The Crawford Effect

Thus far we have considered the so-called masking effects produced by the presentation of brief stimuli in rapid succession to neighboring retinal areas. The perceptual effects produced by this mode of stimulation are not essentially different from those produced by the successive presentation of brief stimuli to overlapping retinal regions.

To illustrate this point, consider again an observer sitting in a movie theater watching an animated cartoon of a moving balloon. But instead of the balloon appearing to move {30} across the screen, as in our earlier example, the balloon now appears to be moving directly toward the observer. If the observer maintains a steady gaze, the objective stimuli falling on his retina will consist of a series of luminous circles, each larger than and concentric to its predecessor. These stimuli will not, of course, be perceived as so many discrete events but rather as a continuous motion of a single “object.” The motion experienced under these conditions is somewhat different from that produced by the successive presentation of a series of identical stimuli to adjacent regions of the screen.   In the former case, movement information is ambiguous in that is may be equally associated with either radial growth or with “approaching” movement of a single “object.” Nevertheless, as in the case of beta movement, the perception of radial movement is not destroyed, but its direction merely reversed, when the order of stimulus presentation is reversed. The perception thus produced will be that of an object either receding or contracting radially.

It is apparent, therefore, that essentially the same conclusion drawn from our earlier example of the horizontally moving balloon is applicable here: It is not true that the earlier frames of the balloon film are “phenomenally absent” or “masked” by subsequent frames. Just as the concepts of {31} successive presentation of brief visual stimuli to overlapping retinal areas.

If a TS and MS consisting of two concentric disks of light are successively presented to a S’s retina, then, in order for the S to be able to detect the presence of the smaller disk (the TS), the intensity of the latter must be gradually increased as the TS and MS onsets approach coincidence. This phenomenon is referred to, in this context, as the Crawford effect. Virtually all attempts to explain this phenomenon have assumed that increasing the intensity of the TS increased its detectability by either allowing the TS neural effect to break through that of the more massive neural onset effect of the MS (Boynton, 1961), or allowing the TS neural effect to overcome some sort of negative MS neural effect, such as lateral inhibition (Robinson, 1966, 1968).

In addition, most theorists have sought to explain the retroactive nature of the Crawford effect in terms of either stimulus “overtake” of the TS by the MS, resulting from differential neural conduction times (Crawford, 1947; Donchin and Lindsley, 1965), or in terms of stimulus integration resulting from temporal quantitization of the neural input. In the latter case, it is assumed that the TS and MS neural effects are integrated or summated whenever both stimuli occur within the same “time frame”. This time frame, furthermore, is regarded as either stationary or moving with respec1 to the sensory surface (Allport, 1968), and as “triggered” either internally, e.g. by a neurophysiological rhythm, or {32} externally, e.g. by threshold energy (cf. Donchin, 1967).

Boynton’s hypothesis of the peripheral on-effect barrier does not explain why masking increases as MS diameter decreases; one would expect the whole nerve on-effect to be smaller and the local on-effect to remain the same in the latter case. Furthermore, Boynton’s hypothesis cannot explain the subsequent “retrieval” of masked stimuli reported by Robinson (1966, 1968) and by Dember and Purcell (1967).

The lateral inhibition hypothesis of Robinson (1966, 1968) cannot account for the fact that successive masking with overlapping stimuli can be produced with dotted visual noise (DVN) which is known to produce little or no lateral inhibition (Uttal, l970a,b,c). This fact invalidates any assumption that lateral inhibition is a necessary condition for the production of visual masking, i.e., conditions sufficient for the production of visual masking need not be accompanied by lateral inhibition.

Explanations of the retroactive nature of the Crawford effect in terms of a neural overtake hypothesis cannot explain why such retroactive masking of a TS can be obtained using a TS of higher luminance than that of the following MS (Krandel, 1958 cited in Boynton, 1961). Neither can the overtake hypothesis explain retroactive masking employing black disks (Kolers, 1962; Sturr et al., 1965) or black patterns (Schiller and Smith, 1965). Any explanations of the retroactive nature of masking that rely upon the assumption that sensory input is temporally {33} quantitized are rendered dubious in light of recent experiments (Haber and Standing, 1970; Efron, l970a, b; Efron, 1973; Bowen et al., 1974) clearly establishing that the durations of auditory, tactile and visual perceptions are not quantitized in the temporal domain. Efron found (1) that the duration of a perception is constant and independent of the duration of the objective stimulus up to some critical value, which Efron refers to as the minimum duration of that particular kind of perception, and (2) that for stimuli longer than this critical duration, the duration of the perception is equal to the objective duration of the stimulus. These findings are impossible to resolve with the notion that sensory input is quantitized in 100 msec frames.

The concept of the neural processing period adequately accounts for both the retroactive and proactive nature of the Crawford effect without having to assume that neural processing involves the packaging of discrete temporal samples of sensory information, and adequately accounts for the “masking” nature of the Crawford effect without having to make any arbitrary assumptions about the number of kinds of neural mechanisms underlying the perception of successive overlapping stimuli. All that is assumed is (1) that the processing of a brief stimulus may be interfered with at any stage during its processing by the introduction of additional stimulus in- formation, (2) that this additional information will be integrated with that from the first stimulus, and (3) that the resulting perception will emerge at the end of the processing {34} of the fastest stimulus component.

