum -II has: moaned oessuwnmmoq mason I'll I'll .msoauoqsm Hmocfla can if aflenuwummoa .noBom now mammamcw new umon mo an IIIIIIIIIII ml]. 0 H w-Hnma 44 Hoo. v Asia - ss Ho. v Anya I s .uoam moHHEmm Mom macaw mumsmm ammoa I macaw .ucoeoewmooo cowumHoHuoo ucoEOEIuUSUonm somnmmm I n .coHnUSSM Hmsom now neocomxm I G -- ms. -- -- has .mm. II- In: mm. om. some -- be” -- -- mom smmn -- -- am. sum. as III «mm III III mma «mm III III em. smm. m -- as. -- -- ems .mm. -- -- mm. som. m III mm. III III ehH ssmm. III III «m. ssmm. 5 III me. III III eHN ssem. III III em. sshm. e -- as. -- -- was «om. -- -- mm. 4mm. m -- mm. uu- -- ems sea. -- -- om. saw. S III Hm. III III va ssmm. III III He. em. III He. III III eom ssem. III Ill be. sHm. m -- as. In: -- omm yam. In: In: oe. . m ssmm H H H omon H emoam n .Im H a H III oumum ncmwmcnns aumnm ucoflmcmne ouwnm S III . museum . museum momoum n oflmcmne Anoucoo II- .II mmov moum m osmoH sum swans use: Hmoaflq oesnnflnmmoq nosom A All// Avoscflucouvr I. 0H wHQMB DISCUSSION High Contrast Effects Several investigators (Brown and Wiesel 1959, Cleland and Enroth-Cugell 1968, Steinberg 1969, Stone and Fabian 1968) have reported, in a semiquantitative manner, that high contrast stimuli have suppressive ef- fects upon the activity of cat retinal ganglion cells. The results of the present study are consistent with these reports but also Show: (1) that high contrast suppression occurs for both the transient and steady state responses of on-center cells; (2) that the sup- pressive effect is more pronounced for off-center cells than on-center cells; (3) that it occurs for a wide range of spot sizes. Spatial Summation at Threshold Area-threshold relationships for cat retinal ganglion cells have been studied by Barlow et a1. (1957) and Wiesel (1960) but most extensively by Cleland and Enroth-Cugell (1968). In the Cleland and Enroth-Cugell study, area-threshold curves for on—center cells were determined with sinusoidally modulated stimuli. Thres- holds were found to first decrease, then level off, as 45 46 spot size was increased. The change in Slope occurred as spot diameter reached that of the receptive field center. Although some of the cells encountered in the present study showed slight increases in threshold at the edge of the receptive field center, the shapes of the functions, were, for the most part, like those reported by Cleland and Enroth-Cugell. Barlow gt_§l. (1957) and Wiesel (1960), who both used square wave stimuli, reported that many of the cells they studied showed "U" shaped functions, where thresholds were found to rise sharply at the border be- tween the center and the surround of the receptive field. These results are difficult to reconcile with those of the present study. Cleland and Enroth-Cugell also determined the best fitting straight line for their area-sensitivity (reciprocal threshold) data plotted on log-log coordi— nates. A line with a slope of two, which is the slope predicted by Ricco's law (area x intensity = constant) for data graphed in this manner, (The slope for a graph of the data of the present study would have a slope of about 1.45 if displayed in this way; Piper's law would predict a slope of +1.) was found to be the best linear prediction rule for the increasing portion of the func— tion, i.e., for spots which did not invade the surround. The rate at which threshold fell with increasing Spot size was found, in the present study, to be more gradual 47 than this. The slape of the best fitting straight line was found to lie between the ones predicted by Ricco's and Piper's (/EEEE x intensity = constant) law. Perhaps the difference in the wave shape of the stimuli used in the two studies, i.e., Sine wave versus square wave, ac- counts for the disparity in the results. Unfortunately, neither Barlow §E_gl, (1957) nor Wiesel (1960) analyzed their data in this manner. Spatial Summation for Supra- threShOIH'StImuli In examining the stimulus—response functions for a variety of spot sizes, one cannot avoid noticing the similarity of their shapes. The likeness of the func- tions becomes even more apparent if the stimulus magni- tudes are described in threshold units. All of the functions, for a given cell, change from an increasing to a decreasing portion at about 2.2 threshold units; this change usually occurs at about the same response magnitude; and the suppression of the response that occurs for subsequent, high contrast, luminance levels is of the same order of magnitude for all of the spots. The negative results of the homogeneity of regression analy- sis suggest that the slopes of the rising phase of the functions cannot, from a statistical standpoint, be 48 distinguished.5 This latter finding points out an even more important aspect of the suprathreshold data. It suggests, in the form of a testable hypothesis, that cat retinal ganglion cells Show linear spatial summation for low contrast levels. It is also reasonable to hypothe- size that the magnitude of the suprathreshold summation is equal to the summation found at threshold. An example will