SUPPRESSION OF MAINTAINED ACTIVITY OF RETINAL GANGLION CELLS ROBERT P. SCOBEY’, LEO M. CHALUPA? and ROBERT W. HAMMON’ Laboratory of Neurophysiology. Department of Behavioral Biology. School of Medicine.
University of California.
Davis..
California 95616. U.S.A.
(Receiced 28 April 1978) Abstract-Recordings were made of the changes in unit discharge rates of on-center and off-center retinal ganglion cells when the luminance of a spot in the receptive field was increased or decreased from a background level. A prolonged suppression of maintained activity to background illumination was observed in both transient and sustained cells of cat and monkey. On-units decreased in rate when the luminance of the spot was reduced: off-units decreased in rate when the luminance of the spot was increased. Long duration silence of maintained activity was dependent upon spot size, position and luminance. The most effective spots covered only the receptive field center. With low contrast levels a response time course was found which did not follow a recovery back to maintained rate along a single exponential function. Increasing contrast silenced a progressively greater proportion of retinal ganglion cells. Key Words-suppression
of discharge: retinal ganglion cells; transient; sustained: cat: primate.
INTRODUCTION
IMETHODS
Since the introduction of single-cell recording techniques to the study of visual processes (Hartline, 1938), most experiments have been concerned with those stimulus parameters which induce an increase in the rate of action potentials. Thus, retinal ganglion cells were classified a on-center or off-center (Ku!?ler. 1953) based on whether a particular cell discharges at the onset or offset of a light-spot flashed in the center of the receptive field. He noted that the spontaneous activity of off-center ganglion cells is suppressed by a luminance increment. The discharges of on-center cells may be decreased by a luminance decrement of a spot positioned on the receptive field (DeValois, Jacobs and Jones, 1962; Kriiger and Fischer, 1975; Barlow and Levick, 1976). In the present study stimulus characteristics which increase, or decrease discharges of retinal ganglion cells were investigated in cat and monkey. In particular, we were interested in determining whether the classification of ganglion cells into transient and sustained types observed with stimuii which increase discharge rates (Cleland, Dubin and Levick. 1971; Cleland, Levick and Sanderson, 1973), could also be obtained with stimuli which diminish spontaneous activity levels. Transient cells in both cat and monkey have been reported to adapt rapidly to stimuli which decrease their spontaneous activity (Cleland er al., 1973, Fig. 8; Marrocco, 1976). but this phenomenon has not been studied in detail. ‘Department of Behavioral Biology. School of Medicine, University of California, Davis, California 95616, U.S.A. ’ Department of Psychology, University of California, Davis. California 95616, U.S.A. ’ Present Address: Stanford Research Institute, 333 Ravenswood, Menlo Park, California, U.S.A. 451
Preparation
Recordings were made from ten cats and I6 monkeys. Following anesthesia (halothane for cats, pentobarbital for monkeys), tracheal and venous cannulas were inserted and the animal was aligned in a stereotaxic instrument. A metal bar system was fastened to the skull with screws and acrylic; this served to hold the animal’s head in stereotaxic coordinates after removal of ear and eye bars. In four of the cat and all of the monkey experiments, responses of ganglion cells were recorded by means of an intraocular approach (Scobey and Horowitz. 1976). In the other six cat experiments microelectrode insertions were made through an opening in the skull over the optic tract. All wounds and pressure points were infiltrated with long-lasting local anesthetic (Nupercaine). ARer surgery, halothane was reduced to between 0 and OS’?, and a mixture- of gallamine triethiodide (5-15 mg/kg. h) and d-tubocurarine (O.l-0.3mg/kg. h) was continuously infused to reduce eye movements. Reduction of the halothane was required only when cardiac arythmia was noted in the electrocardiogram. Artificial respiration was maintained with a 70% N,O and 300,; O2 mixture. A constant volume pump was adjusted to provide a 3.5-57; end-tidal CO1 concentration as determined with an infrared CO2 monitor. The animal’s temperature was monitored by a rectal thermistor which electronically controlled a heating pad to maintain body temperature at 37.5’C. A drop of neosynephrine (10%) and atropine sulfate (1%) retracted the cat’s nictitating membrane, paralyzed accommodation, and dilated the pupils. A 1 D contact lens was fitted to each eye. With cats. the eyes were refracted and optic discs plotted on a tangent screen using the method of Fernald and Chase (1971); with monkeys the procedures described by Scobey and Horowitz (1976) were employed. Electrical activity from axons in the cat tract was recorded with tungsten microelectrodes. Glass micropipettes were used for recording from ganglion cells. Srimularion After isolating the action potentials of a single unit. the
recepttve field was plotted on ~1tangent screen usrng backprojected spots of light as ueII as the edges of hand-held black and white wands. .4 st.mdard background luminance of 8 cd,‘m’. measured with an SE1 phorometrr. was produced with tungsten filament lamps. Units were classified as on-center or off-center. as weil as sustained or transient. Only briskly responding units which proved to be unJmbiguous wtth respect to a sustained or transient classification (Cleland and Levick. 1974) were employed in the quantitative analysis of responses. The terms tonic and phasic 41 be used for monkey units (Debfonasterio and Got%%. 197%. After determining the receptive field and response characteristics of a given unit. the tangent screen was removed and a specially constructed stimulator (Fig. 1) was moved to the site of the receptive field. The front cover screen (D) of the stimulator was illuminated by filtered fiuorescent light (3ed m’t which approxima!ely matched the *‘white” color produced by the PJ sulfide aluminized phosphor of a cathode ray tube (A). Light from the cathode ray tube (CRT) first passed through two holes in a box (B) painted black inside. and then through a rotating disc (C) which permitted selection of various spot sizes. The ratio m area between adjacent holes was 2 to 1. The disc was painted the same white color as the cover screen. A mylar diffusing screen was attached to the rear of the box. and the box was sealed to the stimulator. When a normal 60 cycle interleaved television raster was initially tried. the stimulator continuously excited the cat transient units for the duration of the spot presentation. This artifact was eliminated by employing a
1-rrricol sautooth sueep whtch hdd a -I mjec cycle dura.. and a sgnchronized horrront.tl sweep ;rovtded 2 stable non-interleaved pattern with ;L 50 tis~ horizontal sweep duration. The mylar dil%_tsin$ screen e(:mmated all stationary pattern detail. The luminance of the light from the CRT was Adjusted to match the screen within 0.1 102 units. When the light from the CRT increased. the stimsiltor produced a luminous spot relative to background. and when the light from the CRT decreased. the sput appeared darker than the backgruund. With the CRT OK. the spot NSS 1.200 of the background intensity. .1 cdiibrstlotl curve ‘ms taken from the CRT screen before and aim: edch experiment to contirm stability. The fraction of th? ipot intensity due to background was added to the photometrically measured luminance of the CRT screen to CaItbrate dark SPWS (El. Retinal illumination was reduced tn some experiments b:, the addition of shielded neutral density filters in front of the eye. A subjective estimate of cat neural threshold responses was made for each spot size by manually changing the intensity of a flash (0.5 set duration. one per second rate). The voltage to the Z ;LXIS was adjusted until a just-noticeable increase in the rate of the audible neural pulse train was evident in approximately 50”” of the trials. This “auditory threshold method” was used LO measure sensttivity of the cat‘s units to each of 11 available stimulus sizes. The “equivalent center diameter” ua_s estimated by the method descrtbed b> Cleland. Le\ick and Sanderson I 19731. In order IO decrease the ttme necessary to estimate
A
B
Fig. 1. Stimulator. The stimulus apparatus used to generate dark and light spots relative to background consisted of four parts as shown in the exploded diagram. A raster was produced on a cathode ray tube (A). A box painted black on the inside (B) was sealed to the CRT. The light from the CRT passed through a diffusing screen at the CRT (not shown) and through the box. A series of holes could be selected by means of a rotating disc (C) which was painted the same color as the tangent screen (LX The raster luminance was adjusted to the luminance of C and D with the D-A voltage equal to zero (E). The spot luminance could be adjusted between 1 iog units less than background to 1 log unit above the background of 3.0cd.m’ by voltage control of the raster luminance. The calibration curve (El was constructed by adding the reflected light to the light from the CRT alone.
