Comparison of the responses of Necturus retinal ganglion cells to stationary and moving stimuli

COMPARISON OF THE RESPONSES OF NECTURUS RETINAL GANGLION CELLS TO STATIONARY AND MOVING STIMULI’ JOHN R. T~_!I-~LE’ The Rockefeller University. New York, NY 10021, U.S.A. (Receiced 10 July 1976; in revised form 28 December 1976)

Abstract-A detailed ex~ination was made of the responses of Necturus retinal ganglion cells to stationary and moving stimuli, with the aim of assessing the degree of response specificity of on-off cell “motion detectors”. Responses to stationary and moving strmuli were recorded while changing one stimulus parameter (intensity, size or duration) in the same manner for both stationary and moving stimuli, with ail other stimulus parameters fixed. Sustained-on cells were used as controls, since they are not particularly movement sensitive. No pronounced or systematic differences were observed between the ways the responses of on-off cells to stationary and moving stimuli depended on the three stimulus parameters. in no case were the differences in response dependence more pronounced for individual on-off cells than for some sustained-on cell. Therefore, Nrcturus on-off ganglion cells should not be considered “motion detectors”. Key Words--Secrurus;

ganglion ceil; on-off cell; moving stimuli; “motion detectors”.

Beginning with the earliest studies of optic nerve discharge, it has been noted that moving visual stimuli have special properties. Remarking on the decline in firing rate in bundles of optic nerve fibers after the initial burst of response to a flash of light, Adrian and Matthews (1927) concluded “that a movement will be more readily perceived than a steady pattern of light and shade”. Barlow It9531 noted that frog “on-OK” ganglion cells gave a vigorous response to a moving stimulus even when stimulus motion was confined to a region of the receptive field which had uniform sensitivity for both “on” and “off” responses. Maturana, Lettvin, McCulloch and Pitts (1960) described two types of “movement detectors” which have subsequently been studied by others. Some of their class-2 neurons were “directionally selective” movement detectors. and the rest were treated as “pure” movement detectors. Both kinds of “movement detector” seem to deserve this title, because there is something “special” about their response to moving stimuli. The property of directional selectivity only becomes evident when one examines the responses elicited by moving stimuli. In their report of directionally selective ganglion cefls in the rabbit retina, Barlow and Hill (1963) list four defining characteristics of “units which have genuine directional selectivity”. Retinal ganglion cells satisfying most or all of these criteria have since been reported in many different species: pigeon (Maturana and Frenk, 1963), goldfish (Jacobson and Gaze, 1964: Cronly-Dillon, 1964), lizard (Dejours, 1965), cat (Stone and Fabian, 1965). ground squirrel (Michael. 1966, 1968). grey squirret (Cooper

and Robson. 1966). mudpuppy (Wetblin. 1970: Norton, Spekreijse, Wagner and Woibarsht, 1970) and salamander (Griisser-Cornehls and Himstedt, 1973). Retinal ganglion cells which might be called “pure” movement detectors have also been reported in a large number of different animals: several species of frog (Grtisser-Cornehls, Griisser and Bullock, 1963; Finkelstein and Griisser, 1965; Grtisser-Cornehls and Ludke, 1970), lizard (Dejours, 196.5), mudpuppy (Werblin, 1970), toad (Ewert and Hock, 1972) and salamander (Griisser-Cornehls and Himstedt, 1973). In some of these animals there has been fairfy extensive investigation of the quantitative dependence of ganglion cell response on a variety of stimulus parameters (Griisser, Griisser-Comehls, Finkelstein, Henn, Patutschnik and Butenandt, 1967; Ewert and Hock, 1972; Grtisser-Cornehls and Himstedt, 1973). Notwithstanding these quantitative investigations, the a~ve-mentioned reports only include informal definitions of a “pure” movement detector. The usual comments are simply that the class of cell under study responded much better to moving stimuli than to flashed, stationary stimuli, or that the cells did not respond to changes in the level of general illumination. This study consisted of a detailed examination of the response of Necturus retinal ganglion cells to stationary and moving stimuli, with the aim of assessing the degree of stimulus specificity of the responses of “movement detectors”. Both “pure” and “directionally selective” movement detectors are found in Necrmrcs, so both might be studied in the same anima1 and compared to one another. The relatively extensive knowledge of Necrurus retinal anatomy (Dowling and Werblin, 1969) and the response properties of Ne~&~r~spre-gan~ionic retinal neurons (Werblin and Dowling, 1969; Werblin, 1970, 1971, 1972, 1974; Norton er n/., 1970; Fain and Dowling, 1973: Nelson. 1973: Normann and Werblin. 1974:

’ This work formed part of a dissertation submitted to the faculty of The Rockefeller University in partial- fulfiltment of the requirements for a Ph.D. degree. * Present address: Center for Visual Science. University of Rochester. Rochester. NY 14627. U.S.A. 777

__

tY

JOHS R. TCTTLE

Werblin and Copenhagen. 197-I; Burkhardt. 1974: Fain. 19753 should permit modeling of the retinal network producing the observed stimulus specificity. Finally. the large size of Srcturus retinal neurons xould facilitate further investigation by intracellular recording of the mechanisms great speciticitq-. LI-\TERL4LS

underlying

any observed

tion spectra of Lltamm .X-based photopigments (Bridges. 196’1 ulth peak wa\elengrhs of 525 and 575 nm. the known photopigments

rn ?;rcr~[is

rods

and

cones.

respectivel) 19-Z). The maximum intcnsit! of each beam produced the .ytiL,ftlrlis rod equivalent of 2.9 * IO” quanta cm’-ssc at 25 nm and the .Vrcr~~~.s cone equivalent of 5.2 x 1O’j quanta cm?-set at 5’5 nm. This maximum intensity was taken as 0.0 log relative intensity.

(Brown. Gibbons and V+‘aid. 1963; Llebman.