serve to illustrate this point. Suppose that an experiment were conducted on a cell in which the rising portions of the stimulus-response functions for a number of spots (which activated only the central region of the receptive field) had exactly the same lepe. Suppose, also, that in the experiment on this cell, a suprathres- hold response criterion, of say 150 spikes/second were chosen and the intensity of various Spots was varied until this firing rate was reached for each spot. If the same procedure was repeated for a number of response criteria, according to the hypotheses, one should find functional relationships between spot size and intensity (for a con- stant response) which are essentially the same. This function should be similar to the one found for threshold 5It cannot be stated that the population slapes are equal since this would require accepting the null hypothesis-—an inexcusable act. The argument is still statistically well founded, however, because non-signifi- cant F values were obtained for 24 cells, i.e., the prob- ability of making a type I error decreases when results are repeatable. he _- -.7 r—siy. hr- 49 stimuli which activated only the central region of the receptive field, i.e., increasing spot size should re- sult in a decrease in the intensity necessary to reach the criterion response. The slope of the line depicting these data should have a value of approximately -1.45 for all response criteria associated with low contrast stim- uli. (See Figure 12) Intensity Code Perhaps the most salient feature of the stimulus- response functions is the difference in the Shape of their high and low contrast portions. These differences are very significant from the standpoint of cells in higher visual centers, like the lateral geniculate nu- cleus,.which receive information from retinal ganglion cells. When intensity is increased in the low contrast range, the picture presented by the retinal ganglion cell is quite clear: an increase in luminance, within limits, is signalled by an increased firing rate. The picture given in the high contrast range, on the other hand, is quite different. Sometimes an increase in firing rate results from an increase in luminance; at other times a decrease in firing rate occurs. The amount of change in reSponse magnitude per incremental increase in intensity, is at least for on-center cells, much smaller for high contrast levels than for low ones. Although the increments 50 Figure 12.--Theoretical area-constant response curves plotted on log-log coordinates. Abscissa and ordinate units are arbitrary but logarithmically spaced. The higher curves represent higher response criteria. All curves have slopes equal to -1.45. ,. . LLLLLL Log Intensity n c.) r l // ; p l .5 1.0 1.5 Log Spot Diameter Figure 12 52 in firing rate are~about the same for both ranges in the case of off-center cells, the variability in response magnitude is much greater for the high contrast stimuli. Apparently, cat retinal ganglion cells are designed to give a more precise representation of changes in lumi- nance for the low contrast range. From a functional standpoint this is not an unexpected finding, since lumi- nance differences greater than two log units are rarely encountered in the cat's normal environment. One would exPect, therefore, that these cells would have a more precise intensity code in the contrast range in which they normally function. Several investigators (DeValois 1965, Easter 1968, Stone and Fabian 1968) have attempted to determine the mathematical relationship which gives the best ap- proximation of stimulus-reSponse functions (over monotonic ranges) found for single cells of vertebrate visual systems. Using diffuse achromatic stimuli DeValois (1965) determined stimulus-response relationships for rhesus monkey lateral geniculate (broad band) cells. He reported that monotonic relationships could only be found over one log unit intensity ranges. Although it was concluded that either a logarithmic or power function adequately depicted the data for this range, this was not substanti- ated with statistical tests. 53 Perhaps the most comprehensive study related to this question was conducted on retinal ganglion cells in the goldfish retina (Easter 1968). Easter varied the in— tensity, over restricted ranges, of brief (10 msec.) monochromatic spots (620 nm) presented in the central region of "red, on—center" cells. His statistical analysis unequivocally demonstrated a non—linear intensity code. A power function with an exponent of .5 was found to be superior to other negatively accelerated functions that were examined. Stone and Fabian (1968) pooled low contrast stimulus-response data from ten cat retinal ganglion cells and examined the goodness of fit of logarithmic, power, and linear functions. Only the peak firing rates of the transient component were analyzed and both stimu- lus and reSponse values were measured as percentages of the maximum values encountered. The Pearson product- moment correlation coefficients they reported for the logarithmic, power and linear function