Maintained suppression the size of the receptive tieid of monkey units. the hole size which gave the largest response to small increments or decrements was taken as the center size. For quantitative testing, a minicomputer was employed to control the rate. duration, iterations and intensity of the stimuli. Data for as many as 16 histograms were coilected by randomty intermixing the stimufus triaf values. Each data set was based on eight stimulus presentations. Increments and decrements of spot intensity were controlled by a IO-bit D-A converter over the range of the stimulator. Data were collected as the intervals between sequential action potentials and stimulus events. The inter* vals were measured with a l/8 msec resolution and stored on digital tape for subsequent analysis. During the experiment. the computer presented displays ii~ustrating the number of action potentiats evoked by the onset and offset of a stimulus as a function of the controlling D-A value. Upon comptetion of one set of histograms. another set of intensities was chosen in order to generate responses for which a continuous function could be estimated. Typicatty 10 histograms covering the range of intensities from about 2 log units befow background to 1 log unit above background were computed for a given spot size. Anal_Uis Initially. responses to single stimulus presentations as well as histograms were inspected with either computergenerated oscillascope displays or with paper records produced by a digital plotter. Two measures of response were employed: flf the sum of action potentials for a given stimulus event. and (2) the peak frequency catcutated as the reciprocal of the average of the four shortest successive intervals between action potentials.
maintained level of activity was primarify dependent upon stimulus parameters. The data to be presented are from recordings of 37 cat units (11 on-sustained. 8 on-transient, 9 offsustained, and 9 off-transient) and 33 monkey units (5 on-tonic+ 2 on-phasic. it off-tonic, and 15 offphasic). From these units t342 histograms at difTerent intensitie relative to background were constructed. Briefly recorded units were characterized. but those from which insufficient quantitative data were not obtained to construct stimulus-response curves were excluded. However, the incomplete study of such units supported the conclusions reported. Two groups of histograms, one group from an oncenter unit with only a transient response. and one group from an on-center unit with a sustained component, are illustrated in Fig. 2. The histograms are separated from one another by a distance proportional to the contrast of the stimulus spot to background (value of 0). Responses to stimuh greater than can be displayed on the intensity axis and between the values illustrated were omitted for clarity, These show a continuum between and beyond those illustrated. A conspicuous feature of the illustration is that small decrements in spot luminance with respect to background (negative values) are effective in decreasing the “spontaneous” or “*maintame< activity. Small vaiues of contrast transiently suppressed activity; modest values (0.5 log units) silenced the units for 4sec. The offset of the stimulus then evoked a char&cteristic and strong
RESULTS
The major aim of this study was to investigate the responses of retinal ganglion cells to stimuli which decreased maintains activity. Xn particular, we were interested in the time course of suppression and the extent to which it could be related to the sustained or transient response category (Cleland, Dubin and Levick, 1971). Decrements of light with on-center neurons. or increments with off-center neurons produced a suppression of maintained activity. The time course of responses to stimuli which decreased the
453
burst.
The time course of the response decrement was dependent upon the position and size of the stimulus in the receptive field. For the units shown in Fig, 2 the spot was equal to the equivalent center diameter (defined in Methods). and positioned to cover the center of the receptive field. These conditions had to be fulfilled in order to evoke the maximum number of unit discharges and also were often necessary to produce a sustained response decrement. Under these conditions, and with appropriate contrast. response decrements lasting for the duration of the stimulus were observed in all cat and monkey units recorded.
A
LOG
r/r,
SECONDS
fig. 2. The figure was constructed by displacement of histograms a distance proportional to the togarithm of the ratio between the background and stimulus intensity. Two cat on center units are itiustrated. a transient unit (A) and a sustained unit (IS).The spot size was equal to the equivalent center diameter and centered on the receptive field. With decrements, the maintained activity to background alone was diminished and silenced with increasing contrast. The offset of the decrement evoked a large response. Calibration mark for the ordinate is 100 spikes per see. The histograms were smoothed by convolution of data with a 3 bin rectangular function.
ROBERT
454
P.
S~OBEY.
LEO M.
CHALL’PA and
In order to provide an adequate estimate of the population response. we systematically varied luminance of spots within the center region while keeping background. spot size, and position constant. These parameters were manipuIated in only a small number of units.
Using a spot centered in the receptive fieId which was equal to the equivalent center diameter, 47 units could be silenced by appropriate contrast for the duration of the stimulus (as shown in Fig. 2). Seven units had an obvious sustained decrease in the tiring rate but would not silence. The procedure used to estimate the minimum contrast which would silence each of the 47 units for the duration of the stimulus was standardized as fotlotvs. A stimulus-repose function using a spot covering the center of the receptive field was constructed by plotting either the peak frequency or the number of action potentials during the stimulus as a function of contrast {Fig. 3). A straight line best fitting these data at background (1 = 0) was drawn by eye and the intersection of this line with the abscissa was recorded. The data from cat and monkey, as well as the data from on and off. sustained (tonic) and transient {phasic), were first evaluated separately. Differences were thought not to be significant and the data were combined to give a distribution for onunits and a distribution for off-units (Fig. 3B). The median contrast necessary for silencing the on and
j-
1.0 I”
8
KCXERT
W.
HALW)~
the off System W;LS -0.11 and +0.17 log units. respectively. Not all units were silenced at the same contrast. AS shown in Fig. 39 there is a range over which increasing contrast silences a progressively increasing proportion of retinal ganglion ceils. It is noteworthy that the reSu)tS illustrated in Figs 2 and 3 were obtained with spot sizes which were equal to the equivalent center diameter. Spots which were not centered on the receptive field and covered only a portion of the center as well as the antagonistic surround would not silence the unit. Additionally, for many units spot size was a critical stimuIus dimension. Spots which were substantially larger and centered on the receptive field might not silence the units. Likewise. spots which were substantially smaller were ineffective (Fig. 4). Two sets of 4 histograms were selected from the 26 available to illustrate the responses to increments and decrements of two different spot sizes. In Fig. 4X. responses of an on-unit to increments of a spot equal to the equivalent center diameter are illustrated. For comparison. an equivalent response \vaS selected from the set of histograms that were obtained with a spot only 1;9 of the area (Fig. 4B). Only a small increase in spot luminance (0.03 log, units) was necessary to offset the reduction of spot size. In terms of suppression of maintained activity. a stimulus decrement of 0.78 log units at a spot size equal to the equivalent center diameter was capable of silencing the maintained activity for 4sec. At the same contrast the smaller spot (Fig. 48) viefded maintained partial suppression.
t
OFF
Fig. 3. Synergistic range of on and off units. In (A). a stimulus-response lunction of an an-transient unit illustrates the method used to estimate the minimum increment or decrement necessary to achieve suppression of unit activity. The size and position of the stimulus was chosen for silencing units with least contrast. Each point represents the peak response frequency obtained for a single stimulus presentation at a given intensity. A best fit straight line was drawn to approximate the data near background; the intersection with the abscissa was used as a data point as indicated by the dotted line. in (B), a cumulative histogram of on- and off-center units derived by this method illustrated the range of luminance necessary for silence. The duration of the stimulus used in the cat experiments was 4sec while the duration of the stimulus used for the monkey experiments was either I or isec. Data for monkey and cat units were combined. Background was between 0.5 and l.Ocd;m’. The cumulative histogram to the left of zero indicates the decrement necessary to suppress activity in on cells. The curve to the right of zero is the value of the increment necessary to silence off units. The space between the two curves is an estimate of the minimum range over which both the on and off system could transmit information about increments and decrements.