.A‘UD METHODS

Bioloyicrri prrpwCrrion The subjects of these experiments \vsr2 adult mudpuppies. Secrurus r~crtlos~ts. from 20 to 30 cm in length. shipped air freight from Minnesota. The animals were kept at 16’C in aerated. refrigerated 25 gal aquaria filled with dechlorinated tap ivater, and maintained in a cycle of 11 hr of light. IZ hr of darkness. Each eye was removed from the head, placed on a small piece of black construction paper thoroughly moistened with amphibian Ringer’s. and the anterior media of the eye (cornea. iris and lens) removed. The vitreous humor was absorbed as thoroughlv as possible with strim of filter paper. and the eyecup p&ioneh in a retinal chamber filled with washed. ignited sea sand thoroughly moistened with amphibian Ringer’s, Within the chamber one duct system directed a steady Row of moist oxygen at the eyecup. while another circulared cooling water from a Lauda K2.!R cooler adjusted so that the temperature of the sand was 16’C. The dissection was done under normal room light. Opricnl stimdator The optical stimulator was a two-channel projection system of fairly comentional design. with the exception of two ralvanometer-controlled mirrors. Beyond the projection iens. the light beam was reflected &to the eyecup by two first-surface mirrors mounted on the shafts of optical scanning galvanometers (G-306 and G-310, General Scanning Co.). The axes of rotation of the two galvanometers were perpendicular. providing motion of the image on the eyecup in two orthogonal directions. The galvanometers had a built-in tangent correction, so that image position on the eyecup was linear with rhe input signal ;o rhe galvanome;er driving amplifiers (AX-100. General Scanning Co.). Quartz-halogfn bulbs were used as light sources. Two counter-rotatmg circular neutral-density wedges (Type M Carbon. Kodak) provided a range of adjustment of intensity in excess of 5 log units. An i.r.absorbing heat glass placed before the first lens prevented these potentially damaging rays from reaching the eyecup. A lens on the unused side of the beam-combining cube formed the observation beam used to align the different stimulus transparencies with the center of the optical path. The two stimulator beams were adjusted to obtain equal brightnesses in the focal plane. and then individually calibrated using a silicon photocell radiometer (Optometer 4OA. United Detector Tschnolo=zy). For each beam. the total energy received by the photocell was measured. Then the enerev received was measured while each of 10 narrow band-pa;; interference filters with transmission maxima spaced evenly between 400 and 700 nm was placed in the optical path. These measurements were converted to a received-energy spectrum by (1) correcting all measurements for the fractional area of the photocell illuminated. (2) correcting the reading for each interference filter for its known peak transmission. and (3) scaling-the spectral distribution to agree with the measured iota1 energy received. No correction was made for the slightly different transmission band widths of the various interference filters. since these differences would have had only a slight efTect on the result. The received-energy spectra were used to calculate the visually-effective stimulus by converting them to quanta/cm’-set and convolving them aith the absorp-

The position of each galvanometcr-mounted mirror could be adjusted separately. thus providing independent control of stimulus position along each of the two orthogonal ases. The motion controller pro\ided for stimulus motion along a straight line of continuously variable angular orientation. An external input signal (a linear ramp provided by a function generator) controlled the position of rhe stimulus along the line. The partern of possible motions thus generated resembled the spokes of a wheel. The motion and position signals were summed. making the point set by the position controls the center of rotation of the motion line. i.e. the “hub” of the “wheel”.

Recordings from single retinal ganglion cells were made at rhe bitread surface of the retina with microrlecrrodes fashioned from Z-3 cm lengths of 25 /cm platinum-iridium uirc Hith an insulating layer of Tetlon (10IrlT. Nedwire Corp.. Mount Vernon. N.Y.). One end of the it-ire was soldered to the end of a short brass rod. and the free end cut perpendicular to its length. A silver-silver chloride indifferent electrode was placed within rhe Ringer’s-moistencd sand of the retinal chamber. The potential difference between the recording and indiffersnt electrodes was amplified by a small battery-po\vered. capacitatively-coupled preamplifier (gain of 1000) and then an oscilloscope (502 or 502A. Tektronix. Inc.). The oscilloscope vertical amplitier output was the input to a level detector which produced a uniform output pulse whenever the input exceeded an adjustable threshold level. These pulses were fed into a computer, and also were counted to provide the experimenter with feedback during the experiment. The data collected in this study were recorded on-line using a Control Data Corporation 160-X digital computer. During each stimulus presentation the computer recorded the time of occurrence of each ganglion cell impulse. the times of opening or closing of the shutters, the settings of the neutral density wedges, the stimulus transparency in use, and a continuous record of the stimulus position. Between stimulus presentations the data were written on digltal magnetic tape for later computer-aided analysis. .A digital programmer controlled the repetition rate and timing of stimulus presentations, initiated computer rscoriing of data. and gated the pulse counters used to probide feedback to rhe experimenter. E.~perime~xnl procedwe These experiments were aimed at determining whether or not .Vect~us on-elf retinal ganglion cells were “movement detectors”. i.e. whether or not they were driven bT some mechanism responding specifically fo moving srirnuh. The experimental approach was an examination of the form of the dependence of responses to stationary and moving stimuli on three stimulus parameters-intensity. area and duration. If some motion-specific response mechanism were driviny the on-off gangIion cells. then it should be revealed as a ditTerence in the form of rhe dependence of the response to stationary, and moving stimuli on one or more of the selected stimulus parameters. If the response to stationary and moving stimuli should display rhe same dependence on all three stimulus parameters.