were .94, .86, and .80, respectively. They concluded that stimulus- response relationships were non-linear. The results of the statistical analysis in the present study also indicate a non-linear intensity code for cat retinal ganglion cells. Correlation coefficients for a linear function, which were usually lower than .80, fell below the .01 significance level for 19 of the 20 54 cells studied. On the average, only about 60 per cent of the variance could be accounted for by this type of relationship. In marked contrast to this, both of the negatively accelerated functions that were examined showed highly significant Pearson product-moment correla— tion coefficients. On the average about 80 per cent of the variance could be accounted for by employing either a logarithmic or power function. Analyzing the transient and steady-state com- ponents separately revealed another important point about intensity coding mechanisms. Although the rising portions of both have essentially the same shape, they differ markedly in Slope. The slape of the least squares straight line fitted to the semilog data, on the average, was found to be four times greater for the transient component functions. The firing rate at the peak of the functions was usually three to four times greater for the transient component functions and for brief, 5 msec. intervals, as much as five times greater. In general, it took a much larger increment in intensity to produce a reliable change in firing rate for the steady state component than for the transient component. From a functional standpoint, then, the most information about intensity would be gained by the cat when there is a great deal of movement, either of its eyes or of the visual target. In this way illumina- tion would be constantly entering the receptive fields of . ‘.. __.—_._..4_.‘ —__I.‘\‘~i-'g rum-:Jm'_a ! 55 on-center cells and leaving the receptive fields of off- center cells, thereby constantly causing transient bursts of neural activity. Mechanisms According to Rodieck and Stone (1965), cat-reti- nal ganglion cell activity is controlled by excitatory and inhibitory processes. The relative contribution of these two influences is dependent upon the portion of 19;". l ‘ the receptive field activated and the type of cell, i.e., on-center or off-center. For on-center cells, the excita- tory process is more influential when luminous Spots are- presented in the central region of the receptive field. For off-center cells, inhibitory effects predominate under these conditions. The process which was found to be most influential for spots placed in the center of the recep- tive field was referred to by Rodieck and Stone as the "center mechanism." In a similar way, the less-influential process was referred to as the "surround mechanism." Thus for on-center cells, the excitatory process was called the center mechanism and the inhibitory process was re— ferred to as the surround mechanism; the terms are re- versed for off-center cells. The onset of the surround mechanism, according to Rodieck and Stone (1965), occurs about 50 msec. after the onset of the center mechanism when low contrast stimuli are used. (See Figure 13) _ _ .._._—.‘.——v‘n—(:.P’ t. - . 1i._-,-n --};J——v "Wei fin.‘m'- ! 56 Figure l3.--Receptive field model (reproduced from Rodieck and Stone, 1965) . The dome shaped curves at the top de- pict the strength of the center and surround mechanisms at different positions in the receptive field. Also shown is the time course of the center and surround mechanisms for three positions in the receptive field of an on-center cell. For the position closest to the center of the field the excitatory process predominates. The strength of the center and surround mechanisms are approximately equal at the second position. If their time of onset were the same, no response would be observed. If the surround mechanism is delayed by about 50 msec., however- on-off responses occur--the type of responses actually observed in their experiment. The surround mechanism predominates in the third position. 5:”; RE'I‘I NAI- RECEP'l‘lVE FIELDS > c ofi-eé‘éfiém ' L \E \ / L enter-mechanism l surround-mechanism surround component c N I I I l 1 l l I l l 1 V F W b ‘”"‘ ii A k % f V I..- _ I _____ center type no response on off type surround type response response reSponse Figure 13 58 It seems reasonable to assume, in accordance with the Rodieck and Stone model, that increasing luminance over the low contrast range had the effect of increasing the magnitude of both the excitatory and inhibitory in- fluences upon the ganglion cell, with the inhibitory ef- fects predominating for off-center cells and the excitatory effects predominating for on-center cells. Presumably, the slope of the transient component curves are much greater than the steady state component curves because the onset of the surround mechanism occurs about 50 msec. after the onset of the center mechanism. That is to say, the steady state component represents the sum of both ex- citatory and inhibitory influences whereas the transient component, for the most part, reflects excitatory inputs for on-center cells and inhibitory inputs for off-center cells. As the intensity enters the high contrast range the data suggest that two things occur: (1) the surround mechanism's relative strength, i.e., relative to the center mechanism, increases; (2) the temporal relationship between the onset of the center and surround mechanism changes; their time of onset jet about the same for.high contrast stimuli. The rationale for the necessity of BEER.