ZIaintained suppression B
A
SPOT DIAMETER 0.8 0~ ARC
SPOT DIAMETER 24’ OF ARC
BACKGROUND ALONE 0
-.7a
1 0
1
I
4
SECONDS
8
;1
;
SECONDS
i
Fig. 4. Sustained decrement of maintained activity. The responses of an on-center transient unit are illustrated. Intensities are denoted by a number (log units relative to background of OScd m’t to the right of each PSTH. At the left side. the spot size (2.4 of arc) was equal to the equivalent center diameter. A small increment (0.08) produced a response which rapidly adapted to the maintained rate for background alone (0). Decrements evoked complete suppression of unit activity for the J set duration of the stimuli below background intensity (-0.78 and - 1.8). With the same intensity. a smaller spot size (0.8’ of arc) an initial suppression and rebound was followed by a partial. but mamtained, suppression of spike discharge (9. -0.78). Increasine the decrement to - 1.8 did not increase the effect. Calibration for the ordinate;s 100 spikes per sec.
Increasing the contrast by 1 log unit to - 1.8 log units could not further decrease the maintained activity. Thus, in terms of decreasing maintained activity, increases in contrast could not compensate for reduction in spot size. We considered the possibility that the lack of further decrement in response to the smaller spot could be due to light scatter. However, an off unit recorded in the same eye under similar conditions was able to increase its onset response (peak frequency) 43?/;, in the range -0.95 to - 1.8. If light scatter had been the only factor in the effect illustrated in Fig. 4, off-amter cells in adjacent regions should have shown a saturating onset response in the same stimulus intensity range. It was also possible to demonstrate silence with black spots on a hand-held wand for on-center units, and a projected spot of light for off units. The size of the spot had to be appropriate and a hand rest had to be employed to eliminate small movements of the stimulus. We could not silence units without correct refraction. Under these conditions, units of all types were observed to be silent for up to 30sec (longest time tested). The effectiveness of contrast in silencing units was dependent upon background illumination. Units which could be silenced at our usual background level about 6cd/m2 were also silenced when the illumination was increased to 60cd/m2. However, when the background was dropped from 6.0 to 0.6 cm/m’ by
the insertion of a light-shielded 1 log unit neutral density filter before the eye (3 units), maintained activity could not be silenced by the same size spot at any contrast; it continued at a lower rate throughout the stimulus period. The reduced effectiveness is perhaps expected because a decrease in background illumination may decrease the effectiveness of the antagonistic surround and increase the size of the center region (Barlow, Fitzhugh and Kuffler, 1957). It is noteworthy that at reduced backgrounds a sustained decrease in the maintained activity to background was evident during the stimulus period. III. Response rime course decrements
to equal increments
and
In order to determine if the characteristic time course of increases in activity were similar to the time course of decrements in activity, the small range of luminances between the background level and the level which would silence the activity was studied. Units with little or no spontaneous activity were not suitable for this purpose. Small changes in maintained activity could be most easily evaluated by plotting the responses as cumulative histograms. Data from an on-center transient unit are illustrated in Fig. 5. In Fig. 5A, the spot size was equal to the equivalent center diameter. The responses to equal increments and decrements (kO.071 log units) are overplotted with the maintained activity to back: ground alone. The onset response for the increment
ROBERT
4%
P.
SCOBEY. LEO
M. CHALUPAand
ROBERT
W.
HAMYON
2.4
DEG.
DIA.
1.2
DEG.
DIA.
SECONDS
Fig. 5. Equal increments and decrements. Cumulative histograms illustrate with three overplotted curves the response of maintained activity to background alone and a small increment and decrement which were equal on a log scale. An increment caused a rapidly adapting burst which returned to a maintained state during the increment which was higher than the maintained rate to background alone (at 4sec the slope of trace labelled +0.071 is greater than trace labelled 0). A decrement of 0.071 caused a pause in the firing rate and the maintained activity during the decrement was at a rate less than that of maintained activity to background alone. Whereas for A the size of the spot was equal to the equivalent center diameter. in C. the spot was only l/S the area. The rough appearance in comparison of C with respect to A was due to the smaller number of action potentials per bin collected during a Zsec period. The response to the onset of increments and offset of a decrement was similar and prominent. A sustained decrement began within 100 msec of the stimulus onset.
OFF
PHASIC
Fig. 6. Equal increments and decrements: off phasic unit. In A. density histograms were separated by a distance equal to the stimulus increment or decrement. Phasic units of the monkey adapt to increases of activity in a few hundred msec. In A. the two histograms in the foreground illustrate a maintained decrement. IO B. increments and decrements near threshold evoked a response which adapted to the rate of maintained activity to background alone within a few hundred msec. With more contrast (D). the sustained decrement to an increment of light was evident.
Maintained suppression
sharply increases from the background rate; the onset response for the decrement is a horizontal plateau. A delay of about 1 set was required before action potentials were seen following the onset of the decrement. Thereafter, the response increased at a small constant slope for 4sec until the offset response. The constant slope indicates that there was no progressive increase in activity towards the maintained rate seen with background alone. Subportions of the receptive field center showed a similar time course (Figs 4B and 5C). Monkey phasic units were seen to adapt to increases in action potentials almost an order of magnitude faster than the transient units of the cat. A corresponding time course change was sought in responses to stimuli which decreased the maintained activity to background alone. Density and cumulative histograms of an off physic unit are illustrated in Fig. 6. Increments of the spot suppressed the activity of the off physic unit. Responses to small increments and decrements (kO.032 log units) are overplotted in Fig. 6B together with the response to background alone. With increments of low contrast, only a short pause in the maintained activity was evident; after the pause the maintained activity returned to a rate equivalent to that at background alone (seen as three parallel lines during the stimulus period). With more contrast (kO.23 log units), the pause evoked by the onset of the increment was followed by a sustained decrement in the rate of action potentials (Fig. 6D-note the difference in slope of the curves labelled 0 and +0.23 during the stimulus). In summary, following pauses evoked by the onset of a dark spot (in on-units) or a light spot (in offunits), neural activity resumed at a constant rate. For very low contrasts the maintained rate was similar to the rate observed to background alone. With suppression was maintained greater contrasts, throughout the stimulus period. Even greater contrasts silenced most units. DISCUSSION
Maintained suppression of retina! ganglion activity was most evident when stimuli equal to the central portion of the receptive field were utilized. With contrasts greater than a few tenths of a log unit this suppression showed no evidence of adaptation. Cat and monkey ganglion cells can be classified into sustained (tonic) and transient (phasic) types on the basis of stimuli which increase discharge rates, but both types of cells manifest a sustained suppression. The known organization of the retina may provide a mechanism to account for the observed maintained suppression. Concentric center-surround
organization
The bipolar cell situated between the photoreceptor and the ganglion cell has a receptive field with a concentric center-surround organixation (Werblin and Dowling, 1969). More recent studies by Shantz and Naka (1976) postulated that the bipolar cell represents a filter which removes the common mode signal by summing two inputs of different polarities from the photoreceptor and horizontal cell. Therefore, one polarity of contrast should result in bipolar cell
457
depolarization; the other polarity of contrast should result in hyperpolarization. Bipolar hyperpolarization would be predicted to result in disfacilitation at the inner plexiform layer, A si!ence of retina! ganglion cells for the duration of electrical hyperpolarization of bipolar cells has been seen by Naka (1977). A suppression under appropriate stimulus conditions would be expected given a maintained bipolar response. Postsynaptic inhibition, which may or may not be present, would not have to be implicated. The observation that most units can be silenced by appropriate contrast is of theoretical interest. An adequate mathematical description of the stimulusresponse function of ganglion cells must therefore predict a response value of xero at some small increment or decrement from the background illumination. Simple logarithmic or power functions lack this feature. In order to deal with the maintained activity, one approach used in the past has been to consider it as spontaneous activity, unrelated to the stimulus. This introduces a constant in the mathematical formulation. A reasonable interpretation of the presence of this constant is that maintained activity would persist in the absence of synaptic input. Inhibition therefore would be present at the inner plexiform layer for the duration of the stimulus. However, there is no evidence that bipolar cells release an inhibitory neurotransmitter (Miller and Dacheux, 1976). Further, if post-synaptic inhibition (rather than disfacilitation) were the silencing mechanism, a small spot of high contrast might have been capable of substituting for a large spot of low contrast; this was not seen. Differential pulse code modulation
The small range between the background luminance and the luminance which silences the unit is part of the range over which visual information can be transmitted. In the restricted range near background (Fig. 3), a decrease in the activity of the off system can signal brightness relative to background and a decrease in the activity of the on system can signal darkness. This possibility has been discussed (Jung, 1973) and supporting data have been reported @eValois, Jacobs and Jones, 1962; Kruger and Fisher, 1975). For example, if a cortical neuron were to summate inhibitory and excitatory synaptic current topologically related to overlapping on and off center neurones, the difference signal would have twice the information content and be independent of the common mode signal on the converging axons. This would serve to enhance detection of small luminance changes. A common mode signal consisting of an increase of activity in both on-center and off-center ganglion cells with movement in the visual field far from the center and surround mechanisms has been described (McIlwain, 1964). This response, originally called the periphery effect, was reported as a gradual increase in excitability with continued stimulation. More recently, Fischer, Kruger and Droll (1975) used small displacements of gratings which was excluded from the center and surround. They demonstrated that the response (now called theshift effect) was even more powerful than had been previously suspected. The function of this activity is unknown. Could it be to
increase the maintained activtt): across the retina and increase the range oker which synergistic effects
thus
of both
on and
tern of
subcortical
off cells
would
permit
units
can
occur’!