Responses to stationary and moving stimuli

then one would conclude that the same mechanisms were generating the responses to both stationary and moving stimuli and that no motion-specific response mechanism was present. The absence of such a mechanism would mean that ~Vecturuson-off ganglion cells could not be considered “movement detectors”. At the same time. these experiments examined the relative importance of intensity, area. duration and motion as parameters in determining the response of iVecrrmls onoff ganglion cells. If stimulus motion were the most important determinant of on-off cell response. then they should be considered “movement sensitive” ganglion cells even if they lack the specificity of “movement detectors”. However, if stimulus intensity. area. and duration all have as much effect on on-off cells’ responses as whether or not the stimulus is moving, then there would be no reason to consider Necturus on-off ganglion cells particularly specialized for the detection of moving stimuli. The dependence of ganglion cell response upon each stimulus parameter was examined in the same manner. The responses of single ganglion cells to flashed. stationary stimuli and to moving stimuli were recorded while one of the parameters was varied in the same manner for both stationary and moving stimuli. The other stimulus parameters were set at arbitrary values and held constant. The response measure used in all of these comparisons was the average total number of impulses elicited by several stimulus presentations. The flashed, stationary stimuli were presented between five and ten times, and the moving stimuli were presented one to three times in each of eight directions of motion (see “Directionality of responses to moving stimuli”) for each value of the variable stimulus parameter. The ordering of presentation of the directions minimized sequential adaptation effects. For each value of the variable stimulus parameter, the flashed stimulus was presented both immediately before and immediately after the series of moving stimulus presentations. so that any slow changes in each cell’s sensitivity would affect the responses to stationary and moving stimuli similarly. Similar experiments performed on sustained ganglion cells, primarily sustained-on cells, served as controls. Since there was no reason to believe that sustained ganglion cells were driven by any motion-specific response mechanism, any differences in response variation for stationary and moving stimuli were presumed to be due to problems in experimental design and/or the inherent variability of Necm-usganglion cell responses. If the differences in response variation of on-off cells were no more pronounced than those of sustained cells. then the absence of a motion(a)

AI lilIllu

specific response mechanism in on-off ganglion cells would be demonstrated. After both eyecups from a .Vecrurus had been prepared and placed in the retinal chamber. they were examined with a dissecting microscope and the better of the two was chosen for use. The retinal surface of the chosen eyecup was visually positioned in the stimulator’s plane of focus. Single units were isolated using a full-field test flash (400 msec duration, repeated every 4 set) with a continuous full-field background. Once a single unit was isolated. the center of its receptive Eeld was located using a 100 pm wide line extending across the entire retina in Erst the X-axis direction and then the I-axis. The position of the line was stepped along the axis perpendicular to its extent, and the response to a flash at each position noted. On a number of cells, the receptive field center determined in this manner was checked by presenting a small spot at a variety of positions about the previously determined Eeld center. The maximum response to this stimulus was always at or very near the line-determined Eeld center, thus confirming the efficacy of the line method. Field-mapping stimuli were presented as Zsec flashes every 20 sec. The 2-set Hash permitted easy, reliable separation of “on” and “off” responses. The 20-set interstimulus interval was a compromise between collecting data as rapidly as possible and spacing the stimuli sufficiently to avoid light adaptation effects (Proenza and Burkhardt, 1973). This same regime was used for all flashed stimuli. except for those used to study the effect of stimulus duration. RESULTS Responses to jlashed

stationary stimuli

Ganglion cells of each of the five response types previously described in Necturus (Werblin and Dowling, 1969; Hartline, 1938) were observed. An example of each type is presented in Fig. 1. Two classes of ganglion cells generated long duration (sustained) responses. beginning either at the onset (a, sustained-on) or the offset (d, sustained-off) of illumination. Three other classes of ganglion cells generated short duration (transient) responses at the onset (b, transient-on), the offset (e, transient-off), or both the onset and offset (c, on-off) of illumination. Sustained-off cells generally were not used during this study. since the long duration of their responses

I /I il~lll

Cd) -

779

-

I I II I

II I

I I

I

III

I

I

I

I

Fig. 1. Necturus retinal ganglion cell response types. (a) Sustained-on cell (530pm spot. no background). (b) Transient-on cell (lol,~+ line X dir.. -4.02 log; bgd. -5.07 log). cell (336pm spot, -5.08 log; bgd. -6.55 log). (d) Sustained-off cell (100pm line Y dir., bgd. -7.03). (e) Transient-off cell (530pm spot, -6.24 log; bgd. -9.04). All responses 2 set flashes. Time marks: 500 msec.

-5.52 log; (c) On-off -6.02 log; elicited by

JOHNR. TU-~TLE

750

reduced the number of stimulus presentations ible. Thus only four of the forty cells included in experiments were sustained-off cells. The other were 18 on-off cells, 13 sustained-on cells. 3 sient-on cells, and Z transient-off cells.

possthese cells tran-

(0)

1111 11111 II I

(b)

l/l/ II111111

Responses to moving stimuli Response types. The moving stimuli used were either circular spots of light or circular spots of darkness-i.e. a pattern of light falling everywhere except within a circular spot. Most cells were studied with spots of light, since they involved fewer complications. The spot of light was usually turned on (shutter opened) while the spot center was positioned 1040 m from the center of the receptive field. Coincident with the spot being turned on. it began a smooth traverse through the cell’s receptive field to a point 104Opm beyond the center of the cell’s receptive field. At this point. the spot was turned off (shutter closed). On this 2080pm traverse the spot passed completely through the cell’s receptive field; it was turned on and turned off while illuminating regions beyond the boundaries of the cell’s receptive field. When spots larger than 530pm diameter were used. the traverse began and finished 2080Atm from the receptive field center. Figure 2.presents the responses of one ganglion cell of each response type (the same cells as in Fig. 1) to a moving spot of light. The response of each cell occurred while the motion of the spot produced a change in illumination similar to that which elicited a response with a flashed stationary stimulus. The sustained-on (a) and the transient-on (b) cells responded as the spot entered their receptive fields. The on-off cell (c) fired impulses more or less continuously while the spot tranversed its receptive field. presum(cl)

Cd)

(e)

H

111~11111

Ill II I

I

I

II III

Fig. 2. Responses of Nrcrurus retinal ganglion cells to moving soots of light (same cells as Fig. 1). (a) Sustained-on log; no bgd.). (b) Transient-on cell (530im spot: -5.S cell (54 urn mot. -3.99 loe: bud. -5.07 loe). (c) On-off cell (84 Cm sdot. - 5.08 log:-bgd: - 6.55 log). id1 Sustainedoff cell (168 pm spot. -6.02 log; bgd. - 7.03 log). (e) Transient-off cell (530fim spot. -6.24 log; bgd. -9.04 log). Diagonal line denotes duration of stimulus motion (8 set). Time mark: 2 sec. Central vertical mark denotes the center of each cell’s receptive field; the time mark roughly corresponds to the receptive field center for each cell. Stimulus velocity: 260 Fm,‘sec.

