of the retina is so small, in relation to that of the R-membrane, that no

Transcription

1 J. Physiol. (1965), 176, pp With 1 plate and 12 text-figures Printed in Great Britain SOME PROPERTES OF COMPONENTS OF THE CAT ELECTRORETNOGRAM REVEALED BY LOCAL RECORDNG UNDER OL BY G. B. ARDEN* AND K. T. BROWN From the Department of Physiology, University of California Medical Center, San Francisco 22, California, U.S.A. (Received 4 August 1964) When the local electroretinogram (l.e.r.g.) is recorded by an intraretinal micro-electrode, the electrode tip penetrates the retinal region from which the recorded response is generated. Changes in amplitude should therefore occur as the origin of a given component is passed, and changes in polarity should also occur if the component is generated as a radial dipole; such results provide information concerning the origin of any given component. Responses from distant retinal areas may also be recorded, however, between an intraretinal electrode and a reference electrode behind the eye (Brown & Wiesel, 1961a). These responses have no significance for analysing or localizing components, but they introduce undesirable complexities superimposed upon the l.e.r.g. The vitreous humour acts as a lowresistance shunt which permits current generated in distant regions to flow through the retina at the intra-retinal electrode (Brown & Wiesel, 1961 a). Thus the problem of obtaining strictly local responses is to prevent current from flowing through the vitreous humour or from being recorded. This problem has been solved in mammals by placing the reference electrode in the vitreous humour (Brown & Wiesel, 1961 a). The efficacy of the method has been explained on the assumption that the radial resistance of the retina is so small, in relation to that of the R-membrane, that no voltage drop could be recorded through the retina itself in response to activity generated at distant sites. t follows that the l.e.r.g. will also be attenuated, particularly when the micro-electrode is in the superficial layers of the retina. An alternative method was therefore desired for recording the mammalian l.e.r.g. n the experiments of this paper, the vitreous humour was replaced by a heavy oil. This abolished the low-resistance pathway for current flow to reach the electrode from distant retinal sites. L.e.r.g.'s could then be recorded from the retinal surface, or by an intraretinal micro-electrode, * Present address: nstitute of Ophthalmology, Judd Street, London, W.C. 1, England.

2 430 G. B. ARDEN AND K. T. BROWN regardless of the position of the reference electrode. Responses recorded by this method are larger than have been seen previously, and responses may be detected to weaker stimuli. Certain responses from surface layers of the retina have also been detected for the first time. This technique has been applied in the present paper to further problems concerning the analysis of e.r.g components and identification of the origins of these components. Granit and his collaborators showed that the e.r.g. consisted of at least three components designated P, Pll, and P (see Granit, 1947). P is the c-wave, which all investigators agree is generated by cells of the pigment epithelium. Granit's P has now been shown to consist of two separate components (Brown & Wiesel, 1961a, b). One of these is the b-wave; the other is an essentially square wave which is positive in polarity by conventional recording and negative deep in the inner nuclear layer. Because of its shape it has been designated the d.c. component. t is now well agreed that the b-wave arises from cells of the inner nuclear layer; the d.c. component also arises from cells of that layer (Brown & Wiesel, 1961b). Granit's P is negative by conventional recording methods, and its leading edge in his analysis gives rise to the a-wave. Brown & Watanabe (1962a, b) have shown that P originates from the receptors and have designated this component a receptor potential. t has been shown that this response has a minimum latency of about 1*5 msec, and a separate receptor response has now been described which has no detectable latency; hence these responses have been designated the early and late receptor potentials (Brown & Murakami, 1964). P is the late receptor potential in this terminology, but will be referred to in this paper simply as the receptor potential. Evidence is presented in this paper which further confirms the localizations of the receptor potential and b-wave; a new component of the e.r.g. is also described. METHODS General techniques. Adult cats weighing kg were anaesthetized with pentobarbitone (30 mg/kg, i.p.); the level of anaesthesia was maintained by subsequent injections of further small amounts, both i.p. and i.v. Succinylcholine was given by continuous intravenous infusion in amounts sufficient to immobilize the animal, which was artificially ventilated. Body temperature was maintained at normal levels by an electric heating pad, which was automatically regulated by a rectal thermistor probe acting through a control circuit. The animal's head was fixed in a head-holder and mounted so that the left pupil was at the centre of rotation of a special optical stimulator. The apparatus and techniques have been described in detail by Brown (1964), and the eye was prepared for intraretinal recording as described by Brown & Wiesel (1959). The pupil was dilated by local application of homatropine (2 %) and phenylephrine (10 %). After the eye-ring assembly had been placed in position, three needles were inserted which served as cannulas into the fundus of the eye. The new procedure of replacing the vitreous humour with oil was then carried out. Replacement of vitreous by oil. A transparent radio-opaque contrast medium was used

3 LOCAL ELECTRORETNOGRAM UNDER OL 431 (Pantopaque, U.S.P.). Since its specific gravity of 1-27 is much higher than that of vitreous, it might be thought that the Pantopaque would sink immediately on to the retina. n the cat this usually does not occur, since the vitreous is very viscous and contains organized structures. Usually about 0-1 ml. of oil was injected at a time through a slender needle inserted into the lowest of the three cannulas. Each injection caused a rise of intraocular pressure, and vitreous slowly issued from the upper two cannulas. When the flow stopped, another injection was made. Only about 1/3 of the vitreous could be thus replaced. Long needles were then inserted through the upper two cannulas into the vitreous and rotated rapidly. This procedure seemed to liquefy the vitreous, so that total replacement was possible in favourable cases. The experiments often failed, however, at this stage. The vitreous might fail to liquefy; also retinal bleeding might develop, owing to the needles catching on organized structures in the vitreous, and thus pulling on the retina. t is our impression that the vitreous is easier to replace in old cats and that the technique usually fails in young adults. n successful experiments the optical qualities of the eye were not significantly impaired. n such cases it was possible to see the sharp outline of the stimulating spot on the retina, and to observe the micro-electrode tip as it touched the retina. When a micro-electrode was introduced through the oil to the retina, impulse activity was always detected at the moment the electrode emerged from the oil; a slight additional advance resulted in penetration of the electrode into the retina. These results indicate that the oil was very closely apposed to the retinal surface. Stimulating apparatus and stimulus calibration. Light stimuli were delivered through a special optical stimulator (Brown, 1964) which provided stimulus and background beams. The stimuli were delivered in Maxwellian view, and the cat's eye was made emmetropic with spectacle lenses. A system of stops and diaphragms provided both disk and annular stimuli of various sizes. The stimulus could be placed on any part of the retina; owing to limitations on movement of the micro-electrode carrier, however, experiments were carried out in the superior nasal quadrant of the retina. All filters and wedges were calibrated in the stimulator with a Photovolt densitometer. The maximum output of the stimulator was found by focusing the light through a + 10 dioptre lens on to a freshly prepared MgO surface, and measuring its luminance with an S.E.. photometer. t was found that the luminance was 1-2 log ft--l, From this the effective retinal illumination of the cat eye was calculated. Corrections were applied for preretinal absorption (Ludvigh & McCarthy, 1938), the power of the cat's eye (Marriot, Morris & Pirenne, 1959), and tapetal reflectivity (Weale, 1953). The value obtained was 5-3 x 10-3 m/ mm2 of retina. t was also desired to calculate the quanta of equivalent wave-length 507 m,t in the stimulus. t was assumed that the colour temperature of the tungsten source approximated to that of Standard lluminant A, and that the cat's luminosity curve approximated to the C..E. scotopic function. Since this method of calculation is indirect, it was thought advisable to check it by measuring the dark adapted sensitivity of the human eye with the same stimulator. This was done for an observer (G. B.A.) whose dark adapted threshold had previously been determined with a calibrated adaptometer. The conditions of the measurement were: 40 min dark adaptation, uniocular viewing, peripheral stimulation, head fixed on a biting board, no fixation, 50 target, unlimited duration of exposure, and forced choice method. Using the calibration referred to above, the threshold was found to be reached at a retinal illumination of 4 x lm/mm2 (S.E. = 0-1 log unit). This is a commonly accepted average threshold value which agrees well with the observer's previous thresbold determination. Therefore it is considered that the present calibrations and calculations are accurate. Assuming that the optical density of visual purple in the rod was 0-15, it was then possible to calculate the stimuli in terms of equivalent quanta at 507 m/z absorbed by the receptors per sec (Marriot et al. 1959; Wald, Brown & Gibbons, 1963). The area of retina covered by the stimulus was calculated from the aperture size, focal length of the final lens of the stimulator, and the posterior nodal distance of the cat's eye.