converging
A differential on
a cortical
syscell
changes to be distinguished due to the periphery effect.
luminunce
from the activity Sptfritrl .slrnmtrfiot7
The difference in time course found with increments and decrements from the transient units may be related to the spatial summation properties which characterize Y-tvpe units (Enroth-Cugell and Robson. 1966: Hochsteiiand Shapley. 1976). While the spatial summation characteristic of Y-type retinal _uanglion cells is nonlinear between the visual field and the _eanglion cell. whether or not the Y-type ganglion cell is a non-linear summator is unknown. Suppose that retinal ganglion cells are all linear summators of dendritic current. Consider two sites of equal sensitivity in the receptive field. At one site a spot of li_ght is to increment and generate a transient synaptic Input. i\t the other site. the spot decrements generatin! a maintained suppression of synaptic activity. The mtroducrion and withdrawal of a sinusoidal grating covering a set of sites could be considered similarly. Note that no null response could be found under these conditions. Neither could any position of the spatial sinusoid be found to produce a null response. because two qualitatively different time courses of synaptic activation and suppression would not summate uniformly to give a null response over an entire stimulus cycle. On the other hand. if the time courses aere the same for synaptic activation and suppression. a change in position of the spatial sinusoid could adjust the amplitudes of the maintained acrivation and suppression to equal values, resulting in a null position. Exceptions could be imagined and have been found (Shapley and Hochstein. 1975: DeMonasterio. Gouras and Tollhurst. 1976: Jakiela and Enroth-Cupell. 1976). For example, if all the aforementioned synapses conversed on a retinal ganglion cell. a maintained response and a Y-type characteristic would be predicted. Furthermore. it is possible that adaptation in the spike generating mechanism could occur after summation (Partridge and Stevens. 1976). resulting in a transient response with an X-type characreristic. ._((Xn(,\~(r~lyr,,lrnrs-The work was supported by NIH grant No. EY-01495 to R. P. Scobey. We thank J. Horton for computer programs. and D. Howard and M. Boisen for the stimulator construction and laborator) maintenance.
REFERENCES BarIon H. B.. Fitzhugh R. and Kuffler S. W. (1957) Change of organization in the receptive fields of the cat’s retina during dark adaptations. J. Phniol. 137. 335-351. Barlou- H. B. and Levick W. R. (1976) Threshold setting b! the surround of cat retinal ganglion cells. J. Phgiol.
239. 737-757. Boycott 8. 8. and Wlssle H. (1974) The morphological types of ganglion cells of the domesttc cat’s retina. J. Phxsiol. 240. 397-419. Cleland B. G.. Dubin M. W. and Lewck W. R. (1971) Sus-
talnsd
and
transient neurones in the cat‘s retma and nucleus. J. Phywl. 217. 473-496. B. G. and Lcv!ck LV. R. (1971) Brisk and sluggish
latzrsl geniculate Cleland
concentrically
organized
ganglion
cells in the cat’s retina.
J. Ph!,Gol. DO, 1?1-156. Cl&md B. G.. Lebick W. R. and Sanderson K. J. (19731 Prooerties of sustained and transient cells in -Panelion _ the cat retina. J. Phr.siu(. 228. 649-680. De Xlonasterio F. &I.. and Gouras P. (197% Functional properties of ganglion cells of the rhesus monkey retina. J. Phxsiol. 251. 167-195. De Xlonasterlo F. Xi.. Gouras P. and Tolhurst D. J. (1976) Spatial summation. response pattern and conduction velocity of ganglion cells of the rhesus monkey retina. C’isiorl Rrs. 16, 674-678. De Vslois
R. L.. Jacobs
G.
H.
and
Jones
,A. E. 11962)
Effects of increments and decrements of light on neural discharge rate. Sciww 136. 986988. Enroth-Cugell C. and Robson J. G. (1966) The contrast sensitivitv of retinal ganglion cells of the cat. J. Phxsiol. 187. 517~552. Fernald R. and Chase R. t 1971) An improved method for plotting retinal landmarks and focusing the eyes. Vision
Rr.\. 11. 95-96. Fischer B.. Kriiger J. and Droll W. (1975) Quantitative aspects of the shift-effect in cat retinal ganglion cells. Brciin Rrs. 83, 391-403. Hartltne H. K. (1938) The response of single optic nerve fibers of the vertebrate eye to illumination of the retina. .-In!. J. P/~rsiol. 121, 400-11% Hochstein S. and Shapley R. 41. I 1976) Quantitative analysis of retinal ganglion cell classiiications. J. Ph!~io/. 262. 237-264. Jakiela H. G. and Enroth-Cugell C. (1976) Adaptation and d\namlcs In X-Cells and Y-Cells of the cat retina. E.xp/ Brtritf Rrs. 21. 335-341. Jung R. (19731 Visual perceptIon and neurophysiolog). In Hdhook q/‘Scw.sor~~Plt,rsioio~~~. VII 3A (edited by Jung. RJ. Springer. Berlin. Kriiger J. and Fischer B. (1975) Symmetry between the Visual B- and D-systems and equivalence of center and surround: Studies of light increment and decrement in retinal and geniculate neurons of the cat. Biol. C~hernerics 20, X3-236. KutTler S. W. (1953) Dtschargt patterns and functional organization of mammalian retina. J. .Vrwoph~sir~l. 16, 37-68. Marrocco R. T. (1976) Sustained and transient cells in monkey lateral geniculate nucleus: Conduction velocities and response properties. J. .Veuroph.MJl. 39, 340-353. Mcllwain J. T. (1964) Receptive fields of optic tract axons and lateral geniculate cells: Peripheral extent and barbiturate sensitivity. J. Sewophuiol. 27. 11511173. Miller R. F. and Dacheux R. F. (19761 Synaptic organization and ionic basis of on and off channels in Mudpuppy retina. III. X model of ganglion cell receptive field organization based on chloride-free experiments. J. gem
Phrsiol. Naka
67. 679-690.
K.-I. (197X Functional organization of catfish retina. J. .Vrwophxsiol. 40, 26-43. PartrIdge L. D. and Stebens C. F. (1976) A mechanism for spike frequency adaptation. J. Ph.rsiol. 256. 315-332. Scobey R. P. and Horowitz J. M. (1976) Detection of image displacement by phasic cells in peripheral visual fields of the monkey. C’ision Rex. 16, 15-2-t. Shantz ht. and Naka K.-I. (19761 The bipolar cell. L’ision Rrs. 16. 1517-1518. Shapley R. and Hochstein S. (1975) Visual spatial summation in two classes of geniculate cells. .Vnrure 256. 41 l-413. Werblin F. A. and Dowhng J. E. (1969) Organization of the retina of the mudpuppy. NVrcrwus .Lftlcrriosw. II. Intracellular recording. J. Seurophxsiol. 32. 339-355.