°f these conditions is as follows. If condition (1) were true and condition (2) were false, the steady~state component would have shown high contrast suppression and the tranr sient component would have kept rising in strength. If 59 condition (2) were true but condition (1) false, the transient component would have shown high contrast sup- pression and the steady state component would not have. Although transient and steady state components could not be compared for off-center cells, there is no reason to argue that these assumptions do not apply to them also. One can account for the major difference between the ef- fect of luminance upon the two cell types, i.e., that there was greater high contrast suppression for off-center cells, by making the additional assumption that the in- crease in the relative strength of the surround mechanism is greater for off-center cells. (See Figure 14) One might ask what anatomical structures and neurophysiological events in the retina underlie the mechanisms that have been prOposed. A receptive field model presented by Rodieck (1967a) gives a reasonable answer to this question. According to this model recep- tor activation in any part of the receptive field has the immediate effect of increasing bipolar cell activity. It is presumed that bipolar cells form both presynaptic and/postsynaptic connections with amacrine cells; they excite amacrine cells, but, in turn, are inhibited by these same cells, thus providing a means of lateral in- hibition. Amacrine cells and bipolars have apposite ef- fects on ganglion cells. The effect depends upon the cell type: for on-center cells, bipolar cells are 60 Figure l4.--Schematic of mechanisms proposed to account for observed stimulus-response functions. For low con- trast stimuli the center mechanism predominates and leads the surround mechanism by about 50 msec. The relative strength of the surround mechanism, especially for off-center cells, increases at high contrast levels. The latency of both mechanisms decreases but the latency differential between them disappears. ‘ I‘M- Vb“ 61 ON CENTE R Center I\ Mechanism LOW CONTRAST Surround Mechanism I on stimulus Center Mechanism HIGH CONTRAST Surround l/ Mechanism Figure 14 OFF CENTER N— off if L K‘ 62 excitatory and amacrine cells are inhibitory; for off- center cells, amacrine cells are excitatory and bipolar cells are inhibitory. When illumination causes activa- tion of both bipolar cells and amacrine cells, the effect of bipolar cells predominates. (See Figure 15) Let us now consider the neurological events that might occur under the conditions of the present experi- ment. Presenting a luminous spot in the center of the receptive field would lead to receptor activation, then to increased activity in the bipolar cells. Bipolar cells would excite amacrine cells, which in turn, would inhibit bipolars. For on-center cells, the amacrine cells would inhibit and the bipolar cells would excite, the ganglion cells to which they both make connections. The opposite would occur for off—center cells. For on- center cells the excitatory influences would predominate for all luminance levels and for off-center cells the inhibitory influence would predominate. One could ex- plain the difference in the Slope of the transient and Steady state components curves by simply assuming that the amacrine cell effects are delayed by about 50 msec. at low contrast levels. High contrast suppression could be explained by presuming that amacrine cell effects are relatively greater for high contrast stimuli. One would also have to presume that these effects are no longer de— layed and that they are greater in magnitude for off-center cells than for on-center cells. 63 Figure 15.--Rodieck (1967) receptive field model for an on-center cell. Activity of bipolars can be either in- creased by receptor activation or decreased by amacrine cells acting as inhibitory interneurons. Bipolars are excitatory and amacrine cells inhibitory to ganglion cells. The opposite is true for off-center cells (re- produced from Johnson, Hatton and Goy, 1969). IHIH))II\))IIII IIIIIIIIIIIIIII for s a:- acrre RODS I sax ¢vexcnatory '0“ l lbifory 9' BIPOLARS AMACRINES GANGLION CELLS Figure 15 BIBLIOGRAPHY ['1 A‘A—"A :_ BIBLIOGRAPHY Barlow, H. B., Fitzhugh, R. and Kuffler, S. W. Change of organization in the receptive fields of the cat's retina during dark adaptation. Journal of Physio- logy, 1957, 137, 338-354. Bishop, P. 0. Central nervous system: afferent mecha- nisms and perception. Annual Review of Physio- logy, 1967, 29, 427-484. Bishop, P. 