Cc) 1

,(

I

,

I

I

Fig. 3. Responses of Xrctrrrus retinal ganglion cells to moving spots of darkness. (a) Sustained-on cell (168pm dark spot, - 7.03 log; no bgd.). (b) On-off cell (336pm -6.38 log: bad. -7.03 lee). (c) Sustained-off cell SDOt. (i68 pm dark siot. -5.03 log: b&i. -5.03 log). Diagonal line and time mark as in Fig. 2. Stimulus velocity: 260 jlrn ‘sec.

ably due to the succession of “on” and “off” responses elicited from different parts of its receptive field as the spot passed. The sustained-off(d) and transient-off (e) cells responded as the spot began leaving their receptive fields, causing an effective decrease in illumination. These general characteristics were observed in all cells of each response type. All of the 10 cells studied with dark spots did respond to them. though most only feebly. Figure 3 presents the responses of a sustained-on cell (a). an on-off cell (b), and a sustained-off cell (c) to moving spots of darkness. As with the responses to moving spots of light, the various ceils’ responses to moving spots of darkness occurred when the moving stimulus produced the appropriate change in illumination. Thus sustained-on cells (a) responded while the spot was leaving their receptive field, i.e. while the darklight boundary at the back of the spot entered their receptive field. Conversely, sustained-off cells (c) responded while the dark spot entered their receptive field. On-off cells (b) responded while the dark spot traversed their receptive field. producing a succession of “off” and then “on” transitions at different points within the receptive field. Directionaliry of responsesto moriny stimuli. It has been reported (Werblin. 1970: Norton et al.. 1970) that some but not all on-off cells in .Vecturtrs exhibit directionally selective responses to moving stimuli. All 40 ganglion cells studied in these experiments were examined for directional selectivity by recording the responses to a given stimulus moving in both directions along both X and Y axes and both 45’ diagonals. The average number of impulses elicited by a number of stimulus presentations for each direction of motion was plotted in polar coordinates. The radial coordinate was the average number of impulses elicited; the angular coordinate, the direction of motion. Using this graphical display. the responses of a perfectly non-directional ganglion cell would form the points of a regular octagon. The responses of a directionally selective ganglion celI (such as those in the rabbit retma) would form an oblong figure with one end passing through or very near the origin.

781

Responses to stationary and mov-ing stimuli

(0) /*\. /

x .i \/

l \

\

asymmetry could not be explained on the basis of receptive field asymmetries. One of the requirements for true directional selectivity (Barlow and Hill, 1963) is that the preferred direction be independent of stimulus-background contrast. Since the response pattern for the dark spot was almost a mirror image of that for the light spot. this on-off cell could not be considered directionally selective. Comparison of responses to srati0nar.v and mocing stimuli

Fig. 4. Responses of Necturus retinal ganglion ceils to moving stimuli as a function of the direction of movement. (Average response plotted radially: see text for details.) Sustained-on cells: (a) 84 pm spot of light. (b) 84 pm (dots) and 168 pm (crosses) spots of light. On-off cells: (c) 134 pm (dots) and 268 pm (crosses) spots of light. (d) 168 pm spot of light (dots) and 168 pm spot of darkness (crosses).

None of the 18 on-off cells showed a pattern of response like that expected of a directionally selective cell. However, most of the on-off cells, and many of the other cells did have asymmetric patterns of response. The response patterns of two on-off cells and two sustained-on cells are shown in Fig. 4. The sustained-on cells are included as internal controls, since these non-directional cells were expected to have symmetric response patterns. The sustained-on cell of Fig. 4(a) had the expected nearly octagonal response pattern. The other sustained-on cell’s response pattern was roughly rectangular for an 84pm spot (Fig. 4(b), dots), and more nearly octagonal for a 168 pm spot (Fig. 4(b). crosses). The rectangular response pattern with the small spot was attributable to an inherent asymmetry of the cell’s receptive field as revealed in the field-mapping responses. The second response pattern was more nearly octagonal because the larger spot effectively “averaged out” a large part of the asymmetry. Slight errors in locating the receptive field center along the X and!or Yaxes would have similar effects on the response pattern. since these errors would cause effective field asymmetry. The oval response pattern of the first on-off cell with a 134pm spot (Fig. 4(c), dots) became much less oval with a 268 pm spot (Fig. 4(c), crosses), indicating averaging out of some field asymmetry. All on-off cells with asymmetric response patterns for small spots exhibited this averaging out for larger spots and,/or clear receptive field asymmetries in their fieldmapping responses. with one exception. The response pattern of this on-off cell with a spot of light (Fig. 4(d), dots) was quite asymmetric and might have been considered that of a directionally selective ceK The cell’s response pattern for a dark spot (Fig. 4(d). crosses) ruled out this possibility, even though the

Response dependence on stimulus intensity. The dependence of ganglion cell response to stationary and moving stimuli on stimulus intensity was determined for three sustained cells and nine transient cells, seven of which were on-off cells. Stationary stimuli were presented as two second flashes; moving stimuli, as 2080pm traverses completely through the cell’s receptive field at a rate of 26Opm.sec. All cells were studied with spots of light. The response-intensity functions of a sustained-on cell for stationary and moving stimuli of four different sizes are shown in Fig. 5. In this figure, as in all figures comparing response variations for stationary and moving stimuli, the response functions are plotted together to facilitate comparison. The average responses to flashed. stationary stimuli are plotted as filled circles (dots); the average responses to moving stimuli, as crosses. The straight lines joining points of the same kind w-ere drawn to further facilitate comparison; they have no theoretical significance. The general agreement between the response-intensity functions for stationary and moving stimuli in

oL&‘-7.0

- 6.5

- 6.0

- 5.5 log

relottve

I

-7.0

-6.5

-6.0

-5.5

mtensity

Fig. 5. Comparison of the responses of a ‘Veccurussustained-on ganglion cell to stationary and moving stimuli as a function of stimulus intensity. Shaded circle indicates relative size of stimulus spots; spot sixes: 168 pm, 336 pm, 530 pm and 840 pm. Flashed stimuli, 2 set duration (dots); moving stimuli. 260 pm/set (crosses). (Receptive field dia >84Opm; bgd -7.03 log.).