4 432 G. B. ARDEN AND K. T. BROWN Recording technique. L.e.r.g.'s were recorded by tungsten micro-electrodes with tips of about 0.5gu diameter, or by 125/s silver wires clad in glass. After the latter had been introduced through the oil on to the retina, their impedance at 4 c/s was about 300 kq. The tungsten micro-electrodes had a higher impedance, which could approach 100 MQ2. The silver wires were theoretically preferable for surface records, but it was found that the biological noise was often so high that little difference was noted. The silver wires had diameters equal to that of the smaller stimulus spots, which caused shadowing, and retinal penetrations were desired in most experiments. Hence tungsten micro-electrodes were usually used, and electrode depth was controlled by a hydraulic device (Brown, 1964). The reference electrode was always placed in the mouth, and the input stage was a conventional cathode follower. Except where specifically mentioned, all records were obtained with directcoupled amplification. The upper band pass of the amplifier was limited to either 250 or 500 c/s, to reduce noise. n experiments with penetrating micro-electrodes, retinal electrode depth was determined by the physiological methods of electrode location which have been described (Brown & Wiesel, 1959) and validated by electrode marking studies (Brown & Tasaki, 1961). n all cases the end point of the depth scale was penetration of the R-membrane (Brindley, 1956), which has been shown to be in the complex of Bruch's membrane and pigment epithelium (Brown & Wiesel, 1959; Tomita, Murakami & Hashimoto, 1960; Brown & Tasaki, 1961; Brindley & Hamasaki, 1963). n some cases electrode depth was expressed as a percentage of the micrometer distance which the electrode traversed between the retinal surface and the R-membrane. This percentage scale is independent of the angle at which the electrode approaches the retina, and permits accurate correlation of electrode depth with retinal histology (Ogden & Brown, 1964). RESULTS Recording from the retinal surface Comparison of e.r.g and l.e.r.g. When the electrode was in the oil pool the input was open-circuited, and recordings suddenly became possible when the electrode contacted the retina. The l.e.r.g. obtained in this way differed from the conventional e.r.g. recorded with an active electrode in the vitreous humour. The best condition for comparison of these responses was obtained in cases where the vitreous humour was incompletely replaced. n such cases the e.r.g. could be recorded from the remaining vitreous, and the l.e.r.g. could then be obtained after advancing the electrode through the oil to the retina (Text-fig. 1). Note in Text-fig. 1 that the response polarities and forms are similar in the e.r.g. and l.e.r.g., but the latter is much bigger; the b-wave and off-response are particularly large. The offresponse of the cat, consisting of a rapid negative deflexion, is due to termination of the d.c. component (Brown & Wiesel, 1961a, b). The detailed form of the l.e.r.g proved quite variable, however; these variations in form are described later and were found to depend upon stimulus position relative to the electrode, stimulus intensity, and degree of light adaptation. Effect of stimulus position upon the l.e.r.g. The retinal area from which a response could be recorded by the surface electrode proved to be quite limited. Figure 2 shows results of an imustrative experiment. The effect of

5 LOCAL ELECTRORETNOGRAM UNDER OL 433 stray light was minimized by superimposing the brief stimulus upon a continuous background illumination. The retina, the electrode, and the stimuli were observed during the experiment through the final half-silvered mirror of the optical stimulator. t may be seen that the response was abolished by moving the stimulus spot about 500,u away from the electrode in one direction and about 75Q,z in the other direction. An active electrode in the vitreous humour records responses equally well from all parts of the A B 1\<~~~~~~~~~~~~~~5 luw~~~ 350 msec Text-fig. 1. Comparison of e.r.g. and l.e.r.g. n this experiment the vitreous humour was partially replaced by oil which was against the retina. Records are shown at three electrode locations as the micro-electrode was advanced from the vitreous humour (A) into the oil pool (B) and then on to the retinal surface (C). The reference electrode was in the mouth for all records of this paper. Record A shows a conventional e.r.g. n record B the electrode is in the oil and the input circuit is open, so only 60 cycle current is recorded. Records C shows l.e.r.g. recorded from the retinal surface under the oil. The stimulus was a disk having a retinal diameter of 5704c and giving a retinal illumination of 5-3 x 10-6 m/rnmm2. This disk was focused on the retina and carefully centred on the electrode in the case of Record (C. There was no background illuimination, but the retina was moderately light adapted by repeating the stimulus every 10 sec. This repetition rate was used for all records of this paper, unless otherwise noted. Positive deflexions are displayed upward in all records, following the e.r.g. convention. Amplification was direct-coupled in this and succeeding figures, unless otherwise noted. retina (Brown & Wiesel, 1961 a). Hence Text-fig. 2 shows the degree of response localization which may be achieved by recording under oil, even in the case of a macro-electrode on the retinal surface. The l.e.r.g. as a function of stimulus inten8ity. Text-figure 3 shows the l.e.r.g. as a function of stimulus intensity. These results were obtained after

6 B. ARDEN AND K. T. BROWN dark adapting the animal for 90 min. No attempt was made to eliminate minor sources of extraneous light in the experimental room, such as leaks under doors and pilot lamps; thus some light may have reached the animal through cracks in the metal shield around it, and dark adaptation may not have been complete. Text-figure 3 shows that the weakest stimulus light used (3-5 x lm/mm2) evoked a negative potential having a maximum amplitude of about 160,tV. t was impractical to use weaker stimuli because of the noise level in the records. This stimuli intensity corresponds Stimulus position Response - -. / _, 230a \ /s \ 320 msec Text-fig. 2. Localization of l.e.r.g. recorded under oil. The diagram on the left shows, to scale, the sizes and positions of the stimulating disks of light and the silver-in-glass macro-electrode. On the right are the responses evoked from the illustrated stimulus locations. The stimulus disk gave a retinal illumination of 5-5 x 10-4 m/mm2. A background illumination of 2-4 x 10-5 im/mm2 was used to suppress the effect of stray light. to a quantal flux at 507 m,t of 4957/sec. Assuming 30 % absorption of incident light by the rods, and a summation time of 0-1 sec, the response shown is evoked by the absorption of only about 149 quanta. The stimulus covered about 1-00 mm2 of retina, and such an area would contain about 100, ,000 rods. Hence the evoked response results from about one

7 LOCAL ELECTRORETNOGRAM UNDER OL 435 rod per thousand absorbing one quantum. Under these circumstances the chance of any receptor absorbing more than one quantum is remote. The micro-electrode often detected nerve impulses in addition to slow potentials. Although impulses were usually eliminated from the records by a high frequency filter, they could be heard on the audio amplifier. t was therefore possible to determine the threshold for ganglion cell impulses during an intensity series like that of Text-fig. 3. At the lowest level of >531A 53xlo3-107x x x14 X1- x1o8 5-35x xlo 3-5 x10- E 3- ~x1o x10-11 _ msec 350 msec Text-fig. 3. The effect of stimulus intensity upon the l.e.r.g. under dark adapted conditions. These records were obtained with a micro-electrode on the retinal surface, following a 90 min period of dark adaptation. There was no background illumination; the stimulus spot was centred on the electrode and had a retinal diameter of 1140gu. The sequence of records was from the bottom left (lowest intensity used) to the upper right (maximum intensity available). A 1 min period was used between stimuli, but this may not have been sufficient to permit complete recovery following the highest intensity flashes. Note that amplification was cut to 2/5 for the right column of records; this was necessary to keep the responses within the face of the oscilloscope. retinal illumination used in the experiment of Text-fig. 3, it was impossible to be sure whether impulses were being discharged in response to the stimulus. But impulses were clearly heard when the stimulus intensity was increased 05 log units, to 1 x lm/mm2. A possibility considered was that the higher threshold of ganglion cell discharges was due to some deleterious effect of the oil upon ganglion cell activity. n a separate experiment the vitreous humour was not replaced by oil, and the threshold 28 Physiol. 176