0., Kozak, W., Levick, W. R. and Vakkur, G. J. The determination of the projection of the visual field on the lateral geniculate nucleus in the cat. Journal of Physiology, 1962a, 163, 503-539. Bishop, P. 0., Kozak, W. and Vakkur, G. J. Some quanti- tative aspects of the cat's eyes: axis and plane of reference, visual coordinates and optics. Journal of Physiology, 1962b, 163, 466-502. Brown, K. T. and Wiesel, T. N. Intraretinal recording with micrOpipetee electrodes in the intact cat eye. Journal of Physiology, 1959, 149, 537-562. Cleland, B. G. and Enroth-Cugell, C. Quantitative aspects of sensitivity and summation in the cat retina. Journal of Physiology, 1968, 198, 17-38. DeValois, R. L. Behavioral and electrOphysiological studies of primate vision. In Contributions to Sensornyhysiology. Neff, W. L. (Ed.), Academic Press, New York 1965. Easter, S. S. Excitation in the goldfish retina: evi- dence for a non-linear intensity code. Journal of Physiology, 1968, 195, 253-271. Gouras, P. Identification of cone mechanisms in monkey ganglion cells. Journal of Physiology, 1968, 199, 533-547. Granit, R. Stimulus intensity in relation to excitation and pre and postsynaptic inhibition in isolated mammalian retina. Journal of Physiology, 1944, 103, 103-118. 65 Jrllull 66 Granit, R. Sensory Mechanisms in the Retina. Oxford Press, London 1947. ' Granit, R. Receptors and Sensory Perception. Yale Univ. Press, New Haven, 1955. Hubel, D. H. Single unit activity in the lateral genicu- late body and optic tract of unrestrained cats. Journal of Physiology, 1960, 150, 91-104. Hubel, D. H. and Wiesel, T. N. Integrative action in the cat's lateral geniculate nucleus. Journal of Physiology, 1961, 155, 385-398. Johnson, J. I., Hatton, G. I., and Gay, R. W. The phy- siological analysis of animal behavior. In The gehavior of Domestic Animals. E.S.E. Hafez TEE.) Bailliere, Tindall and Cassell, 1968, Ch. 7. Kuffler, S. W. Discharge patterns and functional organi- zation of mammalian retina. Journal of Neuro- physiology, 1953, 16, 37-68. Martin, A. R. and Branch, C. L. Spontaneous activity of Betz cells in cats with midbrain lesions. Journal of Neurophysiology, 1958, 21, 368-375. Mead, L. C. Visual brightness discrimination in the cat as a function of illumination. Journal of Gene- tic Psychology, 1942, 60, 223-231. Mello, N. M. and Peterson, N. J. Behavioral evidence for color discrimination in cats. Journal of Neurophysiology, 1964, 27, 327-333. Meyer, D. R. and Anderson, R. A. Colour discrimination in cats. In, Colour Vision: Physiology and Ex- pgrimental Psychology, A.V.S. DeReuch and JTI Knight (Eds.) Little, Brown and Co., Boston, 1965. Michael, C. R. Receptive fields of single Optic nerve fibers in a mammal with an all cone retina. III: opponent color units. Journal of Neurophysiology, 1968, 31, 268-282. Rodieck, R. W. Maintained activity of cat retinal gang- lion cells. Journal of Neurophysiology, 1967a, 30, 1043-107IT Rodieck, R. W. Receptive fields in the cat retina: a new type. Science, 1967b, 157, 90-92. . “1...... __Ai.._P 67 Rodieck, R. W. and Stone, J. Analysis of receptive fields of cat retinal ganglion cells. Journal of Neurgphysiology, 1965, 28, 833-849. Rodieck, R. W., Pettigrew, N. D., Bishop, P. O. and Nikara, T. Residual eye movements in receptive- field studies of paralyzed cats. Vision Research, 1967, 7, 107-110. Sechzer, J. A. and Brown, J. L. Color discrimination in the cat. Science, 1964, 144, 427-429. Spinelli, D. N. Visual receptive fields of the cat's retina: complications. Science, 1966, 152, 1768-1770. Steinberg, R. H. High intensity effects on Slow poten- tials and ganglion cell activity in the area centralis of cat retina. Vision Research, 1969, 9, 333-350. Stone, J. and Fabian, M. Specialized receptive fields in the cat's retina. Science, 1966, 152, 1277- 1279. Stone, J. and Fabian, M. Summing properties of the cat's retinal ganglion cell. Vision Research, 1968, 8, 1023-1040. Vakkur, G. J., Bishop, F. O. and Kozak, W. Visual optics in the cat, including posterior nodal distance and retinal landmarks. Vision Research, 1963, 3, 289-314. Walker, H. M. and Lev, J. Statistical Inference. Holt Rinehart and Winston, New York, 1953. Wiesel, T. N. Receptive fields of ganglion cells in the cat's retina. Journal of Physiology, 1960, 153, 583-594. Wiesel, T. N. and Hubel, D. H. Spatial and chromatic in- teractions in the lateral geniculate nucleus of the rhesus monkey. Journal of Neurophysiology, 1966, 29, 1115-1157. Winer, B. J. Statistical Principles in Experimental De- sign. McGraw-Hill, New York, 1962. APPENDIXES Lo- 'I.. 1' Fm» .