University of California.
Davis..
California 95616. U.S.A.
(Receiced 28 April 1978) Abstract-Recordings were made of the changes in unit discharge rates of on-center and off-center retinal ganglion cells when the luminance of a spot in the receptive field was increased or decreased from a background level. A prolonged suppression of maintained activity to background illumination was observed in both transient and sustained cells of cat and monkey. On-units decreased in rate when the luminance of the spot was reduced: off-units decreased in rate when the luminance of the spot was increased. Long duration silence of maintained activity was dependent upon spot size, position and luminance. The most effective spots covered only the receptive field center. With low contrast levels a response time course was found which did not follow a recovery back to maintained rate along a single exponential function. Increasing contrast silenced a progressively greater proportion of retinal ganglion cells. Key Words-suppression
of discharge: retinal ganglion cells; transient; sustained: cat: primate.
INTRODUCTION
IMETHODS
Since the introduction of single-cell recording techniques to the study of visual processes (Hartline, 1938), most experiments have been concerned with those stimulus parameters which induce an increase in the rate of action potentials. Thus, retinal ganglion cells were classified a on-center or off-center (Ku!?ler. 1953) based on whether a particular cell discharges at the onset or offset of a light-spot flashed in the center of the receptive field. He noted that the spontaneous activity of off-center ganglion cells is suppressed by a luminance increment. The discharges of on-center cells may be decreased by a luminance decrement of a spot positioned on the receptive field (DeValois, Jacobs and Jones, 1962; Kriiger and Fischer, 1975; Barlow and Levick, 1976). In the present study stimulus characteristics which increase, or decrease discharges of retinal ganglion cells were investigated in cat and monkey. In particular, we were interested in determining whether the classification of ganglion cells into transient and sustained types observed with stimuii which increase discharge rates (Cleland, Dubin and Levick. 1971; Cleland, Levick and Sanderson, 1973), could also be obtained with stimuli which diminish spontaneous activity levels. Transient cells in both cat and monkey have been reported to adapt rapidly to stimuli which decrease their spontaneous activity (Cleland er al., 1973, Fig. 8; Marrocco, 1976). but this phenomenon has not been studied in detail. ‘Department of Behavioral Biology. School of Medicine, University of California, Davis, California 95616, U.S.A. ’ Department of Psychology, University of California, Davis. California 95616, U.S.A. ’ Present Address: Stanford Research Institute, 333 Ravenswood, Menlo Park, California, U.S.A. 451
Preparation
Recordings were made from ten cats and I6 monkeys. Following anesthesia (halothane for cats, pentobarbital for monkeys), tracheal and venous cannulas were inserted and the animal was aligned in a stereotaxic instrument. A metal bar system was fastened to the skull with screws and acrylic; this served to hold the animal’s head in stereotaxic coordinates after removal of ear and eye bars. In four of the cat and all of the monkey experiments, responses of ganglion cells were recorded by means of an intraocular approach (Scobey and Horowitz. 1976). In the other six cat experiments microelectrode insertions were made through an opening in the skull over the optic tract. All wounds and pressure points were infiltrated with long-lasting local anesthetic (Nupercaine). ARer surgery, halothane was reduced to between 0 and OS’?, and a mixture- of gallamine triethiodide (5-15 mg/kg. h) and d-tubocurarine (O.l-0.3mg/kg. h) was continuously infused to reduce eye movements. Reduction of the halothane was required only when cardiac arythmia was noted in the electrocardiogram. Artificial respiration was maintained with a 70% N,O and 300,; O2 mixture. A constant volume pump was adjusted to provide a 3.5-57; end-tidal CO1 concentration as determined with an infrared CO2 monitor. The animal’s temperature was monitored by a rectal thermistor which electronically controlled a heating pad to maintain body temperature at 37.5’C. A drop of neosynephrine (10%) and atropine sulfate (1%) retracted the cat’s nictitating membrane, paralyzed accommodation, and dilated the pupils. A 1 D contact lens was fitted to each eye. With cats. the eyes were refracted and optic discs plotted on a tangent screen using the method of Fernald and Chase (1971); with monkeys the procedures described by Scobey and Horowitz (1976) were employed. Electrical activity from axons in the cat tract was recorded with tungsten microelectrodes. Glass micropipettes were used for recording from ganglion cells. Srimularion After isolating the action potentials of a single unit. the
recepttve field was plotted on ~1tangent screen usrng backprojected spots of light as ueII as the edges of hand-held black and white wands. .4 st.mdard background luminance of 8 cd,‘m’. measured with an SE1 phorometrr. was produced with tungsten filament lamps. Units were classified as on-center or off-center. as weil as sustained or transient. Only briskly responding units which proved to be unJmbiguous wtth respect to a sustained or transient classification (Cleland and Levick. 1974) were employed in the quantitative analysis of responses. The terms tonic and phasic 41 be used for monkey units (Debfonasterio and Got%%. 197%. After determining the receptive field and response characteristics of a given unit. the tangent screen was removed and a specially constructed stimulator (Fig. 1) was moved to the site of the receptive field. The front cover screen (D) of the stimulator was illuminated by filtered fiuorescent light (3ed m’t which approxima!ely matched the *‘white” color produced by the PJ sulfide aluminized phosphor of a cathode ray tube (A). Light from the cathode ray tube (CRT) first passed through two holes in a box (B) painted black inside. and then through a rotating disc (C) which permitted selection of various spot sizes. The ratio m area between adjacent holes was 2 to 1. The disc was painted the same white color as the cover screen. A mylar diffusing screen was attached to the rear of the box. and the box was sealed to the stimulator. When a normal 60 cycle interleaved television raster was initially tried. the stimulator continuously excited the cat transient units for the duration of the spot presentation. This artifact was eliminated by employing a
1-rrricol sautooth sueep whtch hdd a -I mjec cycle dura.. and a sgnchronized horrront.tl sweep ;rovtded 2 stable non-interleaved pattern with ;L 50 tis~ horizontal sweep duration. The mylar dil%_tsin$ screen e(:mmated all stationary pattern detail. The luminance of the light from the CRT was Adjusted to match the screen within 0.1 102 units. When the light from the CRT increased. the stimsiltor produced a luminous spot relative to background. and when the light from the CRT decreased. the sput appeared darker than the backgruund. With the CRT OK. the spot NSS 1.200 of the background intensity. .1 cdiibrstlotl curve ‘ms taken from the CRT screen before and aim: edch experiment to contirm stability. The fraction of th? ipot intensity due to background was added to the photometrically measured luminance of the CRT screen to CaItbrate dark SPWS (El. Retinal illumination was reduced tn some experiments b:, the addition of shielded neutral density filters in front of the eye. A subjective estimate of cat neural threshold responses was made for each spot size by manually changing the intensity of a flash (0.5 set duration. one per second rate). The voltage to the Z ;LXIS was adjusted until a just-noticeable increase in the rate of the audible neural pulse train was evident in approximately 50”” of the trials. This “auditory threshold method” was used LO measure sensttivity of the cat‘s units to each of 11 available stimulus sizes. The “equivalent center diameter” ua_s estimated by the method descrtbed b> Cleland. Le\ick and Sanderson I 19731. In order IO decrease the ttme necessary to estimate
A
B
Fig. 1. Stimulator. The stimulus apparatus used to generate dark and light spots relative to background consisted of four parts as shown in the exploded diagram. A raster was produced on a cathode ray tube (A). A box painted black on the inside (B) was sealed to the CRT. The light from the CRT passed through a diffusing screen at the CRT (not shown) and through the box. A series of holes could be selected by means of a rotating disc (C) which was painted the same color as the tangent screen (LX The raster luminance was adjusted to the luminance of C and D with the D-A voltage equal to zero (E). The spot luminance could be adjusted between 1 iog units less than background to 1 log unit above the background of 3.0cd.m’ by voltage control of the raster luminance. The calibration curve (El was constructed by adding the reflected light to the light from the CRT alone.