8 436 G. B. ARDEN AND K. T. BROWN of the most sensitive ganglion cells was again at 1.1 x lm/mm2. Barlow, Fitzhugh & Kuffler (1957a) likewise determined thresholds for impulse activity in ganglion cells of the dark adapted cat retina; our technique in this regard is very similar to theirs, except that they used decerebrate preparations. Thresholds of the two studies are in good agreement. t therefore appears that the threshold for ganglion cell activity is slightly higher than the stimulus intensity at which a negative slow potential can be recorded. t seems noteworthy that our ganglion cell thresholds were not detectably altered by the oil, since this would appear to be a sensitive indication that the retina continued to function normally under the oil. Up to an intensity of 1 1 x 10-9 lm/mm2, only a negative response was evoked and no b-wave could be seen. Hence the stimulus intensity required to evoke a distinct b-wave was more than 2*5 log units above that required to evoke the negative response. The form of the negative response is noteworthy atintensity levels which arehigh enoughto evoke a clear response, but not high enough to evoke a b-wave (see the response to 1 1 x 10-10lm/mm2.) n this intensity range the response showed a rapid onset followed by a slow increase of negativity during the stimulus; when the stimulus terminated, a small rapid decay phase was followed by a larger and much slower decay phase. As stimulus intensity was increased, and the b-wave was evoked, a clear a-wave was thereby produced. t seems obvious that the rapid onset of the negative response, which may be isolated at low stimulus intensities, is responsible for the initial phase of the a-wave at higher intensities. Thus the negative response evoked at the lower stimulus intensities in Text-fig. 3 is the component which was designated P by Granit. Since P has now been identified as a receptor potential, the negative response of Text-fig. 3 will be thus designated throughout the remainder of this paper. t should be noted that when the receptor potential of the cat is isolated by selectively clamping the retinal circulation, it shows a small rapid decay phase followed by a larger and much slower decay phase (Brown & Watanabe, 1962b). Thus the rapid and slow decay phases of the receptor potential, when isolated in this work by low stimulus intensity, correspond to those which have been found by the previous method. These rapid and slow decay phases have been explained as a combination of cone and rod receptor potentials, since the cone receptor potential of the fovea of the Cynamolgus monkey decays rapidly whereas the rod receptor potential of the night monkey decays very slowly (Brown & Watanabe, 1962a, b). As the intermediate range of stimulus intensities was approached in Text-fig. 3, the response assumed its familiar form. The b-wave appeared, grew to a maximum amplitude of 8x2 mv peak-to-peak at an intensity of

9 LOCAL ELECTRORETNOGRAM UNDER OL x 1O-5 lm/mm2, and then decreased in amplitude at the highest intensities. This decreased amplitude of the b-wave at very high stimulus intensities has been noted previously by many investigators. Note that the a-wave does not show this effect, but increases in amplitude to the highest stimulus intensity which was used. The off-response became a rapid negative deflexion at intermediate intensities, as usually found in the cat, but at the highest intensity it became a positive deflexion again. t therefore appears that the recorded off-responses of the cat retina are due to termination of the cone receptor potential at both very low and very high intensities, but are due to termination of the d.c. component at intermediate intensities. Note that the latency of the positive deflexion was less, at both the low intensities and the highest intensity, than the latency of the negative deflexion at intermediate intensities. Hence the suggested interpretation of the polarities of these off-responses is supported by the latency values, but this point requires further study. Throughout most of the intensity range there is a slow negative potential which persists after the rapid off-response. This has been referred to in e.r.g. work as 'remnant negativity', which has now been identified as due to the slow decay phase of the rod receptor potential (Brown & Watanabe, 1962b). This slow decay phase first increased with stimulus intensity, attained a maximum at about 1-7 x 1O-7 lm/mm2, and then decreased and disappeared at -an intensity of about 5-3 x 1O-5 lm/mm2. Above this intensity the c-wave appeared and increased in amplitude up to the maximum intensity used. Only the early portions of the slow decay phase or the c-wave may be seen in Text-fig. 3, and since these responses are opposite in polarity the actual amplitudes cannot be measured. t seems evident, however, that the slow decay phase of the rod receptor potential is larger than the c-wave at low and intermediate intensities, whereas the reverse is true at high intensities. Effect of stimulus area on the receptor potential. The effect of stimulus spot size was studied with the stimulus intensity low enough so that only the receptor potential was evoked. The amplitude of the receptor potential increased as the diameter of the stimulus disk increased to 076 mm, but further increase in size of the stimulus disk produced no further effect upon the receptor potential. Thus the retinal area over which the effect of spot size summated (the summation area) was small in the case of the receptor potential. Of course the summation area is determined partly by the retinal area from which the electrode could record activity. The summation area may be larger than the recording area, however, if there are convergent pathways from the receptors to the site at which the recorded response is generated. Since little convergence may be anticipated in the case of a receptor potential, the small summation area which was found for the receptor potential is in agreement with its identification. Although 28-2

10 438 G. B. ARDEN AND K. T. BROWN weak background illumination was used to suppress stray light in these experiments, it might not have been completely effective. f stray light had any effect upon our measurements of summation area, however, it would be due to light scatter from the outer border of the stimulus disk into the centre of the stimulus spot. This would cause the recorded summation area to be larger than the true one; hence the recorded summation area is probably a maximum value. t seems unlikely that the recording area of the electrode could exceed the summation area of the receptor potential, so the summation area of 0-76 mm in diameter provides an estimate of the x10-7 ;, x107 N E 8 x108 i x ([x x msec Text-fig. 4. Relative summation areas of the receptor potential and b-wave, demonstrated with disk and annular stimuli. These records were obtained with a micro-electrode on the retinal surface, following a preliminary period of dark adaptation. The stimulus pattern was carefully focused and centred upon the electrode in this and all succeeding figures of the present paper. The upper three records constitute one series, and the lower three records are another series at higher stimulus intensity. n each series the product of area and intensity was the same for all three stimuli. The background retinal illumination was 2-4 x 10-9 lm/mm2. maximum retinal area from which the electrode can record directly. This value is useful, since any response which yields a larger summation area is probably subject to convergent pathways from the receptors to the site at which the response is generated. Relative summation areas ofthe receptorpotential and b-wave. Experiments were also performed with small exploring stimulus disks at low enough intensities to isolate the receptor potential. f the disk was moved just far enough from the electrode to abolish the receptor potential, it was found that a b-wave could then be evoked by increasing the stimulus intensity. This suggested that the summation area of the b-wave was greater than

11 LOCAL ELECTRORETNOGRAM UNDER OL 439 that of the receptor potential. Thus the relative summation areas of the receptor potential and b-wave were studied by using centred stimulus disks, and annuli of varying size, as shown in Text-figs. 4 and 5. This was done at high enough stimulus intensities to evoke both a- and b-waves; under these conditions the a-wave is an index of the amplitude of the receptor potential. Retinal illumination (lm/mm2) 2 x x 10-8 :Z 190o0 1- : Xo C a b * C J\J - ~~~~~~E _J WV q Ạ 350 msec 350 msec Text-fig. 5. The effect of increasing annulus size upon the l.e.r.g. Responses recorded by a micro-electrode on the retinal surface. The background retinal illumination was 2-4 x 10-9 lm/mm2. n Text-fig. 4 the upper three records show responses to a small disk and two annuli. The product of area and intensity was approximately constant for the three stimuli. The disk evokes a distinct a-wave but only a small b- wave. The small annulus shows little or no change in the a-wave, but a larger b-wave, and the larger annulus shows a smaller a-wave with little or no change in the b-wave. Still larger annuli evoked no distinct response. n the lower three records, the experiment was repeated with three annuli and a higher stimulus intensity. Again, the product of area and intensity was constant for the three annuli. The a-wave decreases in amplitude with

12 440 G. B. ARDEN AND K. T. BROWN annulus size, and is absent in response to the largest annulus. But the b- wave amplitude decreases less rapidly, and the largest annulus which evokes no a-wave still evokes a b-wave. This b-wave is small because of the low stimulus intensity required to maintain a constant product of intensity and area as the annulus size increased. Text-figure 5 shows results of an experiment in which the stimulus intensity was constant for each series of annuli, and results are shown at two levels of intensity. At the higher intensity level, a small centred disk evoked clear a- and b-waves. With annuli of increasing size, the a-wave quickly disappeared. The amplitude of the b-wave first increased, probably due to the increased stimulus area, and then decreased. The results at the two levels of stimulus intensity show that the retinal area from which a b-wave could be evoked was dependent upon the stimulus intensity. n the experiments of Text-figs. 4 and 5, the effects of stray light were reduced by using background illumination. t appears that this effectively prevented stray light from stimulating the area in the centre of the annulus. Otherwise the disappearance of the a-wave when the inside diameter of the annulus reached 0 57 mm (see Text-fig. 4) could not be explained. This value of 0 57 mm diameter for the summation area of the receptor potential is somewhat smaller than the value obtained by spots of increasing diameter. Thus the summation area of the receptor potential appears to have a diameter of the order of mm. These experiments demonstrate that the a-wave can be recorded only from a small retinal area immediately subjacent to the electrode, while the b-wave may be evoked by more distant illumination. t should be noted especially that the b-wave could be evoked by annuli which did not evoke an a-wave, in spite of the fact that the a-wave is more sensitive than the b-wave to centred stimulus disks. t therefore seems evident that the b-wave can be evoked by stimulating receptor areas which are too far from the electrode to evoke an a-wave. Thus the b-wave is subject to a greater convergence of pathways, from the receptors to the site at which the b-wave is generated, than is true for the a-wave. This is not surprising since the b-wave is generated by cells of the inner nuclear layer, where a greater convergence of pathways may be expected than at the receptor level. The actual size of the receptor area from which a b-wave can be evoked, however, is difficult to determine. As annulus size increases, the total stimulus area also increases and stray light becomes more important. This may not be entirely eliminated in the cat because of the highly reflecting tapetum lucidum. Also the measured summation area of the b-wave increases with stimulus intensity. This is analogous to the increased size of ganglion cell receptive fields with stimulus intensity (Kuffler, 1953). Hence our findings for the b-wave may represent detection, at the level of the inner nuclear