JOHN R. TCTTLE

3y

I

-60

-5.5

III -53

-6.5

-60

-55

-6.5

-iO

Fig. 6. Comparison of the responses of Xrcrurrts on-off ganglion cells to stationary and moving stimuli as a function of stimulus intensity. Shaded circles indicate relative size of stimulus spots. Flashed stimuli, 2 set duration (dots): moving stimuli. 260 pm/set (crosses). (a) Spot sizes. 131 ,um. 268 Jim and 530 ,um. (Receptive field dia > 530 ,um: no bgd.) (b) Spot sizes. 168 pm. 336pm and 8JOitm. (Receptive field dia ~(2.S40 ,um: bgd. - 7.03 log.) Fig. 5 is striking. Particularly striking is the almost exact agreement of the two functions for the 530pm spot (lower left). For the 165 pm (upper left) and 840itm (lower right) spots the responses to moving stimuli were consistently greater than the responses to stationary stimuli. In both cases, however, the functions for stationary and moving stimuli have the same form or shape. Since the values of the non-variable stimulus parameters were arbitrary and the relative position of the two functions on the response axis could have been adjusted by changes in stimulus duration, the relative position of the two responseintensity functions was considered unimportant. The significant observation was that, in the presumed absence of a motion-specific response mechanism, the response-intensity functions for flashed, stationary stimuli and for moving stimuli had the same shape. Figures 6(a) and (b) present the response-intensity functions of two on-off cells for stationary and moving stimuli, each cell studied with three spot sizes. The data presented in Fig. 6(a) were representative of those obtained on the on-off cells. There often were consistent differences in the magnitude of the responses to stationary stimuli (dots) and moving stimuli (crosses), as was the case for the two larger spot sizes in Fig. 6(a), though the two functions had the same general shape. Figure 6(b) presents the response-intensity functions for the on-off Cell which displayed the greatest difference between the responses to stationary and moving stimuli. Even for this cell,

the curves for stationary and moving stimuli had a similar form for each spot size used. There was no systematic difference in form between the responseintensitv functions for stationary stimuli and moving stimuli*in any ganglion cell studied. Response dependence on stimdus area. The rsponse-intensity functions of Figs. 5 and 6 indicated that ganglion cell responses to stationary and moving stimuli had the same dependence on stimulus area. since the functions had equally similar forms for a variety of spot sizes. This similarity of dependence on stimulus area was further examined by determining the response of ganglion cells to stationary and moving stimuli of increasing size, presented at the same intensity and duration. The average responses of a sustained-on cell to stationary (dots) and moving (crosses) spots of light of various diameters are plotted in Fig. 7(a). While there were differences between the responses to stationary and moving spots of the same size. the responses to both stationary and moving stimuli showed the same general increase with increasing spot size. Response-stimulus diameter data for stationary and moving stimuli are presented in Figs. 7(b) and (c) for two different on-off cells. The data in Fig. 7(b) are for the same cell whose response-intensity functions were plotted in Fig. 6(a). The similarity between the response-stimulus diameter curves for stationary and moving stimuli confirmed that responses to both had the same dependence on stimulus area. Figure 7(c) presents similar data for another on-off cell. The average response to flashed. stationary stimuli was

32r Ia’ /

1’“’

.

o~o~o

/+

r

24

01

(Cl

I 200

I 400 Diameter,

I

I 800

600

I 1000

pm

Fig. 7. Comparison of the responses of Xecturrrs ganglion cells to stationary and moving stimuli as a function of stimulus diameter. Flashed stimuli, Z set duration (dots); moving stimuli (crosses). (a) Sustained-on cell. Moving stimuli. 260pm,‘sec. (Spots. -5.52 log: no bgd.; receptive field dia ca. 810pm.) (b) On-off cell. Moving stimuli. 260,~m/sec. (Spots, -5.75 log: no bgd.; receptive field dia >53Oblm.) (c) On-off cell. Moving stimuli: upper curve 260,umIsec; lower curve 500 pmisec. (Spots, -6.38 log: bgd. - 7.03 log: receptive field dia en. 530 pm.)

Responses

to

stationary

determined for four spot sizes (dots). Three of these four spot sizes were presented as moving stimuli, each at two velocities. The response-stimulus diameter curves for the two velocities demonstrated the change in the relative positions of the curves for stationary and moving stimuli produced by different, fixed values of the stimulus parameters held constant. Despite differences in response magnitude, the two curves for moving stimuli both showed the same general agreement with the shape of the curve for flashed, stationary stimuli. Response dependence on stimuhs duration. Response-duration data for a sustained-on ganglion cell are shown in Fig. 8(a). The curve for stationary stimuli was generated by averaging the responses elicited by several presentations of a spot of light of constant intensity and size at each of a number of flash durations (dots). A similar variation in duration for the same intensity and size moving spot was achieved by varying the rate of motion. In order to plot the averaged responses to stationary and moving stimuli together. it was necessary to convert the various rates of motion to “effective” durations. This was done two different ways. First, a “spot” duration was computed by dividing the diameter of the stimulus spot by the rate of motion (Fig. 8(a), crosses); this assigned the effective duration as the approximate time a given point within the cell’s receptive field was illuminated

+/ . I-

‘5r

a

(a)

f

12

./*

+

: s

9

f

/

3Cr

i

.-

1-2

z 0 c

(bl

.A

“r

a

/-

.b

/-+.Ad.