Maintained suppression the size of the receptive tieid of monkey units. the hole size which gave the largest response to small increments or decrements was taken as the center size. For quantitative testing, a minicomputer was employed to control the rate. duration, iterations and intensity of the stimuli. Data for as many as 16 histograms were coilected by randomty intermixing the stimufus triaf values. Each data set was based on eight stimulus presentations. Increments and decrements of spot intensity were controlled by a IO-bit D-A converter over the range of the stimulator. Data were collected as the intervals between sequential action potentials and stimulus events. The inter* vals were measured with a l/8 msec resolution and stored on digital tape for subsequent analysis. During the experiment. the computer presented displays ii~ustrating the number of action potentiats evoked by the onset and offset of a stimulus as a function of the controlling D-A value. Upon comptetion of one set of histograms. another set of intensities was chosen in order to generate responses for which a continuous function could be estimated. Typicatty 10 histograms covering the range of intensities from about 2 log units befow background to 1 log unit above background were computed for a given spot size. Anal_Uis Initially. responses to single stimulus presentations as well as histograms were inspected with either computergenerated oscillascope displays or with paper records produced by a digital plotter. Two measures of response were employed: flf the sum of action potentials for a given stimulus event. and (2) the peak frequency catcutated as the reciprocal of the average of the four shortest successive intervals between action potentials.
maintained level of activity was primarify dependent upon stimulus parameters. The data to be presented are from recordings of 37 cat units (11 on-sustained. 8 on-transient, 9 offsustained, and 9 off-transient) and 33 monkey units (5 on-tonic+ 2 on-phasic. it off-tonic, and 15 offphasic). From these units t342 histograms at difTerent intensitie relative to background were constructed. Briefly recorded units were characterized. but those from which insufficient quantitative data were not obtained to construct stimulus-response curves were excluded. However, the incomplete study of such units supported the conclusions reported. Two groups of histograms, one group from an oncenter unit with only a transient response. and one group from an on-center unit with a sustained component, are illustrated in Fig. 2. The histograms are separated from one another by a distance proportional to the contrast of the stimulus spot to background (value of 0). Responses to stimuh greater than can be displayed on the intensity axis and between the values illustrated were omitted for clarity, These show a continuum between and beyond those illustrated. A conspicuous feature of the illustration is that small decrements in spot luminance with respect to background (negative values) are effective in decreasing the “spontaneous” or “*maintame< activity. Small vaiues of contrast transiently suppressed activity; modest values (0.5 log units) silenced the units for 4sec. The offset of the stimulus then evoked a char&cteristic and strong
RESULTS
The major aim of this study was to investigate the responses of retinal ganglion cells to stimuli which decreased maintains activity. Xn particular, we were interested in the time course of suppression and the extent to which it could be related to the sustained or transient response category (Cleland, Dubin and Levick, 1971). Decrements of light with on-center neurons. or increments with off-center neurons produced a suppression of maintained activity. The time course of responses to stimuli which decreased the
453
burst.
The time course of the response decrement was dependent upon the position and size of the stimulus in the receptive field. For the units shown in Fig, 2 the spot was equal to the equivalent center diameter (defined in Methods). and positioned to cover the center of the receptive field. These conditions had to be fulfilled in order to evoke the maximum number of unit discharges and also were often necessary to produce a sustained response decrement. Under these conditions, and with appropriate contrast. response decrements lasting for the duration of the stimulus were observed in all cat and monkey units recorded.
A
LOG
r/r,
SECONDS
fig. 2. The figure was constructed by displacement of histograms a distance proportional to the togarithm of the ratio between the background and stimulus intensity. Two cat on center units are itiustrated. a transient unit (A) and a sustained unit (IS).The spot size was equal to the equivalent center diameter and centered on the receptive field. With decrements, the maintained activity to background alone was diminished and silenced with increasing contrast. The offset of the decrement evoked a large response. Calibration mark for the ordinate is 100 spikes per see. The histograms were smoothed by convolution of data with a 3 bin rectangular function.
ROBERT
454
P.
S~OBEY.
LEO M.
CHALL’PA and
In order to provide an adequate estimate of the population response. we systematically varied luminance of spots within the center region while keeping background. spot size, and position constant. These parameters were manipuIated in only a small number of units.
Using a spot centered in the receptive fieId which was equal to the equivalent center diameter, 47 units could be silenced by appropriate contrast for the duration of the stimulus (as shown in Fig. 2). Seven units had an obvious sustained decrease in the tiring rate but would not silence. The procedure used to estimate the minimum contrast which would silence each of the 47 units for the duration of the stimulus was standardized as fotlotvs. A stimulus-repose function using a spot covering the center of the receptive field was constructed by plotting either the peak frequency or the number of action potentials during the stimulus as a function of contrast {Fig. 3). A straight line best fitting these data at background (1 = 0) was drawn by eye and the intersection of this line with the abscissa was recorded. The data from cat and monkey, as well as the data from on and off. sustained (tonic) and transient {phasic), were first evaluated separately. Differences were thought not to be significant and the data were combined to give a distribution for onunits and a distribution for off-units (Fig. 3B). The median contrast necessary for silencing the on and
j-
1.0 I”
8
KCXERT
W.
HALW)~
the off System W;LS -0.11 and +0.17 log units. respectively. Not all units were silenced at the same contrast. AS shown in Fig. 39 there is a range over which increasing contrast silences a progressively increasing proportion of retinal ganglion ceils. It is noteworthy that the reSu)tS illustrated in Figs 2 and 3 were obtained with spot sizes which were equal to the equivalent center diameter. Spots which were not centered on the receptive field and covered only a portion of the center as well as the antagonistic surround would not silence the unit. Additionally, for many units spot size was a critical stimuIus dimension. Spots which were substantially larger and centered on the receptive field might not silence the units. Likewise. spots which were substantially smaller were ineffective (Fig. 4). Two sets of 4 histograms were selected from the 26 available to illustrate the responses to increments and decrements of two different spot sizes. In Fig. 4X. responses of an on-unit to increments of a spot equal to the equivalent center diameter are illustrated. For comparison. an equivalent response \vaS selected from the set of histograms that were obtained with a spot only 1;9 of the area (Fig. 4B). Only a small increase in spot luminance (0.03 log, units) was necessary to offset the reduction of spot size. In terms of suppression of maintained activity. a stimulus decrement of 0.78 log units at a spot size equal to the equivalent center diameter was capable of silencing the maintained activity for 4sec. At the same contrast the smaller spot (Fig. 48) viefded maintained partial suppression.
t
OFF
Fig. 3. Synergistic range of on and off units. In (A). a stimulus-response lunction of an an-transient unit illustrates the method used to estimate the minimum increment or decrement necessary to achieve suppression of unit activity. The size and position of the stimulus was chosen for silencing units with least contrast. Each point represents the peak response frequency obtained for a single stimulus presentation at a given intensity. A best fit straight line was drawn to approximate the data near background; the intersection with the abscissa was used as a data point as indicated by the dotted line. in (B), a cumulative histogram of on- and off-center units derived by this method illustrated the range of luminance necessary for silence. The duration of the stimulus used in the cat experiments was 4sec while the duration of the stimulus used for the monkey experiments was either I or isec. Data for monkey and cat units were combined. Background was between 0.5 and l.Ocd;m’. The cumulative histogram to the left of zero indicates the decrement necessary to suppress activity in on cells. The curve to the right of zero is the value of the increment necessary to silence off units. The space between the two curves is an estimate of the minimum range over which both the on and off system could transmit information about increments and decrements.
