Optokinetic Responses of the Crab, Carcinus to A Single Moving Light

These and other difficulties are overcome if a small point source of light is used as a stimulus, but no one, apart from Rademaker & ter Braak (1948) on the rabbit, seems to have attempted to use this for optokinetic responses. To do so, the eye movements must be recorded in a light proof room when the only light is the minute stimulus bulb of a few milliwatts. The crab then sees no stationary objects on which it might otherwise stabilize its eye. The recording system must operate in the dark and be sensitive to movements down to 0·01°. These methods having been developed for the studies on optokinetic memory, they have been used with almost no modification. The result, that optokinetic responses are made to a single point source, makes possible many new types of experiment on the visual system. The electrophysiological responses to lights are better known than those to stripes; lights may be switched to distant points, substituted for others of differing colour, flickered in a sinusoidal or on-off sequence, moved in two dimensions, transformed from spots into lines, moved in paths which are circular, elliptical or square, and so on. Some of these possibilities are explored here but much further work remains to be done.

The apparatus described in the first paper (Horridge, 1966) was modified as follows. The drum was dispensed with; the crab was clamped as before, with a flag on the eyestalk. The flag moves over a pair of photocells which are actuated by a beam of infra-red light. For this stimulus, a subminiature bulb (1·5 V., 45 mA., Pinlight, Kay Electric Co., Fairfield, New Jersey) is mounted on the end of a light arm of a pen-recorder solenoid, so providing a linear movement over small distances, with an accuracy of better than 1 % and frequency response up to 20 cyc./sec. The lamp can be arranged to follow any specified waveform at any position relative to the crab. A low-frequency generator of sine waves and ramp-functions was used for most of the experiments. To provide a sequence of flashes the lamp was operated from a generator which provided up to 50 mA. of current.

The stimulating light was held at a distance of 57·3 cm. from the eye, to facilitate calculations of angle and to avoid illuminating objects near the eye. The crab was surrounded by black card and saw the lamp through a hole in this screen. The light illuminates the eye surface with a brightness of about 0·05 lux at 57 cm. In most experiments reported here the crab was placed directly facing the stimulus and on the same level.

Responses are smaller than corresponding responses to brightly illuminated striped drums at the same stimulus speeds. However, the relation between the velocity of the response and that of the stimulus is strikingly similar to the result previously obtained for striped drums (Horridge & Sandeman, 1964), so that evidently the response to a drum is a summed response which is greater than that evoked by each part of the eye separately. With a sinusoidal oscillation of the light of amplitude 0·4° in a horizontal plane the response is up to 40 % of the stimulus at a frequency of about 0·005 eye./sec. At lower frequency the response falls off steeply but with increasing frequency it slowly falls away until nothing is seen above 5 or 10 per sec. (Fig. 1). This confirms the extraordinary sensitivity to low speeds even with single lights. The maximum velocity contained in the sinusoidal oscillation of the light is equal to π × angular amplitude subtended at the eye × frequency, which, for a frequency of 0·005/sec. and amplitude 0·4° works out at 0·0063°/sec. or nearly 1·4 full rotations per 24 hr. period. The response falls to nothing at about a third of this and so appears to be specially adapted to follow the movements of the sun or moon. More on this topic will follow in the next paper. Tests with ramp-functions in the horizontal plane at these velocities confirm that the optimum response lies between 0·002 and 0·005°/sec. To compare with responses from striped drums the slip speed of the single light must be calculated, because the slip speed is the true average movement stimulus across the eye. For a drum the highest velocity gain obtained was nearly 15 at a slip speed of 0·oo1°/sec.; for a single small light the best was a velocity gain of 0·5 at a slip speed of 0·002°/sec., on a different specimen.

For fast jumps of the light, linearity is maintained for stimulus amplitudes up to about 1°, but for greater angles the ratio of response to stimulus falls off, suggesting that the mechanism for movement perception operates best over short distances. At low frequencies the response is maintained for many oscillations but adaptation becomes noticeable over the same range as for drums. At periods from 30 sec. down to about 3 sec. the first response is larger than the succeeding ones. At periods of 3 sec. or less the response is commonly more in one direction than in the other, with a small superimposed oscillation. These features are illustrated in Fig. 2, where it is also apparent that the responses are much less perfect that those to drums. One cannot distinguish whether the imperfections arise on the sensory side, or from the inability of the muscles to perform the small movements. Responses are frequently better in one direction than the other, and this causes a net movement during the period when the stimulus is oscillating, as in Fig. 2 C. When the stimulus stops the eye moves back toward the former position, although it will not necessarily reach it. This shows, as in the memory experiments, that there is some factor which favours the return to the former position despite the intervening activity, and that in the recovery movement the eye passes across a stationary stimulus.