I

IO

I

20

I 30

I

4.0

Cc) f

+ _+./*\-

3t :./ 8

01/

c

8)

I

1.0 Duration,

I 2.0 set

I

3.0

Fig. 8. Comparison of the responses of Necturus ganglion cells to stationary and moving stimuli as a function of stimulus duration. Responses to flashed, stationary stimuli (dots) plotted at flash duration. Responses to moving stimuli plotted at “spot duration” (crosses) and at “field duration” (circled crosses). (a) Sustained-on cell stimulated with spots of light. (336pm spot, -6.01 log; bgd., -7.03 log; receptive field dia ca. 84Opm.) (b) On-off cell stimulated with spots of light. (336pm spot, -6.28 log; bgd., -7.03 log; receptive field dia ca. 530pm.) (c) On=off cell stimulated with spots of darkness. (168 pm spot, -6.50 log; bgd.. -7.03 log; receptive field dia ca. 53Opm.)

and

783

moving stimuli

by the moving spot. Second. a ‘*field” duration was computed by dividing the diameter of the cell’s receptive field by the rate of motion (Fig. 8(a). circled crosses); this assigned the effective duration as the total time the center of the stimulus spot was within the cell’s receptive field. Responses for the two slowest rates of motion are not plotted at their “field” durations. since these exceeded four seconds. Whether plotted as a function of spot duration (crosses) or field duration (circled crosses), the response-duration curve of this sustained-on cell for moving stimuli has a form quite similar to that of the response-duration curve for the stationary stimulus (dots). Shown in Figs. 8(b) and (c) are response-duration curves for two on-off ceils. The first cell rig. 8(b)] was examined using a spot of light. The averaged responses to the stimulus presented at various rates of motion, plotted at the spot duration defined above (crosses), showed fairly close agreement with the average response to flashed, stationary stimuli of various durations (dots). The agreement between these curves for stationary and moving stimuli was fairly good for all five on-off cells studied with spots of light. One cell was also studied with spots of darkness [Fig. 8(c)]. For the stationary stimulus condition, a 168 pm spot of light was centered in the cell’s receptive field and “flashes of darkness” produced by turning the spot off for a variable amount of time. The moving stimulus was a 168 pm spot of darkness, i.e. a pattern of light which illuminated the entire retina except for a 168pm diameter spot; it was presented at three rates of motion. The average responses to the moving spot of darkness, whether plotted at the spot duration (crosses) or at the field duration (circled crosses), were in relatively good agreement with the responseduration curve for Rashes of darkness. This was a further indication that the responses of on-off ganglion cells to flashed, stationary stimuli and to moving stimuli show the same form of dependence on stimulus duration. Summar) of comparisons. Using the response measure of this study, none of the on-off ganglion cells studied showed any systematic difference in the way responses to stationary and moving stimuli depended on any one or more of the stimulus parameters of intensity, area, and duration. The differences between individual pairs of response-stimulus parameter curves were no more pronounced for onoff cells than for sustained cells. Furthermore, such differences as there were could have been largely eliminated by adjusting the arbitrary values of the non-variable stimulus parameters [cf. Fig. 7(c)]. Thus the experimental results support the conclusion that the responses of all non-directionally selective types of Necturus retinal ganglion cells to flashed, stationary stimuli and to moving stimuli are generated by the same response mechanisms, under the conditions of these experiments. DlSCUSSlON

Properties of ganglion cell receptive fields The properties of the receptive fields of Necturus retinal ganglion cells reported here are in general agreement with those reported previously by other investigators. The responses of ganglion cells to

x

JOHS R. -I L’TTLE

Hashed.stationary stimuli fell into the hve response-type classes (Fig. 1) originally reported by Werblin and Dowling. (1969). Similarly. the responses of these five classes of ganglion cells to moving spots of light (Fig. 2) were the same as those reported by Werblin (1970). But contrary to his conclusion that ganglion cell responses to moving stimuli could not be predicted from their responses to flashing stimuli, it appeared in this research that the responses of the various types of ganglion cells to moving spots of light could be predicted from their responses to flashed. stationary stimuli. The appropriate response of each cell type occurred when the moving spot of light produced a change in the illumination falling within the cell’s receptive field similar to the change(s) of illumination which elicited responses when flashed stimuli were employed. That is. cells responding at the onset of a flash responded as a spot of light entered their receptive field: cells responding at the offset of a flash. as the spot left their receptive field. Responses elicited by flashed, stationary stimuli and moving stimuli were either of long duration (sustained) or of short duration (transient), with responses of each cell having the same character for both types of stimuli. The responses of ~Vecrurus retinal ganglion cells to moving spots of darkness (Fig. 3) also could be predicted from their responses to flashed stimuli. using the principles just outlined. Xone of the 1S on-off ganglion cells examined in these experiments, however. was of the directionally selective type reported in Vecturus by Werblin (1970) and by Norton et a/. (1970). This was presumably due to the small proportion of directionally selective ganglion cells. and/or an electrode-sampling problem. Using smaller-tipped glass micropipettes and tungsten microelectrodes. Karwoski (1975, personal communication) has reported that only 49; of Nectrlrus ganglion cells are directionally selective, satisfying the criteria of Barlow and Hill (1963). Werblin (1970) and Norton er al. (1970) also used smaller-tipped glass micropipettes. The effects of stimulus intensity, area. and velocity on the response of Necrunls retinal ganglion cells to moving stimuli closely parallel thetr effects on responses to flashed, stationary stimuli. This was to be expected from the explanation of the ganglion cell responses to moving stimuli in terms of the changes in illumination they produced and the responses elicited by similar changes in stationary stimuli. The effects of these stimulus parameters in Vrct~~rt~s are qualitatively similar to the effects in frog (Griisser et al.. 1967) toad (Ewert and Hock. 1972) and salamander (Griisser-Cornehls and Himstedt. 1973). Necturus

on-ojf ganglion cells as mocemenr detectors

Werblin’s conclusion that the on-off ganglion cells of the Nectrrrus retina are “motion-detecting ganglion cells” (Fig. 9. Werblin. 1970) apparently was based on a simple, seemingly convincing comparison: on-off ganglion cells responded to a moving stimulus with a prolonged discharge of many impulses, while they responded to a flashed stimulus with short bursts of just a few impulses. Similar comparisons had been considered evidence for movement-detecting ganglion cells in other animals. But it is potentially misleading to conclude that these ganglion cells are .‘movement