- APPENDIX A TRACINGS OF THE ELECTRODE TRACKS As a rule the plane of the electrode track was found to be very different from the plane of the brain section. Thus the tracks had to be traced through a large number of sections in order to determine their starting and finishing points. The sections shown in Figures Al and A2 are the ones in which the lesions were found. Figure A3 shows a photomicrograph of the section where the lesion was found for experiment 68337. In general, recordings were only made from one puncture per experiment. One of the nine brains was lost. Abbreviations used in the figures are taken from A. L. Berman, The Brain Stem of the Cat; A Cytoarchitec- tonic Atlas with Stereotaxic Coordinates. University of Wisconsin Press, Madison 1968 and are as follows: AF - Ammon's formation F - fornix IC - inferior colliculus LGD - dorsal lateral geniculate nucleus LP - lateral complex of the thalamus PUL - pulvinar SC - superior colliculus OT - Optic tract VB - ventral basal complex of the thalamus 68 69 Figure Al.--Electrode track tracings for parasagittal sections from experiments 68339, 67328, 68344, and 68320. Section shown in each case is the one in which the lesion was found. Dotted lines represent electrode tracks and the stars indicate lesion sites. 70 60339 «344 l” D) «no 1 f‘ m Figure A1 6732. 71 Figure A2.--Electrode track tracings for parasagittal sections from experiments 68346, 68400, 68401, and 68337. Section shown in each case is the one in which the lesion was found. Dotted lines represent electrode tracks and the stars indicate lesion sites. 66666 72 Figure A2 w 73 F' -- ‘ igure A3. Photomicrograph of the section, in which the lesion was found for experiment 68337. 74 Figure A3 APPENDIX B EXPANDED VERSION OF THE METHOD SECTION .- -. Mi METHOD Subjects Single unit recordings were made from 81 (44 on- center cells and 37 off-center cells) optic tract fibers in nine normal, adult cats. Apparatus The cats faced a diffusely lit, (either 0 or 1 log candle/m2) 1-1/2 meter, translucent tangent screen (Pola Coat). The photic source, a modified 750 watt slide projector, was placed 130 cm behind the screen. The size of luminous Spots presented on the screen was controlled by placing metal slides in the projector. A metal neutral density filter holder and a variable speed_ mechanical shutter (Wollensak Pi alphax) were attached to the movable lens of the projector. (See Figure B1.) Recording System Glass insulated tungsten microelectrodes were used in a conventional capacitance coupled system (See Figure B2). After being amplified by a Tektronix 122 Preamplifier the biological signal was visually displayed on the upper beam of a Tektronix 502 A oscillosc0pe and 75 76 Figure Bl.--Photograph of the photic source. Intensity was controlled by placing neutral density filters in the holder (with2 slots) attached to the movable lens of the slide projector. The metal Slides provided a means of controlling spot Size. The alligator clips attached to the shutter are connected to two poles of a microswitch. The wires leading from the battery and the shutter go to the Tektronix 162 waveform generator. IMm ters i: amt mmf tacks: :rosut'i ter 4 77 I 9 «~17 -1 Figure B1 I-' 5 s21.“ 78 Figure BZ.--Block diagram of the recording system. the audio monitor and 502 A Scape No. 1 were used for moni- toring the unmodified signal from the brain. The 5021\ Scope No. 2 was used to compare the output of the peak detector with the low frequency output of the audio monitor. This insured that pulses going to channel 1 of the tape recorder and to the second audio amplifier represented the activity of the cell being studied. fflw 162 waveform generator triggered the 161 pulse generaUn? at stimulus onset. The pulse from the 161 pulse genera- tor was recorded on channel 2 of the tape recorder. 79 T! A": I _ I... [1'39“ .| .3303: saga f Sam—fies; flaw j .3 Eat; 95:33. oaoum < .mem $222.: a ’ 3:32-— SEE-fl & .o1_ooc¢ 0.61 :53... y :— ell lili.’ _aE_=< 80 passed to an audio monitor. The low frequency filter output of the audio monitor was connected to the upper beam of a second 502 A oscillosc0pe. After the biologi- cal signal was amplified here it was fed to Technical Instrument Corp. Model 607 peak detector. This instru- ment, which served as a Schmitt trigger, permitted the experimenter to isolate the spikes of a single unit from activity of lower amplitude. The three volt peak detector output was amplified through the lower channel of an oscilloscope (which was used as a visual monitor), whose output was connected to a second audio amplifier and to one channel of a tape recorder (Magnecord 1028). The signal heard on the loudspeaker connected to the second audio amplifier was used for mapping receptive fields. The onset of the stimulus was marked on a second channel of the tape by utilizing a microswitch which is present in the Wollensak Pi alphax shutter. When the shutter Opened it closed the microswitch which was linked in a 45 volt battery circuit. The output of this circuit was used to trigger a Tektronix 162 waveform generator. Its sawtoothwave output, in turn, activated a Tektronix 161 pulse generator. The pulse from the 161 generator was recorded on the second channel of the tape. Procedure Initially an intraperitoneal injection of atropine sulfate (.5 mg/kg.) was given to minimize secretions. 