13 ~~~~~~~~~~~~~E E LOCAL ELECTRORETNOGRAM UNDER OL 441 layer, of at least part of the mechanism which underlies the similar findings in the case of ganglion cells. Responses of the light adapted retina. When strong background illumination was employed, which light adapted the retina, responses were recorded which differed in detail from more dark adapted responses. The response form depended somewhat upon the intensity of the background illumination and the intensity and size of the stimulus disk. As illustrated in Text-fig. 6, an a-wave was often not visible and the rising phase of the c S g c > E e i msec Text-fig. 6. The effect of increasing annulus size upon the l.e.r.g. recorded from a strongly light adapted retina. Records obtained by a micro-electrode on the retinal surface. The retina was maintained in the light adapted state by a background retinal illnation of 2 4x10x4 /mmn2. The stimulus intensity was 5 x 10-3 hn/mn2 b-wave was rather fast. Also the negative off-response was especially large, in agreement with the general finding that off-responses of the e.r.g. are larger in the light adapted state. n Text-fig. 6 the area from which the b-wave was elicited was about the same as in previous experiments, but the larger annuli evoke a slow negative potential which develops during the stimulus and persists after termination of the stimulus. This response is distinct from the negative receptor potential, and will be referred to as surround negativity. Whereas the receptor potential has a shorter latency than the b-wave, the latency of

14 442 G. B. ARDEN AND K. T. BROWN surround negativity is longer than that of the b-wave. Also the onset of the receptor potential is quite rapid, while that of the surround negativity is much slower. n addition, the surround negativity is best evoked by large annuli which do not evoke a-waves. t therefore seems clear that surround negativity is a separate component of the e.r.g. which has not been described previously. Surround negativity becomes more prominent with light adaptation. Of course the most efficient method of light adapting the retina is background illumination, but this is not necessary. Surround negativity has also been noted in the early stages of dark adaptation. t likewise appears when the retina is light adapted by the repetition of a rather intense stimulus (see responses to annuli in the left column of Text-fig. 5). tr Text-fig. 7 shows responses of the light-adapted retina to a small disk, large disk, and large annulus; such responses were obtained at a variety of stimulus intensities. Note that responses to the small disk cannot be matched with responses to the large disk by varying relative stimulus intensities. Thus the illustrated effects of stimulus area are not mimicked by stimulus intensity. At any given stimulus intensity, changing from the small to the large disk brought in surround negativity but caused a decrease in the b-wave and off-response. The surround negativity evoked by the annulus, however, appears even greater than with the large disk. These effects were found over the entire range of stimulus intensity illustrated. Note that surround negativity could not be seen clearly at any stimulus intensity in response to the small disk, and that surround negativity elicited by the annulus increased in amplitude with stimulus intensity. A careful scrutiny of Text-fig. 7 strongly suggests that there is some type of interaction between the b-wave and off-response, on the one hand, and surround negativity on the other hand, which cannot be attributed to the simple summation of potentials elicited by the central and peripheral portions of a large stimulus disk. This possibility was tested by recording responses to disk and annular stimuli of equal area, and responses to a combination of these two stimuli. llustrative results are shown in Text-fig. 8. n this experiment the disk and annulus were contiguous, and the annulus elicited a small b-wave in addition to surround negativity. Each response in Text-fig. 8 was obtained as the average of five records, traced in an enlarger. n the lower pair of records the tracing marked 'sum of responses' is the algebraic sum of the responses to the disk and annulus, when presented separately. Hence this tracing gives the result predicted on the assumption that there is only direct electrical summation of the responses to disk and annulus, with no interaction between these responses. The tracing marked 'both' is the actual response obtained when the disk and annulus were presented

15 LOCAL ELECTRORETNOGRAM TJNDER OL 443 simultaneously in a large disk with outer diameter equal to that of the annulus. Note that in the actual response the b-wave, off-response, and surround negativity are all smaller than predicted if there is only direct electrical summation. This result was confirmed in cases where the inner diameter of the annulus was considerably greater than the outer diameter of the disk, and with various levels of background illumination. The interaction was found only when the background illumination was sufficiently 4 x10-3 Disk 760/s Disk 2280/s A/ Annulus /s 2 x _a x1o3 4 x10-4 um JKAV7S ~AWAV 8 x10-5, u o, 1, 350 msec Text-fig. 7. The effects of stimnulating with a small disk, large disk, and large annulus upon the shape of the l.e.r.g. Responses were recorded by a microelectrode on the retinal surface andwere obtainedover a range of stimulus intensity. The retina was maintained in a light adapted state by a background retinal illumination of 2-4 x 10-4 im/mm2. intense strongly to light adapt the retina, and produce a marked degree of surround negativity. Since the interaction was found only under these conditions of background illumination, which reduced the effectiveness of stray light, this finding seems to preclude the possibility that the observed

16 444 G. B. ARDEN AND K. T. BROWN interaction may be attributed to stray light. t therefore appears that the annular stimulus reduced the b-wave and off-response evoked by the centred disk, and that the centred disk reduced the surround negativity evoked by the annulus. Thus the cells stimulated by the disk must interact in some manner with cells stimulated by the annulus, so that there is mutual inhibition between responses evoked by the disk and annulus S _~.,.. Disk - ~~~~ % C u: 0*0 *.. Annulus Both 04F. _ 1 <:i Sumof responses L_ ~~~~~~~ ~~~~~~~ 100 msec time marks Text-fig. 8. nteraction between responses to disk and annular stimuli. Responses were recorded by a micro-electrode on the surface of the light adapted retina. The background retinal illumination was 2-4 x 10-6 lm/mm2, and the stimulus intensity was 2 x 10-5 lm/mm2. Each plotted curve is the average of 5 responses. The stimulus disk had a retinal diameter of 1330g; the annulus had an inside diameter of 1330/s and an outside diameter of 1900/t. The curve marked 'sum of responses' is the algebraic sum ofthe responses to disk and annulus, when presented separately. The curve marked 'both' is the actual response to the disk and annulus, when presented simultaneously in a single large disk with an outside diameter of 19004i. See text for further explanation. Recordings with penetrating micro-electrodes General findings and sources of variability. t was found by Brown & Wiesel (1961a) that an intraretinal micro-electrode near the pigment

17 LOCAL ELECTRORETNOGRAM UNDER OL 445 epithelium recorded all components of the l.e.r.g., except for the c-wave, at inverted polarity with respect to the conventional e.r.g. This was true regardless of whether the reference electrode was in the vitreous humour or at a remote location (on the back of the head); in the latter case, however, the inverted local responses were contaminated by responses from distant retinal areas. n this work the l.e.r.g. was recorded without such contamination, even when the reference electrode was at a remote location (in the mouth). The polarity and general form of the deep intraretinal l.e.r.g., recorded under these conditions, was as previously reported. t was noted that the form of the l.e.r.g. changed markedly with light adaptation, as illustrated in Text-fig. 9. Response A was taken from about the level of the inner nuclear layer after 30 min of dark adaptation. This response shows a large negative b-wave, followed by a large positive c- wave. No a-wave is visible, since this is best seen when the electrode tip is near the retinal side of the pigment epithelium. Response B was then obtained after turning on the background illumination; this greatly reduced both the b- and c-waves, and a small a-wave became visible. Responses A and B proved to be extremes, between which various proportions of a- and b-waves could be found. Record C was taken sec after turning off the background illumination. At this early stage of dark adaptation the a-wave was still visible, though the b- and c-waves had increased in amplitude. Hence the effects of light adaptation seem to include a marked reduction in amplitude of the b-wave, thereby revealing (or perhaps increasing) the a-wave. This finding suggests that light adaptation reduces the amplitude of the b-wave much more than that of the receptor potential; this has been demonstrated in monkeys by studying isolated receptor potentials (Brown & Watanabe, unpublished). Thus greater effects of light and dark adaptation may be observed in studies of the b-wave than in studies of the a-wave or isolated receptor potential. This subject was not pursued in the present study; it is of special interest, however, and accounts for some of the variations among responses illustrated in this paper. Another source of variation were unpredictable differences in response forms during different penetrations. Satisfactory penetrations were especially difficult to obtain in this study; the reasons for this are not clear, but the oil may make the surface membranes of the retina more difficult to penetrate. Attempts to penetrate often led to the appearance of a pressure spot on the tapetum; this indicates poor penetration (Brown & Wiesel, 1959). Even electrodes which could penetrate successfully showed considerable variability in response forms. For example, one electrode which seemed to penetrate easily was used three times in succession. The l.e.r.g.'s on the first and third occasions resembled record A of Text-fig. 9, even