ZIaintained suppression B
A
SPOT DIAMETER 0.8 0~ ARC
SPOT DIAMETER 24’ OF ARC
BACKGROUND ALONE 0
-.7a
1 0
1
I
4
SECONDS
8
;1
;
SECONDS
i
Fig. 4. Sustained decrement of maintained activity. The responses of an on-center transient unit are illustrated. Intensities are denoted by a number (log units relative to background of OScd m’t to the right of each PSTH. At the left side. the spot size (2.4 of arc) was equal to the equivalent center diameter. A small increment (0.08) produced a response which rapidly adapted to the maintained rate for background alone (0). Decrements evoked complete suppression of unit activity for the J set duration of the stimuli below background intensity (-0.78 and - 1.8). With the same intensity. a smaller spot size (0.8’ of arc) an initial suppression and rebound was followed by a partial. but mamtained, suppression of spike discharge (9. -0.78). Increasine the decrement to - 1.8 did not increase the effect. Calibration for the ordinate;s 100 spikes per sec.
Increasing the contrast by 1 log unit to - 1.8 log units could not further decrease the maintained activity. Thus, in terms of decreasing maintained activity, increases in contrast could not compensate for reduction in spot size. We considered the possibility that the lack of further decrement in response to the smaller spot could be due to light scatter. However, an off unit recorded in the same eye under similar conditions was able to increase its onset response (peak frequency) 43?/;, in the range -0.95 to - 1.8. If light scatter had been the only factor in the effect illustrated in Fig. 4, off-amter cells in adjacent regions should have shown a saturating onset response in the same stimulus intensity range. It was also possible to demonstrate silence with black spots on a hand-held wand for on-center units, and a projected spot of light for off units. The size of the spot had to be appropriate and a hand rest had to be employed to eliminate small movements of the stimulus. We could not silence units without correct refraction. Under these conditions, units of all types were observed to be silent for up to 30sec (longest time tested). The effectiveness of contrast in silencing units was dependent upon background illumination. Units which could be silenced at our usual background level about 6cd/m2 were also silenced when the illumination was increased to 60cd/m2. However, when the background was dropped from 6.0 to 0.6 cm/m’ by
the insertion of a light-shielded 1 log unit neutral density filter before the eye (3 units), maintained activity could not be silenced by the same size spot at any contrast; it continued at a lower rate throughout the stimulus period. The reduced effectiveness is perhaps expected because a decrease in background illumination may decrease the effectiveness of the antagonistic surround and increase the size of the center region (Barlow, Fitzhugh and Kuffler, 1957). It is noteworthy that at reduced backgrounds a sustained decrease in the maintained activity to background was evident during the stimulus period. III. Response rime course decrements
to equal increments
and
In order to determine if the characteristic time course of increases in activity were similar to the time course of decrements in activity, the small range of luminances between the background level and the level which would silence the activity was studied. Units with little or no spontaneous activity were not suitable for this purpose. Small changes in maintained activity could be most easily evaluated by plotting the responses as cumulative histograms. Data from an on-center transient unit are illustrated in Fig. 5. In Fig. 5A, the spot size was equal to the equivalent center diameter. The responses to equal increments and decrements (kO.071 log units) are overplotted with the maintained activity to back: ground alone. The onset response for the increment
ROBERT
4%
P.
SCOBEY. LEO
M. CHALUPAand
ROBERT
W.
HAMYON
2.4
DEG.
DIA.
1.2
DEG.
DIA.
SECONDS
Fig. 5. Equal increments and decrements. Cumulative histograms illustrate with three overplotted curves the response of maintained activity to background alone and a small increment and decrement which were equal on a log scale. An increment caused a rapidly adapting burst which returned to a maintained state during the increment which was higher than the maintained rate to background alone (at 4sec the slope of trace labelled +0.071 is greater than trace labelled 0). A decrement of 0.071 caused a pause in the firing rate and the maintained activity during the decrement was at a rate less than that of maintained activity to background alone. Whereas for A the size of the spot was equal to the equivalent center diameter. in C. the spot was only l/S the area. The rough appearance in comparison of C with respect to A was due to the smaller number of action potentials per bin collected during a Zsec period. The response to the onset of increments and offset of a decrement was similar and prominent. A sustained decrement began within 100 msec of the stimulus onset.
OFF
PHASIC
Fig. 6. Equal increments and decrements: off phasic unit. In A. density histograms were separated by a distance equal to the stimulus increment or decrement. Phasic units of the monkey adapt to increases of activity in a few hundred msec. In A. the two histograms in the foreground illustrate a maintained decrement. IO B. increments and decrements near threshold evoked a response which adapted to the rate of maintained activity to background alone within a few hundred msec. With more contrast (D). the sustained decrement to an increment of light was evident.
Maintained suppression
sharply increases from the background rate; the onset response for the decrement is a horizontal plateau. A delay of about 1 set was required before action potentials were seen following the onset of the decrement. Thereafter, the response increased at a small constant slope for 4sec until the offset response. The constant slope indicates that there was no progressive increase in activity towards the maintained rate seen with background alone. Subportions of the receptive field center showed a similar time course (Figs 4B and 5C). Monkey phasic units were seen to adapt to increases in action potentials almost an order of magnitude faster than the transient units of the cat. A corresponding time course change was sought in responses to stimuli which decreased the maintained activity to background alone. Density and cumulative histograms of an off physic unit are illustrated in Fig. 6. Increments of the spot suppressed the activity of the off physic unit. Responses to small increments and decrements (kO.032 log units) are overplotted in Fig. 6B together with the response to background alone. With increments of low contrast, only a short pause in the maintained activity was evident; after the pause the maintained activity returned to a rate equivalent to that at background alone (seen as three parallel lines during the stimulus period). With more contrast (kO.23 log units), the pause evoked by the onset of the increment was followed by a sustained decrement in the rate of action potentials (Fig. 6D-note the difference in slope of the curves labelled 0 and +0.23 during the stimulus). In summary, following pauses evoked by the onset of a dark spot (in on-units) or a light spot (in offunits), neural activity resumed at a constant rate. For very low contrasts the maintained rate was similar to the rate observed to background alone. With suppression was maintained greater contrasts, throughout the stimulus period. Even greater contrasts silenced most units. DISCUSSION
Maintained suppression of retina! ganglion activity was most evident when stimuli equal to the central portion of the receptive field were utilized. With contrasts greater than a few tenths of a log unit this suppression showed no evidence of adaptation. Cat and monkey ganglion cells can be classified into sustained (tonic) and transient (phasic) types on the basis of stimuli which increase discharge rates, but both types of cells manifest a sustained suppression. The known organization of the retina may provide a mechanism to account for the observed maintained suppression. Concentric center-surround
organization
The bipolar cell situated between the photoreceptor and the ganglion cell has a receptive field with a concentric center-surround organixation (Werblin and Dowling, 1969). More recent studies by Shantz and Naka (1976) postulated that the bipolar cell represents a filter which removes the common mode signal by summing two inputs of different polarities from the photoreceptor and horizontal cell. Therefore, one polarity of contrast should result in bipolar cell
457
depolarization; the other polarity of contrast should result in hyperpolarization. Bipolar hyperpolarization would be predicted to result in disfacilitation at the inner plexiform layer, A si!ence of retina! ganglion cells for the duration of electrical hyperpolarization of bipolar cells has been seen by Naka (1977). A suppression under appropriate stimulus conditions would be expected given a maintained bipolar response. Postsynaptic inhibition, which may or may not be present, would not have to be implicated. The observation that most units can be silenced by appropriate contrast is of theoretical interest. An adequate mathematical description of the stimulusresponse function of ganglion cells must therefore predict a response value of xero at some small increment or decrement from the background illumination. Simple logarithmic or power functions lack this feature. In order to deal with the maintained activity, one approach used in the past has been to consider it as spontaneous activity, unrelated to the stimulus. This introduces a constant in the mathematical formulation. A reasonable interpretation of the presence of this constant is that maintained activity would persist in the absence of synaptic input. Inhibition therefore would be present at the inner plexiform layer for the duration of the stimulus. However, there is no evidence that bipolar cells release an inhibitory neurotransmitter (Miller and Dacheux, 1976). Further, if post-synaptic inhibition (rather than disfacilitation) were the silencing mechanism, a small spot of high contrast might have been capable of substituting for a large spot of low contrast; this was not seen. Differential pulse code modulation
The small range between the background luminance and the luminance which silences the unit is part of the range over which visual information can be transmitted. In the restricted range near background (Fig. 3), a decrease in the activity of the off system can signal brightness relative to background and a decrease in the activity of the on system can signal darkness. This possibility has been discussed (Jung, 1973) and supporting data have been reported @eValois, Jacobs and Jones, 1962; Kruger and Fisher, 1975). For example, if a cortical neuron were to summate inhibitory and excitatory synaptic current topologically related to overlapping on and off center neurones, the difference signal would have twice the information content and be independent of the common mode signal on the converging axons. This would serve to enhance detection of small luminance changes. A common mode signal consisting of an increase of activity in both on-center and off-center ganglion cells with movement in the visual field far from the center and surround mechanisms has been described (McIlwain, 1964). This response, originally called the periphery effect, was reported as a gradual increase in excitability with continued stimulation. More recently, Fischer, Kruger and Droll (1975) used small displacements of gratings which was excluded from the center and surround. They demonstrated that the response (now called theshift effect) was even more powerful than had been previously suspected. The function of this activity is unknown. Could it be to
increase the maintained activtt): across the retina and increase the range oker which synergistic effects
thus
of both
on and
tern of
subcortical
off cells
would
permit
units
can
occur’!