When the velocity across the eye is reduced to the limit by making the amplitude small and the period of a sinusoidal oscillation long, the response fails in a characteristic way. The eye now follows only the most rapid parts of the movement and otherwise stabilizes on the light as if it were stationary. The response to a sinusoidal stimulus thus comes to approximate to a series of square waves (Fig. 2D). Linear analysis of the low-frequency components is not convenient with such a system. With reduction of the intensity of the light, even for larger amplitudes and higher frequencies, the response to a sinusoidal movement fails in exactly the same way, persisting only when the movement is optimum, for a light giving about 0·0003 lux at the eye. Every stimulus on the eye must excite numerous ommatidia only marginally, by oblique rays, and although the overall response may appear as a straightforward function of the input, it is clear that contributing to it are non-linear components which, like the velocity of the stimulus, cannot be restricted to a small range for an oscillatory analysis.

When the light is moved from one place to another by a ramp function the eye follows as it does a drum and continues to move for a time after the stimulus has stopped. An overshoot is always found but takes two forms, both of which are illustrated in the asymmetrical responses of Fig. 3 A. Responses towards the left have two stages in this example—one follows the movement, the other is the subsequent follow up—but percentage following is low for both movements. Movements to the right in Fig. 3 A are to a larger extent achieved by the immediate response, whereas those to the left are not, and eye movement continues for many seconds after the drum movement. Crabs vary greatly in the form of the response curve and in the amplitude for a constant stimulus, but this is mainly a matter of relative content of slow and fast components. For slower movements the distinction into immediate and subsequent responses is not so obvious, and in any case is only seen when a ramp function is employed as stimulus. With an oscillatory stimulus the overshoot seen with single ramps contributes to the latency of the change in direction, so that plots of lag against frequency can only be superficially analysed as if they originate in a single linear system. Actual latencies range from 0·9 sec. for movements of 2°/sec., to 5 sec. for 0·05°/sec., and can be as long as 10–15 Sec. at 0·005°/sec. Clearly they allow time for averaging numerous nerve impulses over relatively long periods.

When a stationary light is switched off and a similar one appears nearby, the crab sees the effect as motion, just as we do ourselves. Responses are large for small angles subtended at the eye and are fairly constant for all larger angles up to 30° or more. Results for small angles are plotted in Fig. 4, in which the smooth curves are intended to suggest two classes of response, those on the left near 1° ± 0·5° and all others to the right. Without any real proof, the suggestion is that the first class is of outputs from a movement-perception system and the second is derived from a position-sensitive system. This idea would agree with the distinction into immediate and subsequent responses when the stimulus is a single ramp function.

Actual responses to apparent movement have a rapid initial phase with a slower follow-up to a plateau. The fast initial phase frequently overshoots but recovery from the overshoot is distinct from the subsequent slow phase. The new equilibrium takes up to 30 sec. to reach (Fig. 5). For the whole of the response the crab sees only one stationary light at a time. The visual feedback loop is closed for the whole time so that during the response the light must appear to sweep across the eye in the opposite direction. A rapid movement of one light produces a much bigger response for stimulus angles even as small as 1°, and for large stimulus angles the stimulation of intervening receptors along the path of the light makes for much larger responses than are obtained by only an apparent motion over the same distance. Clearly there is here a flexible system which is open to many types of experiment, including tests for the numerous illusions of movement which are known for man (Bartley, 1941). A characteristic feature is that the eye moves in steps, the size of which depend on its rate of movement since the steps come at rather regular intervals.

To produce a dark period between the disappearance of one light and the appearance of the other is to combine the previous experiments on apparent motion with the earlier ones on optokinetic memory. During the intervening period of absolute darkness the eye drifts, which itself imposes a relative movement of the eye on the light (Fig. 6C). The results are exactly comparable with those obtained with striped drums. Movement of the light during the dark period is followed on re-illumination by an eye movement in the appropriate direction. As with all responses to single lights, the percentage following is rather small, rarely reaching 15% even for short dark periods. As in movement perception the percentage following is much less for a single light than for a striped drum, showing that the total forward gain in the latter case is summed from many reduplicated sensing devices, each of lower gain.

The longer the period of darkness before the light reappears, the slower is the subsequent response (Fig. 6). There is an enormous scatter for periods longer than a few seconds and there is no indication as to how the results can be averaged, so all measurements for one crab have been plotted in Fig. 7. These results can be interpreted in at least two ways: (a) the longer the period of darkness the smaller the restoring force against a constant inertia; (b) after longer periods of darkness only circuits of longer time-constants are excited. The second explanation is preferred because it agrees closely with experiments to be described in which the circuits of short time-constant are fatigued while those of longer time-constant remain functional. Responses to sudden movements are then slow but the same equilibrium is finally reached. Circuits which require long integration times for perception of movement produce correspondingly slow eye movement commands.

When the light is intermittent the responses described above are reduced. Without a large series of experimental runs, for which no useful purpose could be seen at this stage, it is not possible to give exact results, but the percentage following falls off approximately in proportion to the time that the light is off, in the frequency range 5/sec. down to 1 in 3 sec. For lower flash frequencies the results are complicated by the drift in the dark periods. At frequencies greater than about 5/sec. the intermittent light is not distinguished from a continuous one of equivalent brightness.