18 446 G. B. ARDEN AND K. T. BROWN when the electrode was rather close to the pigment epithelium, while on the second occasion a response similar to record B of Text-fig. 9 was obtained even when the electrode was near the inner nuclear layer. Thus the b-wave was selectively reduced on certain penetrations, whereas the a-wave was selectively reduced on other penetrations. Such effects have been noted to some extent in other series of intraretinal recordings, but seemed exaggerated in the present series of experiments. The most likely explanation A _ f C 350 msec Text-fig. 9. Relative effects of adaptation upon the a- and b-waves. These records were obtained by an intraretinal micro-electrode at about the level of the inner nuclear layer. Note that the a- and b-waves are inverted with respect to those recorded from the retinal surface. Record A was obtained after 30 min dark adaptation. Record B was obtained during light adaptation with a background retinal illumination of 2-4 x 10-5 lm/mm2. Record C was obtained sec after turning off this background illumination. The stimulus was a disk with a retinal diameter of 950p4, providing a retinal illumination of 10-4 lrn/mm2. of this variation is that the electrode sometimes causes local interference with the retinal circulation, and on other occasions causes local interference with the choroidal circulation. Since the b-wave is dependent upon the retinal circulation, whereas the a-wave can be supported entirely from the choroidal circulation (Brown & Watanabe, 1962 a), this explanation seems

19 LOCAL ELECTRORETNOGRAM UNDER OL 447 plausible. Some interference with local circulation by intraretinal electrodes in this series of experiments is also suggested by the finding that maximum light sensitivity was 1x5 log units lower when determined with intraretinal electrodes than when determined by surface electrodes. L.e.r.g.'s as a function of electrode depth. Studies of the l.e.r.g. as a function of electrode depth are best performed during withdrawal of the electrode from the retina (Brown & Wiesel, 1961 a, b). Text-figure 10 shows such a sequence of depth records; the responses were elicited by light flashes of about 20 /tsec total duration and of very high but unspecified intensity. These stimuli were generated by a Grass photo-stimulator and were led to the eye by a fibre optics bundle (Brown & Murakami, 1964). The intense brief stimulus produced large responses in relation to the amount of light adaptation caused by the flash, and therefore both the a- and b-waves were quite large. No background illumination was used. As the electrode was withdrawin, both the a- and b-waves inverted polarity while the electrode was still in the retina. Note that the latency of the a-wave, the peak time of the a-wave, and the peak time of the b-wave all remained the same as the electrode was withdrawn through the retina. The lower straight line in Text-fig. 10 represents the R-membrane. The lowest record, which was closest to the R-membrane, shows a large positive a-wave and a negative b-wave. As the electrode was withdrawn, the a-wave became smaller. t then inverted between the third and fourth records from the bottom. Note that the b-wave was still negative at the point where the a-wave had inverted. As the electrode was further withdrawn, the b-wave also inverted polarity. t is therefore clear that the a-wave inverts polarity deeper in the retina than the b-wave. Electrode depths were not measured accurately in this experiment, so no depth scale is given. The ordinate in Text-fig. 10 represents relative electrode depth, with the deepest record at the lowest position. The most superficial record in Text-fig. 10 was from or very near the inner nuclear layer. This record shows degenerated spikes superimposed on the l.e.r.g., which frequently occurred in the inner nuclear and ganglionic layers. Responses which are superficially similar to those of Text-fig. 10 have been obtained in isolated cold-blooded retinas (Tomita et al. 1960). n such experiments, the reversal in polarity with electrode depth only occurred when the retina was illuminated with diffuse light and was not found when the stimulus was confined to the electrode tip (Tomita & Torihama, 1956; Tomita et al. 1960). Brown & Wiesel (1961 a, b) found comparable polarity reversals in the cat retina, which they showed were due to the changing proportions of gross and local responses which occurred as the electrode penetrated the retina. They point out that such polarity inversions have no significance for localizing e.r.g. components, and that the only inversion

20 4. B. 448 ARDEN AND K. T. BROWN of significance for localization would be the inversion of a given component of the local response as the electrode penetrates the retina. A true inversion of the local response was first found for isolated receptor potentials of monkeys (Brown & Watanabe, 1962a, b), and this was found to hold for both rod and cone cases. Since the a-wave is the rising phase of the receptor potential, the depth distribution of the a-wave should correspond to that for the isolated receptor potential. The maximum positive receptor -: fi E ~~~~~~~~~25 msec R-membrane Text-fig. 10. Polarity reversals of local a- and b-waves during withdrawal of micro-electrode from retina. The bottom line represents the R-membrane, and records from the bottom to the top were taken at increasing distances from the R-membrane. Electrode depths were not measured in this experiment, so the vertical scale represents only relative electrode depth. The top record was obtained in or very near the inner nuclear layer. The stimulus was an intense 20/sec flash which was delivered to the eye from a Grass photo-stimulator by a fibre optics bundle. No background illumination. The time of the stimulus is indicated by the vertical interrupted line. The amplifier band pass was c/s. potential occurs about 30,t from the pigment epithelium (Brown & Murakami, 1964) while the maximum negative receptor potential occurs at the proximal receptor terminals which are presynaptic membranes to second order neurones (Brown & Watanabe, 1962 a, b). Although the absolute levels of the maximum positive and negative a-waves have not been

21 LOCAL ELECTRORETNOGRAM UNDER OL 449 localized as precisely in the cat, Text-fig. 10 and 11 show that the findings of this work correspond well with results on the isolated receptor potential of the monkey. Since the b-wave recorded in the vitreous humour is positive, while the b-wave previously recorded in the intraretinal l.e.r.g. has been negative, it was anticipated that the positive pole of the local b-wave would be closer to the retinal surface than the negative pole. Text-fig. 10 confirms this expectation; inversion of the local b-wave has thus far been found only by recording under oil, which permits the detection of responses which are close to the retinal surface. Perhaps the most significant new finding of Text-fig. 10 is that the b-wave inverts polarity at a more superficial level than the a-wave. Since the method of recording in Text-fig. 10 gives highly local responses, the demonstrated inversions should be true inversions of local responses, regardless of the fact that the stimulus covered the entire retina. This point was confirmed by experiments with local illumination, as illustrated in Text-fig. 11. The stimulus for this experiment was a disk 0-76 mm in diameter and centred on the electrode tip. A stop was inserted into the background beam so that a large area of retina outside the stimulus disk was constantly illuminated. This suppressed the influence of stray light in regions outside of the stimulus disk. The responses obtained are similar to those of Text-fig. 10, in that polarity inversions may be seen for both the a- and b-waves. The longer but less intense stimulus in this experiment evoked a smaller a-wave, however, and a larger b-wave. The depth value of each record is shown in terms of the micrometer distance of the electrode from the retinal surface. The deepest record was taken with the electrode against the retinal side of the R-membrane where the electrode is next to the pigment epithelium (Brown & Tasaki, 1961). Note that the maximum amplitude of the positive a-wave was found after the electrode had been withdrawn a small distance from the pigment epithelium, in agreement with the finding of Brown & Murakami (1964) for the isolated receptor potential. n the region of the b-wave inversion, there was a complication of wave form because the electrode was recording a small S-potential in addition to the b-wave. Hence the b-wave, in the region of its inversion, was superimposed upon a sustained negativity. Note that the latencies of both the a- and b-waves remained constant throughout the retina, but the peak time of the b-wave changed. The peak of the negative b-wave corresponds to a well-marked shoulder on the rising phase of the positive b-wave. This asymmetry may be due partly to the c-wave. The amplitude of the c-wave is greatest near the pigment epithelium; hence it will decrease the amplitude and shorten the peak time of the negative b-wave recorded near the pigment epithelium (Brown & Wiesel, 1961 b). The asymmetry of negative and positive b-waves persists, however, and may even become

22 450 G. B. ARDEN AND K. T. BROWN E 50 msec -- e _M '4.5 0 ta :4,g : PC 02@' 240 -% J Text-fig. ll. The le.r.g. as a function of electrode depth. This series of records was obtained during withclrawal of the electrode from the retinal side of the R-membrane (upper left record) to the retinal surface (lower right record). The depth of each record is inclicated as the distance of the electrode tip from the retinal surface, along the electrode track, read directly in /t from the micrometer scale. Since the penetration was oblique, these distances are greater than during a vertical penetration. The stimulus was a centred disk having a retinal diameter of 760,u and a duration of 350 misec. The retinal illumination within the stilmulus spot was 5-3 xj1o-3h-m/mm2. The retinal area outside the stimulus was constantly illuminated by an annulus with inside diameter equal to that of the stimulus disk; this annulus gave a retinal illumination of 2-4 x 10-5 hn/mm2. The solid vertical time line marks the beginning of the stimulus. The first interrupted line is drawin through the rise of the a-wave from the base line. The second interrupted line is drawn through the peak of the deep negative b-wave, which corresponds in time to a wellmarked shoulder on the rise of the more superficial positive b-wave. The amplifier band pass was c/s.