81 Fifteen minutes later the animal was lightly etherized to facilitate cannulation of the femoral vein. The venous cannulation served two purposes: (1) It was used during surg- ery for injection of Pento Short--a short acting barbituate (.67cc/kg); (2) In a later state it served as a route for continuous infusion of a muscle relaxant, Flaxedil (galla- mine triethiodide). Following the cannulation, a bilateral cervical sympathectomy and tracheotomy were performed. The sym- pathectomy was done in order to decrease eye movements during the eXperiment (Rodieck gt_al., 1967). A "Y" shaped glass tube was inserted in the trachea to permit artificial respiration. The procedure for mounting the animal's head in a head holder was at variance with the standard one. In- stead of inserting metal pins in the ears, bars were at- tached to the zygomatic arches with dental cement. Once the head was secured, a craniotomy was performed, the dura mater was removed from the exposed area, and Inogar II (2.5 grams to 50 ml of physiological saline) was placed over the Opening (See Figure B3). Upon completion of the surgery the animal was electrically decerebrated by a technique similar to the one described by Martin and Branch (1958). Using the tentorium bone as a landmark, an electrode with a 1 mm uninsulated tip was advanced to the midbrain reticular 82 Figure B3.--Photograph of cat in head holder. The hori- zontal pins of the head holder are cemented to the zygo- matic arches. The substance below the cat'e nose is dental cement which was used to prevent vertical movement of the head. The wire leading from the back of the neck is connected to earth ground through the Tektronix 122 preamplifier. The tubing in the lower right of the picture is attached to a "Y" shaped, glass tracheal can- nula and to a respirator pump (not shown in the picture). 83 Figure B3 84 formation at a medial-lateral position about 2 mm from the midline. A cathodal current of 500 milliamps was delivered for 20 seconds to effect the decerebration. This procedure was repeated for a position 1 mm lateral to the initial lesion.- Two lesions were also placed in the same relative position on the opposite side of the midline. In order to immobilize the eyes an intravenous injection of 80 mg. of Flaxedil was given. A mixture of this drug (28 mg/hour) and physiological saline (6 cc/ hour) was continuously infused through the femoral vein for the duration of the experiment. Respiration was maintained by a Harvard Apparatus respirator pump (stroke volume 45 cc; respiration rate 20/min.). Body temperature was held at 38° C by a heating blanket connected to a 12 volt D.C. circuit. Corneal contact lenses ranging from 37 to 42.25 diopters, were fitted to prevent corneal clouding. A 10 per cent solution of neosynephrine was used to keep the pupils dilated. Retinoscopy was performed in the usual mannerwith spherical and cylindrical lenses ap- propriately placed in front of the eyes to correct re- fractive errors. Upon completion of the preliminaries an electrode was advanced to the Optic tract. The procedure for as- certaining its location was relatively straightforward. 85 First, the location Of the lateral geniculate nucleus (LGN) was determined. Then, using it as a landmark, the optic tract could usually be found by advancing the elec- trode about 5 to 8 mm below the ventral surface Of its anterior portion. In terms Of evoked activity the optic tract could be distinguished from the LGN in two ways: (1) Its overall response (as heard over the audio monitor) to a moving flag is one Of a soft "S" sound, rather than the harsh swish found in LGN (Bishop gt_§l. 1962a); (2) single Optic tract fibers give stronger responses to dif— fuse illumination than lateral geniculate cells (Hubel and Wiesel 1961). Once the tract was located the electrode was allowed to remain stationary for 15 minutes. Then it was advanced in small increments until a fiber could be isolated. The fiber's receptive field was mapped from the rear Of the tangent screen with .2° dim, luminous spots presented on pieCes Of tracing paper taped to the screen. The papers were kept as permanent records of the receptive fields. Threshold Experiments.--Thresholds for various sized spots--placed in the most sensitive portion Of the receptive field-~were determined by varying their inten- sity in .1 log unit steps while listening tO changes in activity over the loudspeaker, i.e., the signal which had been filtered for low frequencies and fed through 86 the Schmitt triggercircuit. For 8 Of the 21 cells studied, thresholds were also determined by analyzing changes in maintained (Spontaneous) firing with a computer Of average transients (C.A.T.). The threshold values from the two techniques were never more than .l.log unit apart. Suprathreshold Experiments.