This simple picture is complicated by several new features. First, after a period of flashing during which it is cut down, the response takes many seconds to recover its former size, as best seen when the stimulus is a constant oscillatory response (Fig. 8). Secondly, experiments with apparent motion also give reduced responses when the two lights are intermittent, and similar small responses when only one of the lights is intermittent. It does not seem to make much difference whether the first or the second light is intermittent (Fig. 9). The reasons for this are not simple because the result depends, not on straightforward considerations of the stimulating effectiveness of the lights or the ability to infer movement during flicker, but upon the familiarity of the crab with the experimental situation.

During these experiments it became obvious that the crab learns to distinguish between a continuous light and an intermittent one. When a crab, which has been responding well in apparent motion experiments, is presented for the first time with a series of apparent motions in which one of the lights is intermittent, the response falls off progressively at each succeeding test (Fig. 10). The intervals between tests can be up to several minutes and the progressive change still occurs, but if the crab is left for an hour it forgets the previous series and begins again with a large response and has to re-learn the difference between the two lights which look so obviously different to the human eye. When the response to, say, continuous light followed by flashes has fallen to zero after four or five repeats, the response to flashes followed by continuous light will also have fallen off, although not completely, and the response to apparent motion between two continuous lights will still be normal. Therefore the crab is making a generalized distinction between the intermittent and the continuous light, irrespective of their order of presentation. These observations strongly suggest that a primitive form of learning is superimposed on the optokinetic response and emphasize that we are dealing with a whole animal, which is much more than a system for movement perception with muscles attached.

In all the above experiments the pin light was operated at maximum brightness, giving an illumination of 0-05 lux at a surface in the position of the crab’s eye. When the intensity of the light is reduced with Kodak Wratten filters the response falls by approximately 20% for each tenfold reduction in intensity (Fig. 11). The lamp emits no ultraviolet light and the eye is insensitive to infra-red, so that one can have confidence in the results obtained with gelatine filters. The results do not immediately agree with a theory which assumes that the response depends on the square of the illumination (cf. Reichardt, 1961). For sinusoidal oscillations of periods in the range 10-1000 sec. which achieve high percentage responses, as in Fig. 1, the nature of the failure at the lowest effective intensities is that the low-velocity components of the response drop out first so that responses to sinusoidal movements are flat-topped, almost rectangular, movements, resembling the response in Fig. 2D. This happens at an illumination of less than 0·0005 lux at the eye. The reduction in the response is interpreted as due to the loss of the contribution of the off-axis ommatidia and there is at present no way of estimating how many off-axis ommatidia are in a geometrical position to be stimulated above their threshold.

The response to movement of a single light opens an avenue to many types of experiment, as outlined in the introduction. Being two-dimensional arrays of receptors and synapses, the eye and optic lobe can be stimulated in two dimensions by a single light, which is not possible with a drum. Further experiments in progress show that the optokinetic responses which illustrate movement perception are all fully operative in two dimensions, although a stimulus which moves along a line inclined at 45° to the horizontal evokes a response which is not necessarily at the same angle. Similarly, the number of lights can be increased; preliminary experiments along these lines show that when an additional light accompanies the first the response is increased. Experiments with two lights lead to tests of the acuity when the two lights will merge into one. However, the ability of the eye to respond depends on the change in intensity over a pattern of receptors which are differently stimulated according to their acceptance angle and inclination to each other. When these parameters are known for a crab it may be possible to relate the small responses to a single light with the larger ones to a series of stripes.

The effectiveness of a small light of brightness 0·05 lux in eliciting eye movements down to stimulus velocities of 0·003°/sec. shows that the sun and the moon (full moon gives up to 0·24 lux) must also be effective. So far this has been demonstrated only for the sun (see the succeeding paper).

The deleterious effect of intermittent light suggests that temporal coding plays an essential part in movement perception, but the possibilities for devising models of the mechanism are so numerous that we must await electrophysiological records from the region where movement is abstracted. The movement-receptor units so far isolated in optic tracts are all wide-field units (Waterman, Wiersma & Bush, 1964). This fact is compatible with the present findings, and with those which show an optokinetic memory to an accuracy of a fraction of a degree, only if the crab has refined information of the absolute position of a stimulus. This in turn is compatible with a system in which the fine discrimination at small angles lies only in the input to wide-angle movement receptors.

The ability of the crab to appreciate the difference between an intermittent and a continuous light when the alternation between these is repeated a few times suggests a new preparation for the study of learning, but more usefully it shows that the optokinetic response can be employed as a test of discrimination, as between coloured lights, slightly different patterns and so forth. It also warns us that the optokinetic mechanism may be modified by repetition of the stimulus.

1. A crab in an otherwise dark room will stabilize its eye position by reference to a single small light, so long as the illumination at the eye exceeds about 0·0003 hix.

2. The eye movements follow the movements of the light.

3. Responses to a light moving in a horizontal plane resemble those to a striped drum, but at lower percentage following.

4. Apparent motion is an effective stimulus; with intermittent light the response is reduced. If there is a period of complete darkness after the first light the subsequent movement, when the second light comes on, is slower for longer dark periods.

5. The crab learns, after some repetitions, to discriminate between a continuous light and an intermittent one, as shown by its eventually stabilizing them at different points on its retina.

I am indebted to Dr G. McCann, California Institute of Technology, for the gift of the first pinlights used in these experiments.