23 LOCAL ELECTRORETNOCRAM UNDER OL 451 more obvious, when stimulus intensities are employed which are so low that the c-wave is almost absent. Amplitude of the b-wave as a function of electrode depth. t seems clear from Text-figs. 10 and 11 that the local b-wave inverts polarity as the electrode passes through that part of the retina in which the b-wave is generated. The slight asymmetry of the positive and negative b-waves in Text-fig. 11 indicates, however, that some portion of this response may not invert polarity or may invert polarity at a slightly different level. Hence it must be borne in mind that the findings reported here pertain clearly only to the major part of the b-wave. Plate 1 shows the b-wave amplitude as a function of electrode depth for 5 separate electrode withdrawals. Since the micro-electrode penetrated the retina at an angle, which varied from one experiment to another, the depth of the micro-electrode tip is given as a percentage of the total measured distance which the electrode traversed between the R-membrane and the retinal surface. Although the micrometer measurement of retinal thickness varied from 300 to 480,u, the results showed good agreement when plotted on the percentage depth scale. This indicates that the electrode moved smoothly through the tissue. The maximum amplitude of the negative b-wave was found near the outer margin of the inner nuclear layer. This agrees with the result of Brown & Wiesel (1961 a, b) which was confirmed by electrode marking (Brown & Tasaki, 1961). Tomita and his co-workers have stressed the concept of a radially oriented dipole, since the b-wave inverts polarity between the conventional recording method and an intraretinal lead (see Tomita et al. 1960). Assuming that the b-wave is due to a radial dipole, in which the poles are separated by a short distance, Tomita & Torihama (1956) point out that the maximum slope of the curve relating b-wave amplitude to electrode depth should occur between the two poles. The assumptions of this criterion are validated by the results of Text-figs. 10 and 11, and P1. 1 shows that the slope of b-wave amplitude as a function of electrode depth is maximum in the inner nuclear layer. Thus the polarity inversion occurs quite sharply in the inner nuclear layer. The amplitude of the positive b-wave increases, however, to a maximum which is reached at about the inner margin of the inner plexiform layer. This suggests that the cells which generate the b-wave have processes which extend through the inner plexiform Jayer. After the positive b-wave attained maximum amplitude during electrode withdrawal, it usually remained constant through the remainder of the surface layers of the retina. Thus if a reference electrode is in the vitreous humour, its potential level is that of the positive b-wave and no positive b-wave can be recorded by a penetrating microelectrode. This was the result obtained by Brown & Wiesel (1961 a, b) when the reference electrode was in the vitreous humour. 29 Physiol. 176

24 B. ARDEN AND K. T. BROWN Results of the two methods are in good agreement, but results of the present method are more informative. This is because the present method permits the recording of local responses with a more remote location of the reference electrode, so that both the positive and negative poles of the b- wave may be detected. Responses to annular stimuli as afunction of electrode depth. Responses to annular stimuli could only be studied when the retina was illuminated with 20 Z E o msec Text-fig. 12. The l.e.r.g. as a function of electrode depth, using annular stimuli. These records were obtained during withdrawal of a micro-electrode from the retinal side of the R-membrane (bottom record) to the retinal surface (top record). Electrode positions are indicated by the distance of the electrode tip from the retinal surface, along the electrode track, read directly in,u from the micrometer scale. The retina was constantly light adapted with a background retinal illumination of 2-4 x 10-6 m/mm2. The stimulating annulus had an inside diameter of 2090A and an outside diameter of 22805l; it gave a retinal illumination of 10 3 hm/mm2.

25 LOCAL ELECTRORETJNOGRAM UNDER OL 453 a bright background illumination, to avoid the effects of scattered light in the centre of the annulus. Under such conditions the negative b-wave which could be recorded by an intraretinal electrode was very small. Text-figure 12 shows, as a function of electrode depth, responses which were evoked under the described conditions by large andular stimuli. This experiment was performed primarily to study the depth distribution of surround negativity. During this electrode withdrawal the responses to a centred stimulus disk changed in basically the same manner as previously described, and served as controls for the illustrated annulus series. Near the pigment epithelium the responses to a centred disk showed a receptor potential, which could not be evoked by annuli with inside diameters exceeding 0 57 mm, in agreement with the findings by surface electrodes. Even the largest annuli evoked a c-wave, however, as shown in Text-fig. 12. As the electrode was withdrawn, the annulus evoked a slow negative potential which reached maximum amplitude near the retinal surface. t seems obvious that this slow negative potential is the component designated surround negativity, which may be recorded by a surface electrode. Although the steady decrease in amplitude of the c-wave as the electrode is withdrawn would cause an apparent increase of surround negativity, this effect must have been small because the amplitude of the c-wave was small even near the pigment epithelium. Text-figure 12 shows that no positive potential was found deep in the retina which corresponded in time course to surround negativity. The onset of surround negativity appears to have been recorded, however, at least half of the way through the retina. Hence if a positive pole of this response exists, the results indicate it would have to be in the receptor layer. At that level it might theoretically be masked by the positive receptor potential and/or c-wave. But the only positive potential which can be seen at the receptor level is a small c-wave, so this possibility seems remote. Also if surround negativity has its positive and negative poles separated by such a great distance, it could only be generated by the Muller cells. Unless this is the case, it would appear that this component is not generated as a radially oriented dipole. DSCUSSON Technique for recording local responses Local potentials have been recorded previously from the isolated retinas of cold-blooded vertebrates. n such cases the vitreous humour and pigment epithelium may both be removed, and either retinal surface may be dried by exposure to air (see Tomita et at. 1960). Local responses have also been obtained in mammals by placing the reference electrode in the vitreous humour. There is no evidence that the oil technique yields more 29-2

26 B. ARDEN AND K. T. BROWN highly localized responses than previous methods, but there are other advantages. Compared with isolated retinas of cold-blooded animals, the preparation has its normal circulation. t is therefore stable over long periods, and the retina is very sensitive to light. Compared with the previous mammalian technique, advantages are: (1) Much larger potentials may be recorded, and therefore responses may be obtained to lower stimulus intensities. (2) Responses may be recorded from the surface layers of the retina, as well as the deeper layers. (3) Local responses may be recorded with a surface macro-electrode, thereby eliminating the necessity for micro-electrodes in certain kinds of work. The use of surface electrodes also makes this tecbnique applicable to the open eye, which may prove useful for small eyes where the closed preparation is difficult or impossible. The major limitation of this method is that the consistency of the vitreous humour varies greatly between animals, and replacement of the vitreous by oil may not be feasible in all animals. ts replacement has so far not been possible in the Cynamolgus monkey. Hence both methods for recording l.e.r.g.'s in mammals have special applications. Convergent retinal pathways indicated by the l.e.r.g. The receptor potential is affected only by stimulation of a small retinal area centred around the electrode. ts recorded amplitude summates over a smaller retinal area than the S-potential (Watanabe & Tosaka, 1959) or the ganglion cell (Barlow et al. 1957b). Thus all responses which are known to be generated by single cells of the second order or higher, show summation over a larger retinal area than the receptor potential. This indicates a functional convergence of pathways from receptors to higher order neurones, which may be expected on anatomical grounds. This study shows that convergence may also be demonstrated in the l.e.r.g. by comparing the a-wave (which is the leading edge of the receptor potential) with the b-wave which is generated in the inner nuclear layer. ndications of convergent retinal pathways, detected by recording e.r.g. components, have also been obtained in retinas of cold-blooded vertebrates. t has been shown in the isolated and inverted carp retina that there is a surface-positive potential which is more localized than a surface-negative potential (Motokawa, Oikawa, Tasaki & Ogawa, 1959; Motokawa, Tamashita & Ogawa, 1961). Because of the retinal inversion, the polarity of their positive potential corresponds to that of the receptor potential in this work. Evidence from depth recordings indicated that the localized positive potential was probably produced by the receptors, and that the less localized negative potential was produced by cells of a slightly deeper layer (Motokawa et al. 1959). Hence these findings are in general