--The luminance Of various spots, ranging from .1° to 7°, was varied in .3 log unit increments over a 4.2 log unit range. Because most diffusing surfaces are directional, luminance values have been expressed in threshold units, i.e., log units above threshold intensity; or in terms Of neutral density filter values (where smaller numbers represent greater luminance levels). The duration Of stimulus pulses was 920 msec.; their rise time was 5 msec. Each stimulus was repeated 15 times at a rate of .2 cycles/second and always presented in the most sensitive portion of the receptive field. Upon completion Of recording from a unit its receptive field location, with respect to the area centralis, was ascertained by the reversible Ophthal- mOSCOpic technique (Bishop et al. 1962b). In this tech- nique (See Figure B4) the rectangular beam Of an opthal- moscOpe is fixed on the edge Of the Optic disc while viewing the fundus. The Ophtahlmoscope is then revolved 180° and the rectangular beam is traced onto the tangent screen. This procedure is repeated for several points tangent to thecptic disk until a suitable map is Obtained. 87 Figure B4.--Schematic showing the reversible Ophthalmo- SCOpiC technique. The fundus (C) was viewed by the experimenter through an ophthalmoscope (B). The slit from the OphthalmOSCOpe was maneuvered until it was tangent to the Optic disc (D). The opthalmoscope was then rotated 180° and the slit was traced onto the 1-1/2 meter tangent screen. This procedure was re- peated until the projected Optic disc was adequately mapped. 88 omfi Ozao .\ x K \ - I \ \‘ C y x A o. a . . I I. . .v 1 cl. \ . \D .. . a u t. (A \ a A . . .1 . « .. . , .cl. ..0 r. . I .., .. . ‘ . A . . 1 . ...~ . t A I... ... ..,. .u‘ o..,, V .. .1.:A r . . . . ‘ .. oh I a. s} ' . .6. n . . . . a . I . . . . u... , I x .. . O. .; .. . u. a . . . . r . . .. ,. . . l , I \ . I. I . I , r . .u o o p £333 “55%. 2m 29. mt. 89 The position Of the receptive field relative to the disc is then measured. This measurement can be converted to one where the area centralis is the reference point by presuming that, on the average, the area centralis is located 13.6° temporal and 5.5° superior to the Optic disc (Vakkur gE_§l. 1963). Typically data from 5 to 15 fibers were collected in a single puncture. After completing work in a punc- ture, a microlesion was usually placed at the recording site. When the experiment was terminated the cat was intracardially perfused with .9 per cent saline, followed by a mixture of 10 per cent formalin and .9 percent sa- line. The brains were embeeded in celloidin and sectioned parasagitally. Alternate sections were stained with thionin for cell bodies and hematoxylin (Weil or Heiden- hain method) for myelinated fibers. Recording locations were confirmed by tracing electrode tracks and looking for the microlesions which had been placed at the record- ing site during the experiment. Data Analysis The system for analyzing data included the apara- tus shown diagrammatically in Figure BS. There were two major components Of the system: one that was concerned with processing the neural activity which was recorded 90 Figure B5.--Block diagram Of data analysis system, The audio monitor, 502 A Scope NO. l and the peak detector were concerned with processing the data from channel 1 of the tape recorder. The lOOp between the peak detec- tor and the 502 A Scope NO. 1 provided a means of moni- toring the input to the C.A.T. The amplitude discriminaflx and 502 A Scope NO. 2 were concerned with the processing of the stimulus marker, i.e., the pulse that correspondai to the stimulus onset, recorded on channel 2 of the tape recorder. This signal was used to trigger the analysis sweep of the C.A.T. The S-4 stimulator was used to con- trol its sweep speed. 91 mm musmflm 52?; 31.22 2.5 {3:2an ® .3».— P _ H ‘ 2:: 5.3.31— _ BEBE; ._.___._.__.4_ a I 32:55:: ~62 a cow w==:_._E<. 2.3m Add @ LL fl < Sm .r 7 ...2_=az. .62 2...; - <2: hostages each 0 O 92 on channel 1 of the tape recorder and one that dealt with processing the time marker signal that was recorded on channel 2. The spike activity from channel 1 was led through an audio monitor before being visually displayed on a 502 A oscilloscope. The vertical output Of the oscillo- scope was connected tO a peak detector which in turn fed square-wave signals to both the C.A.T. 400 B, where they served as the data for generating average response histo- grams, and back to the oscilloscope. The loop between the oscilloscope and the peak detector enabled the experi- menter to monitor the input to the C.A.T. After being amplified by a second oscilloscOpe the signal from channel 2 was fed to an amplitude dis- criminator. (In this context the amplitude discriminator is functionally equivalent to the peak detector.) The amplitude discriminator's output was connected to the C.A.T., in order tO trigger its analysis sweep. The analysis time (bin width size), which was usually 5 msec., was controlled by an S-4 stimulator. After being analyzed by the C.A.T. the data were usually printed out on a teletype. The blocks of printed numbers, which represented the sum (over 15 stimulus cycles) Of spikes during 5 msec. time intervals, were used as the basis for quantifying the magnitude Of the transient and steady state responses. 93 Average reSponse histograms could be visually dis- played and photographed by connecting the "Y" axis output of the C.A.T. to the vertical amplifier Of an oscilloscope. S E R A R m L V H s R E W N U E T A Tlllllz 5 4| WIHIHIHIHIII