converging
A differential on
a cortical
syscell
changes to be distinguished due to the periphery effect.
luminunce
from the activity Sptfritrl .slrnmtrfiot7
The difference in time course found with increments and decrements from the transient units may be related to the spatial summation properties which characterize Y-tvpe units (Enroth-Cugell and Robson. 1966: Hochsteiiand Shapley. 1976). While the spatial summation characteristic of Y-type retinal _uanglion cells is nonlinear between the visual field and the _eanglion cell. whether or not the Y-type ganglion cell is a non-linear summator is unknown. Suppose that retinal ganglion cells are all linear summators of dendritic current. Consider two sites of equal sensitivity in the receptive field. At one site a spot of li_ght is to increment and generate a transient synaptic Input. i\t the other site. the spot decrements generatin! a maintained suppression of synaptic activity. The mtroducrion and withdrawal of a sinusoidal grating covering a set of sites could be considered similarly. Note that no null response could be found under these conditions. Neither could any position of the spatial sinusoid be found to produce a null response. because two qualitatively different time courses of synaptic activation and suppression would not summate uniformly to give a null response over an entire stimulus cycle. On the other hand. if the time courses aere the same for synaptic activation and suppression. a change in position of the spatial sinusoid could adjust the amplitudes of the maintained acrivation and suppression to equal values, resulting in a null position. Exceptions could be imagined and have been found (Shapley and Hochstein. 1975: DeMonasterio. Gouras and Tollhurst. 1976: Jakiela and Enroth-Cupell. 1976). For example, if all the aforementioned synapses conversed on a retinal ganglion cell. a maintained response and a Y-type characteristic would be predicted. Furthermore. it is possible that adaptation in the spike generating mechanism could occur after summation (Partridge and Stevens. 1976). resulting in a transient response with an X-type characreristic. ._((Xn(,\~(r~lyr,,lrnrs-The work was supported by NIH grant No. EY-01495 to R. P. Scobey. We thank J. Horton for computer programs. and D. Howard and M. Boisen for the stimulator construction and laborator) maintenance.
REFERENCES BarIon H. B.. Fitzhugh R. and Kuffler S. W. (1957) Change of organization in the receptive fields of the cat’s retina during dark adaptations. J. Phniol. 137. 335-351. Barlou- H. B. and Levick W. R. (1976) Threshold setting b! the surround of cat retinal ganglion cells. J. Phgiol.
239. 737-757. Boycott 8. 8. and Wlssle H. (1974) The morphological types of ganglion cells of the domesttc cat’s retina. J. Phxsiol. 240. 397-419. Cleland B. G.. Dubin M. W. and Lewck W. R. (1971) Sus-
talnsd
and
transient neurones in the cat‘s retma and nucleus. J. Phywl. 217. 473-496. B. G. and Lcv!ck LV. R. (1971) Brisk and sluggish
latzrsl geniculate Cleland
concentrically
organized
ganglion
cells in the cat’s retina.
J. Ph!,Gol. DO, 1?1-156. Cl&md B. G.. Lebick W. R. and Sanderson K. J. (19731 Prooerties of sustained and transient cells in -Panelion _ the cat retina. J. Phr.siu(. 228. 649-680. De Xlonasterio F. &I.. and Gouras P. (197% Functional properties of ganglion cells of the rhesus monkey retina. J. Phxsiol. 251. 167-195. De Xlonasterlo F. Xi.. Gouras P. and Tolhurst D. J. (1976) Spatial summation. response pattern and conduction velocity of ganglion cells of the rhesus monkey retina. C’isiorl Rrs. 16, 674-678. De Vslois
R. L.. Jacobs
G.
H.
and
Jones
,A. E. 11962)
Effects of increments and decrements of light on neural discharge rate. Sciww 136. 986988. Enroth-Cugell C. and Robson J. G. (1966) The contrast sensitivitv of retinal ganglion cells of the cat. J. Phxsiol. 187. 517~552. Fernald R. and Chase R. t 1971) An improved method for plotting retinal landmarks and focusing the eyes. Vision
Rr.\. 11. 95-96. Fischer B.. Kriiger J. and Droll W. (1975) Quantitative aspects of the shift-effect in cat retinal ganglion cells. Brciin Rrs. 83, 391-403. Hartltne H. K. (1938) The response of single optic nerve fibers of the vertebrate eye to illumination of the retina. .-In!. J. P/~rsiol. 121, 400-11% Hochstein S. and Shapley R. 41. I 1976) Quantitative analysis of retinal ganglion cell classiiications. J. Ph!~io/. 262. 237-264. Jakiela H. G. and Enroth-Cugell C. (1976) Adaptation and d\namlcs In X-Cells and Y-Cells of the cat retina. E.xp/ Brtritf Rrs. 21. 335-341. Jung R. (19731 Visual perceptIon and neurophysiolog). In Hdhook q/‘Scw.sor~~Plt,rsioio~~~. VII 3A (edited by Jung. RJ. Springer. Berlin. Kriiger J. and Fischer B. (1975) Symmetry between the Visual B- and D-systems and equivalence of center and surround: Studies of light increment and decrement in retinal and geniculate neurons of the cat. Biol. C~hernerics 20, X3-236. KutTler S. W. (1953) Dtschargt patterns and functional organization of mammalian retina. J. .Vrwoph~sir~l. 16, 37-68. Marrocco R. T. (1976) Sustained and transient cells in monkey lateral geniculate nucleus: Conduction velocities and response properties. J. .Veuroph.MJl. 39, 340-353. Mcllwain J. T. (1964) Receptive fields of optic tract axons and lateral geniculate cells: Peripheral extent and barbiturate sensitivity. J. Sewophuiol. 27. 11511173. Miller R. F. and Dacheux R. F. (19761 Synaptic organization and ionic basis of on and off channels in Mudpuppy retina. III. X model of ganglion cell receptive field organization based on chloride-free experiments. J. gem
Phrsiol. Naka
67. 679-690.
K.-I. (197X Functional organization of catfish retina. J. .Vrwophxsiol. 40, 26-43. PartrIdge L. D. and Stebens C. F. (1976) A mechanism for spike frequency adaptation. J. Ph.rsiol. 256. 315-332. Scobey R. P. and Horowitz J. M. (1976) Detection of image displacement by phasic cells in peripheral visual fields of the monkey. C’ision Rex. 16, 15-2-t. Shantz ht. and Naka K.-I. (19761 The bipolar cell. L’ision Rrs. 16. 1517-1518. Shapley R. and Hochstein S. (1975) Visual spatial summation in two classes of geniculate cells. .Vnrure 256. 41 l-413. Werblin F. A. and Dowhng J. E. (1969) Organization of the retina of the mudpuppy. NVrcrwus .Lftlcrriosw. II. Intracellular recording. J. Seurophxsiol. 32. 339-355.