27 LOCAL ELECTRORETNOGRAM UNDER OL 455 agreement with our results on the receptor potential and b-wave, but this comparison must be made with reservations. A more direct comparison may be made with the work of Cone (1963) who found evidence for interaction at or before the site of generation of the b-wave in the rat retina. Sensitivity Previous studies of both the l.e.r.g., and the e.r.g. by standard recording methods, have shown that the b-wave can be evoked by lower stimulus intensities than the a-wave. The findings of this study indicate that a receptor potential, the leading edge of which contributes the a-wave at high stimulus intensity, can be evoked by stimulus intensities which are too low to evoke a detectable b-wave. This raises the question of why the receptor potential has not been detected at lower intensities than required for the b-wave in earlier studies. By comparison with previous work on the l.e.r.g. of the cat (Brown & Wiesel, 1961 a, b), the signal level in this study has been more than doubled and the noise level has been somewhat lowered by using electrodes of lower resistance. mprovement of signal-noise ratio, combined with the long periods of dark adaptation which could be employed with surface recording, probably account for our ability to record receptor potentials in this study at stimulus intensities where the b-wave had disappeared. By comparison with standard methods of recording the e.r.g., our signal-noise ratio is little if any greater (see Cone, 1963). Hence a different explanation must be sought. n experiments on the l.e.r.g., activity is recorded from a small retinal area of given size surrounding the electrode. When the size of the stimulus disk is increased beyond the summation area of the receptor potential, the additional receptors stimulated would not contribute to the recorded receptor potential but they would contribute to the recorded b-wave because of the convergent pathways to the site where the b-wave is generated. Thus for each unit of retinal area, the b-wave is favoured relative to the a-wave when large stimuli are used. Large stimuli must be used to evoke a clear response in the e.r.g., if recorded by standard methods. Correspondingly, we have noted that the b-wave is larger in relation to the a-wave in the e.r.g. than in the l.e.r.g. evoked by small stimulus spots. This means that in the e.r.g. it would be even more difficult to detect the receptor potential at stimulus intensities low enough to abolish the b-wave. This seems to have been accomplished, however, by Armington, Tepas, Kropfl & Hengst (1961), using averaging techniques to increase greatly the signal-noise ratio of corneal recordings. These authors used small stimuli confined to the human fovea, and found that when stimulus intensity was too low to evoke a clear b-wave a negative response (less than 21,V in amplitude) could be detected. Although the

28 456 G. B. ARDEN AND K. T. BROWN authors did not identify this negative response, it seems evident that it was the receptor potential. t had the appropriate polarity and latency, and was larger in response to foveal than to peripheral stimuli. t therefore appears that stimulus intensities which are too low to evoke a detectable b-wave may still evoke a receptor potential, in the e.r.g. as well as the l.e.r.g., providing that recording methods are sufficiently sensitive. The lowest threshold previously reported for an e.r.g. component appears to be that of Cone (1963) who detected the b-wave of the rat retina at a stimulus intensity where it was computed that a quantum was absorbed by only about 1 rod in 200. n the present work the receptor potential was detected at a stimulus intensity where only about 1 rod in 1000 absorbed a quantum. This receptor potential of the cat was evoked at a retinal illumination which was only about one log unit above the absolute dark adapted visual threshold, as determined on one of us (G.B.A.) with the same apparatus. Since the receptor potential had an amplitude of 160QV, it is evident that improved recording methods will make the receptor potential detectable at even lower intensities. Since the lowest stimulus intensity which evoked a clear receptor potential in this work did not give a detectable b-wave, it is not surprising that this intensity was subthreshold for ganglion cells. Our findings for the summation area of the receptor potential, in combination with receptive field studies on ganglion cells in the dark adapted state (Barlow et at. 1957b), indicate that with increase of stimulus disk size beyond about 0*76 mm the recorded receptor potential would be unaffected but the threshold of the ganglion cell would fall. This is in accord with evidence that maximum sensitivity of the dark adapted human retina is due to summation of activity over a large retinal area (Arden & Weale, 1954; Weale, 1958). The fact that receptor activity could be detected at intensities only slightly above the human dark adapted threshold means that the receptors pass significant amounts of current into the extracellular circuit. This could be due to strong electrical activity of each independent receptor and/ or to interaction between receptors. Evidence for the latter view is provided by Text-fig. 3. The receptor potentials which were recorded at the low stimulus intensities were evoked by such low intensities that only a small fraction of the receptors could have absorbed 1 quantum each, and the chance of any receptor absorbing two or more quanta was extremely remote. f the effect of increasing stimulus intensity under these conditions were simply the activation of a greater number of independent receptors, the evoked potential should simply get bigger. The latency of the receptor potential decreased, however, which indicates interaction between receptors. This evidence for interaction at the receptor level is similar to that of Cone (1963) in the case of the b-wave. Since latency of the receptor

29 LOCAL ELECTRORETNOGRAM UNDER OL 457 potential is affected by interaction between receptors, this must be responsible for at least part of the change of b-wave latency. Component analysis of the e.r.g. The major features of response forms recorded in this study can be accounted for by components previously isolated, with the exception of the component which we have designated surround negativity. Surround negativity is opposite in polarity to the b-wave, d.c. component, and c- wave. Thus the only previously isolated components, from which surround negativity must be distinguished, are the rod and cone receptor potentials. n contrast to the negative receptor potentials, surround negativity is evoked by annular stimuli which are too large to evoke an a-wave, and is enhanced by light adaptation. Likewise surround negativity has maximum amplitude close to the retinal surface, and does not appear to be generated by a radially oriented dipole. These criteria seem sufficient to ensure that surround negativity is distinct from the negative receptor potentials, and is therefore a new component of the e.r.g. Origins of e.r.g. components Certain findings of this work relate to the conclusion that the a-wave is generated by the receptors, while the b-wave is generated by cells of the inner nuclear layer (Brown & Wiesel, 1961 b; Brown & Watanabe, 1962 a, b). t may be expected from this conclusion that the a-wave will sum over a smaller retinal area than the b-wave. t may also be expected that a receptor potential will be generated at any level of stimulus intensity which elicits a b-wave. Both of these expectations are now confirmed. n this study the oil technique revealed a polarity inversion of the local b-wave in the inner nuclear layer, at a shallower depth than the polarity inversion of the a-wave on the same penetration. Since the negative b- wave was maximum at the outer margin of the inner nuclear layer, while the positive b-wave became maximum at the inner margin of the inner plexiform layer, the cells which generate the b-wave must be radially oriented and probably extend through the inner nuclear and inner plexiform layers. The bipolar cells are radially oriented and extend through the retinal depth indicated by these results (see Polyak, 1957, p. 232). Thus the anatomy of the bipolar cell fits our results, but this does not seem true for any other cell of the inner nuclear layer. The bipolar cells have frequently been suggested as possible generators of the b-wave, but experimental evidence to support this suggestion has been lacking. The origin of the new component, surround negativity, has been investigated in a preliminary manner by intraretinal recording. This com-

30 458 G. B. ARDEN AND K. T. BROWN ponent appears to be absent in the deep layers of the retina and maximum in amplitude at a level which is closer to the retinal surface than the origin of the b-wave. The candidates for cells which generate this response would therefore include especially the amacrine cells and ganglion cells. E.r.g. components which have been investigated previously have been shown not to be generated by ganglion cells, but this possibility has not been eliminated for surround negativity. Mode of generation of e.r.g. components Tomita et al. (1960) have emphasized that the polarity inversion of e.r.g. components between the conventional recording method and an intraretinal lead constitute evidence that such components are generated by radially oriented dipoles. This concept has been confirmed by local recording in the case of rod and cone receptor potentials (Brown & Watanabe, 1962 a, b); these potentials are generated deep enough in the retina so that both the positive and negative poles of the local response can be detected when the reference electrode is in the vitreous humour. Our findings for the a-wave are in agreement with those on the isolated receptor potential, and the concept has also been shown in this work to apply to the b-wave. The c-wave was not studied in this work, and inversion of the rapid offresponse of the cat was not successfully demonstrated. Polarity inversions of both the c-wave and d.c. component have now been demonstrated in the cat retina, however, at the expected levels (Brown & Murakami, unpublished). Thus most components of the e.r.g. are generated as radial dipoles, but this may not be true for all components. The peak times and wave forms of the positive and negative intraretinal b-waves were not always identical, which raises the possibility that some portion of the b-wave is not generated as a radial dipole. Also no source has been demonstrated which corresponds to the sink of surround negativity. f surround negativity is not generated as a radial dipole, this would provide an explanation for why it has not been demonstrated by recording the e.r.g. from the cornea. Comments on surround negativity The b-wave and off-response evoked by a small disk are reduced in the presence of an annulus, and surround negativity evoked by the annulus is reduced in the presence of a centred stimulus disk. t therefore appears that there is some form of mutual inhibition between the cells stimulated by the centred disk and those stimulated by the annulus. Surround negativity is enhanced by light adaptation, and the mutual inbibition has only been found in the light adapted state. Hence these findings are analogous to the mutual inhibition demonstrated between 'on' and 'off'