detectors on the basis of this comparison for two reasons. First. while the appeal of this comparison is that it is simple and direct. it is in reality not so simple a comparison. The results reported herein demonstrate that the responses elicited by stationary and moving stimuli are strongly influenced by the stimulus intensity, area, duration, and rate of motion. Therefore, the result of such a “simple” comparison will depend almost as much on the particular stimulus parameters chosen as on the properties of the cell involved. For instance. consider the responses of the on-off of Fig. 7(c) to the 168 /cm spot. A two second flash elicited an average of 3.75 impulses. Presented as a moving stimulus. this spot elicited 2.63 impulses at 490 htm%ec and 6.00 impulses at 260 pm,‘sec. Thus. whether the stationary or moving stimulus elicited more impulses depends upon which rate of motion one cares to use for the comparison. Second, the simple fact that a given ganglion cell responds “best” to a given stimulus does not seem sufficient reason to label that cell a “detector” of that stimulus. The word “detector” implies great specificity of response. a specificity infrequently displayed by retinal ganglion cells. Allowing that “movementdetecting” ganglion cells might respond to stationary stimuli, one would expect them to have some pronounced, motion-specific property-hopefully. some response mechanism excited only by moving stimuli. If such a motion-specific response mechanism were present. it would almost undoubtedly reveal itself as a difference in the way the responses to stationary and moving stimuli depend on the various stimulus parameters. The experiments reported here have shown that there are no significant or systematic differences between the dependence of the responses to stationary and moving stimuli on stimulus intensity, area. and duration for either the sustained-on ganglion cells or the on-off ganglion cells of the ~Vrctru-us retina. Thus Srcturus on-off retinal ganglion cells do not appear to possess a motion-specific response mechanism. Furthermore. the results of these experiments indicate that Necrtrrrls on-off cells are not even particularly “movement sensitive”. As illustrated above, the stimulus parameters of intensity. area. and duration were at least as important in determining the response of on-off cells as w-hether or not the stimulus was moving. This might or might not be the case under different experimental conditions or using a different response measure. That it was true under these conditions using this response measure argues strongly against A’rcrm[s on-off retinal ganglion cells being specialized “movement sensitive” ganglion cells. There are three reservations about the generality of the conclusion that non-directionally selective ganglion cells do not have a motion-specific response mechanism. One reservation concerns the search procedure. Using a full-field flash may have excluded one or more response-classes of ganglion cells, possibly including a class of movement-specific cells which do not respond to Rashes. Since no other Xecrtrrrrs ganglion cell study has specified a search procedure (Werblin and Dowling. 1969: Werblin. 1970: Norton et trl.. 1970; Burkhardt. 1974: Werblin and Copenhagen. 1974: Fain. 1975; Copenhagen. 1975). no

785

Responses to stationary and moving stimuli assessment of search procedure effect is possible. A second reservation concerns the response measure, total number of impulses. Another response measure, such as the amplitude of a PSTH. might have revealed differences not obvious in a total impulse count. Presenting moving stimuli with eight directions of movement for each set of stimulus parameters prechided collecting enough data for each direction-parameter combination to generate a reasonable PSTH, so this and related response measures could not be used. The third reservation concerns the low light intensities used in these experiments, which were definitely in the scotopic (rod) range. Considering the rod and cone properties reported by Normann and Werblin (1971). the steady background probably had a weak adapting effect on the rods and virtually no effect on the cones. The majority of the stimuli used were below cone threshold (Fain and Dowling, 1973; Normann and Werblin, 1974). Moreover, all of the stimuli were below the level of rod saturation, and resulted in about twice as many quanta absorbed per rod as per cone. Thus the ganglion cell responses were in all probability rod-controlled. While the truly relevant thresholds are those for rod and cone input to the ganglion cells, this discussion is in terms of receptor thresholds since this is the only information available. It is possible that transient ganglion cells might possess a motion-specific response mechanism which receives input only from the cones. This is most unlikely. however. since all the transient ganglion cells studied by Fain (1975) had spectral sensitivities indicating both rod and cone inputs. In addition, it would make little functional sense for a photophobic amphibian like Necrurus (Reese, 1906; Pearse, 1910) which prefers a dimly lit habitat (Harris, 1959) to possess a motion specific response mechanism which was only functional at fairly bright levels of ambient illumination. The assertion that Necturus on-off retinal ganglion cells are not “motion detectors” is not an assertion that these cells play no role in the detection of moving stimuli by the visual system as a whole. Indeed, although the extent to which Necmnts detect and react to moving visual stimuli has not been determined. the on-off ganglion cells most likely play a dominant role in such detection, since they appear to be the ganglion cells whose receptive field properties are most appropriate for responding to the series of changes in retinal illumination produced by such stimuli. What is asserted is that the lable “movement detector” is a misleading and inadequate characterization of the receptive field properties of individual Nectilrils on-off ganglion cells. The stimulus parameters of intensity. area. and flash duration or rate of motion affect the responses elicited from on-off cells at least as much as the parameter of whether the stimulus is stationary or moving. Indeed, for the particular conditions employed in these experiments, the moving stimuli generally elicited fewer impulses than the flashed, stationary stimuli (cf. Figs. 7 and 8). Thus an adequate characterization of properties of ~Vecrtrrta on-off ganglion cells must include-a description of the effects of a large number of stimulus parameters on the responses of these cells.

Similar considerations apply to other classes of ganglion cells. and indeed to “specialized” units in all parts of the visual system. The class-names of these cells usually depict or emphasize the most prominent response characteristic or the most effective stimulus configuration. One should not forget, however, that these cells will respond to a variety of stimulus configurations, and that a variety of stimulus parameters will affect the response magnitude. Acknowledgements-This investigation was supported

in part by NIH Training Grant No. GM 1789 from the National Institute of General Medical Science. and by research grants EY 00188 from the National Eye Institute, and GB 36168 from the National Science Foundation. Preparation of this manuscript was supported in part by grant PO1 EY 01319 from the National Eye Institute. I wish to express my appreciation to Drs. Floyd Ratliff, James Gordon and Robert Shapley for their encouragement and discussions during the course of this study.

REFERENCES

Adrian E. D. and Matthews R. (1927) The action of light on the eye-Part I: The discharge of inpulses in the optic nerve and its relation to the electric changes in the retina. J. Physiol.. Land. 63. 378-414. Barlow H. B. (1953) Summation and inhibition in the frog’s retina. 1. Physiol.. Lond. 119. 69-88. Barlow H. B. and Hill R. M. (1963) Selective sensitivity to direction of movement in ganglion cells of the rabbit retina. Science 139. 412-414. Bridges C. D. B. (1967) Spectroscopic properties of porphyropsins. Vision Res. 7. 349-369. Brown P. K., Gibbons I. R. and Wald G. (1963) The visual cells and visual pigment of the mudpuppy. Necrilrus. J. CeI/ Biol. 19, 79-106. Burkhardt D. A. (1974) Sensitization and center-surround antagonism in Nectrrrrts retina. J. Physiol., Land. 236. 593-610.