31 LOCAL ELECTRORETNOGRAM UNDER OL 459 responses which are elicited by stimulating central and peripheral portions of the receptive field of a light adapted ganglion cell of the cat (Kuffler, 1953). Under dark adapted conditions the stimulation of any part of the ganglion cell receptive field produces a response like that from the centre of the field (Barlow et al b). Only after light adaptation does the opposite response type become evoked by stimulating the peripheral parts of the field, and only then does mutual inhibition occur between the central and peripheral zones. Hence the two cases are so closely analogous that our findings may be functionally related to those with ganglion cells. SUMMARY 1. The technique of intraretinal recording in the unopened cat's eye was modified by replacing the vitreous humour with a transparent nonconducting heavy oil. n favourable cases this did not degrade the natural optics, but prevented lateral current spread from distant retinal points to the recording site. Under these conditions an electrode on the retinal surface recorded responses only to stimulation of a small zone immediately around the electrode, regardless of the position of the reference electrode. 2. Under dark adapted conditions the most sensitive component of this local e.r.g. (l.e.r.g.) was the receptor potential, whose leading edge contributed the a-wave at higher intensities. The receptor potential was evoked by stimulus intensities only slightly above the dark adapted human threshold. Higher intensities were required to evoke the b-wave or ganglion cell discharges. 3. The a-wave could be evoked only by stimulation of a small region subjacent to the electrode; the b-wave could be evoked by stimuli farther from the electrode which did not evoke a detectable a-wave. Thus the b- wave is influenced by convergent pathways from distant receptors. 4. Records with penetrating micro-electrodes revealed a previously undetected polarity inversion of the local b-wave in the inner nuclear layer. This polarity inversion was closer to the retinal surface than that of the a-wave during the same penetration. 5. These findings are all in agreement with earlier conclusions that the a-wave is the leading edge of a receptor potential, while the b-wave is generated in the inner nuclear layer. These findings also show that the b- wave is developed primarily by cells which give rise to a radially oriented dipole. This b-wave dipole appears to extend through both the inner nuclear and inner plexiform layers. Hence the b-wave is probably generated by the bipolar cells. 6. n response to large annular stimuli, an electrode at the retinal surface recorded a slow negative potential (herein designated 'surround

32 460 G. B. ARDEN AND K. T. BROWN negativity'). This response is opposite in polarity to all e.r.g. components previously isolated, except for the receptor potential. t was distinguished from the receptor potential by its longer latency, slower rate of rise, and the much larger retinal area from which it was elicited. Also penetrating electrodes showed its amplitude maximum to be near the retinal surface. Hence surround negativity is a new component of the l.e.r.g. The depth distribution of surround negativity makes it unlikely that this response is generated as a radially oriented dipole. Surround negativity appears only in the light adapted state; under these conditions mutual inhibition was demonstrated between e.r.g. components evoked by a small centred spot and surround negativity evoked by an annulus. We would like to thank Dr M. Murakami for assisting in some of the experiments, Mrs K. M. Howell for making electrodes, and Mrs G. K. McDonnell for the histological preparation. We are also indebted to Mr J. R. Wall and Mr S. Winston for help with the apparatus. This research was supported by Grant No. B-1903 from the National nstitutes of Neurological Diseases and Blindness, U.S. Public Health Service. REFERENCES ARDEN, G. B. & WEALE, R. A. (1954). Nervous mechanisms and dark-adaptation. J. Physiol. 125, ARMNGTON, J. C., TEPAS, D.., KROPFL, W. J. & HENGST, W. H. (1961). Summation of retinal potentials. J. opt. Soc. Amer. 51, BARLOW, H. B., FTZHUGH, R. & KUFFLER, S. W. (1957a). Dark adaptation, absolute threshold and Purkinje shift in single units of cat's retina. J. Physiol. 137, BARLOW, H. B., FTZHUGH, R. & KUFFLER, S. W. (1957b). Change of organization in the receptive fields of the cat's retina during dark adaptation. J. Physiol. 137, BRNDLEY, G. S. (1956). The passive electrical properties of the frog's retina, choroid and sclera for radial fields and currents. J. Physiol. 134, BRNDLEY, G. S. & HAMASAK, D.. (1963). The properties and nature of the R membrane of the frog's eye. J. Physiol. 167, BROWN, K. T. (1964). Optical stimulator, microelectrode advancer, and associated equipment for intraretinal neurophysiology in closed mammalian eyes. J. opt. Soc. Amer. 54, BROWN, K. T. & MURAKAM, M. (1964). A new receptor potential of the monkey retina with no detectable latency. Nature, Lond., 201, BROWN, K. T. & TASAK, K. (1961). Localization of electrical activity in the cat retina by an electrode marking method. J. Physiol. 158, BROWN, K. T. & WATANABE, K. (1962 a). solation and identification of a receptor potential from the pure cone fovea of the monkey retina. Nature, Lond., 193, BROWN, K. T. & WATANABE, K. (1962 b). Rod receptor potential from the retina of the night monkey. Nature, Lond., 196, BROWN, K. T. & WESEL, T. N. (1959). ntraretinal recording with micropipette electrodes in the intact cat eye. J. Physiol. 149, BROWN, K. T. & WESEL, T. N. (1961 a). Analysis of the intraretinal electroretinogram in the intact cat eye. J. Physiol. 158, BROWN, K. T. & WESEL, T. N. (1961b). Localization of origins of electroretinogram components by intraretinal recording in the intact cat eye. J. Physiol. 158, CONE, R. A. (1963). Quantum relations of the rat electroretinogram. J. gen. Physiol. 46, GRANT, R. (1947). Sensory Mechanisms of the Retina. London: Oxford University Press.

33 The Journal of Phy,siology, Vol. 176, N,o. 3 Plate 1 0L * *o 0 0 0@0 SO8 un t-, U*ma0U m ~ 0%~~~~~~~~~0 0 0 a ~~~~~~4 O > 0 OS 4 u1) _~~~~ t _ 0 o LA Electrode position as percentage of total retinal thickness. G. B. ARDEN AND K. T. BROWN (Facing p. 461)

34 LOCAL ELECTRORETNOGRAM UNDER OL 461 KUFFLER, S. W. (1953). Discharge patterns and functional organization of mammalian retina. J. Neurophysiol. 16, LUDVGH, E. & MCCARTHY, E. F. (1938). Absorption of visible light by the refractive media of the human eye. Arch. Ophthal., N. Y., 20, MARRoTT, F. H. C., MORRS, VALERE B. & PLRENNE, M. H. (1959). The absolute visual threshold recorded from the lateral geniculate body of the cat. J. Physiol. 146, MOTOKAWA, K., OxAWA, T., TASAX, K. & OGAWA, T. (1959). The spatial distribution of electric responses to focal illumination of the carp's retina. Tohoku J. exp. Med. 70, MOTOKAWA, K., TAMASHTA, E. & OGAWA, T. (1961). The physiological basis of simultaneous contrast in the retina. n The Visual System: Neurophysiology and Psychophysics. Berlin: Springer. OGDEN, T. E. & BROWN, K. T. (1964). ntraretinal responses of the Cynamolgus monkey to electrical stimulation of the optic nerve and retina. J. Neurophysiol. 27, POLYAK, S. (1957). The Vertebrate Visual System. Chicago: University of Chicago Press. TOMTA, T., MURAKAM, M. & HASHMOTO, Y. (1960). On the R membrane in the frog's eye: ts localization, and relation to the retinal action potential. J. gen. Phys-jol. 43, Part 2, TOMTA, T. & TORHAMA, Y. (1956). Further study on the intraretinal action potentials and on the site of ERG generation. Jap. J. Physiol. 6, WALD, G., BROWN, P. K. & GBBONS,. R. (1963). The problem of visual excitation. J. opt. Soc. Amer. 53, WATANABE, K. & TOSAKA, T. (1959). Functional organization of the Cyprinid fish retina as revealed by discriminative responses to spectral illumination. Jap. J. Physiol. 9, WEALE, R. A. (1953). The spectral reflectivity of the cat's tapetum measured in situ. J. Physiol. 119, WEALE, R. A. (1958). Retinal summation and human visual thresholds. Nature, Lond., 181, EXPLANATON OF PLATE Amplitude of the b-wave as a function of retinal electrode depth. Five separate electrode withdrawals are plotted on a scale of percentage electrode depth. The b-wave amplitudes were measured from the peak of the b-wave to the peak of the a-wave, but a-wave amplitudes were small under the conditions of these experiments. No background illumination was used; the stimulus duration was 350 msec and the stimulus was repeated every 10 sec. The diameter of the centred stimulus disk was 1140,u. The stimulus intensity was constant in any given depth series, but varied between series from 4-5 to about 5-5 log units below the maximum stimulus intensity.