Cooper G. F. and Robson J. G. (1966) Directionally selective movement detectors in the retina of the grey squirrel. J. PhysioL, Lond. 186, 116P-117P. Copenhagen D. R. (1975) Time course of threshold elevation in on-off ganglion cells of lVectttrrts retina: effects of lateral interactions. Vision Res. IS, 573-581. Cronly-Dillon J. R. (1964) Units sensitive to the direction of movement in goldfish optic tectum. Xorrrrr, Lond. 203, 214-215. Dejours S. F. (1965) Receptive fields of optic tract fibers in lizards (Sceloporus ssp.). Doctoral dissertation, Harvard University. Cambridge, Mass. Dowling J. E. and Werblin F. S. (1969) Organization of retina of the mudpuppy. Xecrrrrus maculosus. I. Synaptic structure. J. Neurophpiol. 32, 315-338. Ewert J. P. and Hock F. (1972) Movement-sensitive neurones in the toad’s retina. Expf Brain Rex 16, 41-59. Fain G. L. (1975) Interactions of rod and cone signals in the mudpuuov . .._ retina. J. Phvsiol.. Lond. 252. 735-769. Fain G. L. and Dowling J. E. (1973) Intracellular recordings from single rods and cones in the mudpuppy retina. Science 180, 1178-1181. Finkelstein D. and Grtisser O.-J. (1965) Frog retina: detection of movement. Science 150, 103C&1051. Griisser 0. J., Griisser-Cornehls U.. Henn V.. Patutschnik M. and Butenandt E. (1967) A quantitative analysis of movement detecting neurons in the frog’s retina. Pfliigers Arch. ges. Phpiol.

293, 100-106.

Grtisser-Cornehls U., Griisser O.-J. and Bullock T. H. (1963) Unit responses in the frog’s tectum to moving and nonmoving visual stimuli. Science 141, 820-822.

Y$f,

JOHS

R.

Grusser-Cornehls L’. and Himstedt W. (1973) Responses of retinal and tectal neurons of the salamander (Salamandra xdumandra L.) to moving visual stimuli. Bruin Behac. Erol. 7, l-l5-168. Grtisser-Cornehls ti. and Ludcke &I. (1970) Comparative neurophysiological experiments on signal processing in the retina of anurics. I??iigers .4rch. ges. Physiol. 319. RIJS. Harris J. P.. Jr. (1959) The natural history of .Vrcrurus: I. Habitats and habits. Field anti Lab. 27, 11-20. Hartline H. K. (1938) The response of single opttc nerve fibers of the vertebrate to the illumination of the retina. Itm. J. PhJsiol. 121, 3oo-j15. Jacobson M. and Gaze R. M. (1961) Types of visual response from single units in the optic-t&turn and optic nerve of the ooldfish. 0. JI eso. Phvsiol. 49. 199-209. Liebman P. A. (1972) Mi~rospectrophdtometry of photoreceptors. In Handbook ofSensoryPh,vsiologJ. Vol. VII#i (Edited by Dartnall H. J. A.). pp. 481-528. Springer. Berlin. hlaturana H. R. and Frenk S. (1963) Directional movement and horizontal edge detectors in the pigeon retma. Science IQ. 977-979. Uaturana H. R., Lettvin J. Y.. McCulloch W. S. and Pitts W. H. (1960) Anatomy and physiology of vision in the frog. J. gen. Physiol. 43 (Suppl. 2). 129-175. Michael C. R. (1966) Receptive fields of directionally selective units in the optic nerve of the ground squirrel. Science 152, 1092-1094. hlichael C. R. (1968) Receptive fields of single optic nerve fibers in a mammal with an all-cone retina. II: Directionally selective units. J. Neurophysiol. 31, X7-267. Nelson R. (1973) A comparison of electrical properties of neurons in .Vecturrrs retina. J. Zieurophysiol. 36, 519-535.

TCTTLE

Normann R. A. and Werblin F. S. I 1974) Control of retinal sensitivity. I: Light and dark adaptation of vertebrate rods and cones. 2. +VI. Ph!sio[. 63; 3761. Sorton A. L.. Spekreijse H.. Wagner H. G. and Wolbarsht 41. L. (1970) Responses to directional stimuli in retinal preganglionic units. J. Physiol.. Land. 206, 93-107. Pears; A_ S. (1910) The reactions of amphibians to light. Proc. .-Im. .4cud. .4rts 9: Sci. -5, 160-208. Proenza L. bl. and Burkhardt D. A. (1973) Proximal negative response and retinal sensitivity in the mudpuppy, Srctrrrus mucnlosus. J. .Vetrroph.vsiol. 36, 502-518. Reese A. M. (1906) Observations on the reactions of Cryptobrunchus and .Vecrurus to light and heat. Biol. Bull. 11. 93-99. Stone J. and Fabian iL1.(1966) Specialized receptive fields of the cat’s retina. Science 152, 1277-1279. %‘erblin F. S. (1970) Response of retinal cells to moving spots: intracellular recording in Xecttrrus muculostrs. J. .Vewoph_vsiol. 33, 342-350.

Werblin F. S. (1971) Adaptation in a vertebrate retina: intracellular recording in Mectrws muculosrrs. J. Netrrophysiol. 31, 228-241. 1Verblin F. S. (1972) Lateral interactions at inner plexiform layer of vertebrate retina: antagonistic response to change. Science 175, looa-1010. _ Werblin F. S. (1974) Control of retinal sensitivitv. II. Lateral interactions at the outer plexiform layer. J: gen. Physiol. 63, 62-87.

Werblin F. S. and Copenhagen D. R. (1974) Control of retinal sensitivity. III. Lateral interactions at the inner plexiform layer. J. gee”, Phxsiol. 63, K-110. Werblin F. S. and Dowling J. E. (1969) Organization of the retina of the mudpuppy. Srcrrrrtts mncrrlosus. II. Intracellular recording. J. Netrrophysiol. 32. 339-355.