Since neural processing time decreases as a function of increasing stimulus intensity, any study of successive masking that uses TS intensity as a measure of masking necessarily confounds the effects of TS-MS contrast and TS-MS asynchrony. Thus, in the case of the Crawford effect, increasing the intensity of the TS facilitates its detection in at least three ways. In the first place, a TS of higher intensity is more likely to be identifiable in any perception produced by the integration of TS and MS energy information (i.e., contrast effects). In the second place, the processing time of the TS is reduced, thereby reducing the period during which TS processing may be interfered with by subsequent stimulation (i.e., interaction effects). Thirdly, increases in TS intensity, by producing an incremental separation between the TS and MS processing periods, supplies the S with cues relating to temporal asynchrony (i.e., asynchrony effects).

As a consequence of this latter fact, TS detection under conditions known to produce the Crawford effect, where TS intensity is the independent variable, may be compared with studies of two-pulse resolution experiments (Keitzman, 1967; Lewis, 1967; Nilsson, 1969) in that cues available in the latter (Keitzman and Sutton, 1968) are probably also avail- able in the former. Thus the TS intensity becomes an index of the amount of asynchrony information available to the S at detection threshold. When considered from this point of view, the Crawford effect may be regarded as a measure of the {35} S’s ability to detect the TS in the presence of TS-MS asynchrony information.

Similarly, as in the case of experiments measuring two- pulse resolution, the presence of asynchrony information while producing some perceptual effects will not necessarily produce the perception of temporal discontinuity. When paired with the TS, the MS may be reported as altered in duration, brightness, or configuration (Keitzman and Sutton, 1968).

The fact that apparent movement may effectively serve as a cue for the presence of multiple stimuli renders immediately explicable the results of experiments purporting to demonstrate the perceptual “retrieval of masked stimuli” (Robinson, 1966, 1968; Dember and Purcell, 1967).

Robinson’s (1966) procedure for the production of backward masking of a central luminous disk (TS) by a larger con- centric luminous disk (MS1) was similar to the stimulus arrangement employed by Crawford (1947). Using 20 msec stimuli, Robinson found that while TS masking was readily obtained under these conditions, the introduction of a second mask (MS2), larger than and concentric to M81 and made to follow MS1, produced as much as a 75% reduction in the masking of the TS. Robinson’s measure of masking was percentage of correct TS detection, a procedure differing from that employed by Crawford (l947) in that in the former case, TS intensity was not manipulated. The advantage of Robinson’s procedure, for the purpose of our analysis, is that the processing times of each of his stimuli can be regarded as essentially constant from {36} trial to trial.

Robinson argued that the phenomenon of successive masking and the phenomenon of masked stimulus retrieval could both be accounted for in terms of the concept of lateral inhibition. However, Robinson’s explanation of the retrieval of masked stimuli by appeal to the concept of “disinhibition,” i.e., inhibition of inhibition, was made plausible by subtle equivocation about the meaning of disinhibition. The concept of disinhibition was introduced by Robinson in order to save his theory that visual masking is caused by lateral inhibition. In the presence of lateral inhibition produced by a double mask, the earlier TS was not masked. What is the reason for this failure to obtain masking? Robinson answers: Inhibition! The inhibitory influence of the first mask was inhibited by the inhibitory influence of the second.

Thus the ad hoc nature of the concept of inhibition (as used by Robinson) becomes clear. Given the concept of inhibition, Robinson can “explain” in advance any possible outcome to his experiment. If the TS is masked by the subsequent introduction of a double mask, his explanation is “inhibitory summation”; if the TS is not masked, his explanation is “disinhibition” or “inhibitory inhibition.”

It might be argued that the concept of disinhibition is not introduced ad hoc, since Robinson defended the physiological reality of disinhibition by appealing to the now classic experiments of Hartline and Ratliff (cf. Ratliff, 1961). But the phenomenon of inhibition-of-inhibition at the receptor {37} level, reported by the latter authors, cannot be used to account for the phenomenon of disinhibition, as this concept is used by Robinson. In the context of Robinson’s experiments, “disinhibition” refers to the recovery of information previously blocked by neural inhibition (specifically lateral inhibition). Thus the concept of disinhibition implies that while the neural effect produced by a local stimulus is inhibited, the information present in that stimulus is somehow retained, making possible its later retrieval (supposedly via disinhibition). The phenomenon of inhibition-of-inhibition studied in the experiments of Hartline and Ratliff, on the other hand, does not entail the possibility that stimulus information falling on an inhibited receptor is somehow “stored” for later “release.”

Furthermore, the appeal to local retinal inhibition as the basis for visual masking is rendered considerably dubious in light of recent experiments by White (1976), who found that the MS is more effective when perceived as in the same place as the TS but stimulated remotely different parts of the retina, than when they stimulated immediately adjacent parts of the retina but appeared spatially displaced. White reports:

Visual masking during pursuit eye movements thus depended on the apparent position of the stimuli, not their retinal positions as such, which is in disagreement with previous studies of visual mask- ing during saccadic eye movements. . . . Retinal position masking after saccadic eye movements may erase previous images, and apparent position mask- ing during pursuit eye movements may make moving targets more visible. (White, 1976, p. 469) {38}

We are, therefore, justified in concluding that Robin- son’s use of the concept of disinhibition is not supported by physiological evidence, and is entirely ad hoc. Robinson’s use of the concept of disinhibition in his “explanation” of visual masking involves the logically unjustifiable procedure of arbitrarily positing the existence of opposing physiological processes, viz. inhibition and disinhibition, thereby opening himself to the same charge leveled against Pavlov’s use of these concepts:

A theory which, without experimental support, posits forces in contrary directions evidently escapes being contradicted by experience since it can al- ways bring into play at the right moment one of the two principles in default of the other. For the same reason it is not susceptible to any experimental justification. (Merleau-Ponty, 1942, p. 59)

The “disinhibition” phenomenon and the conditions under which it may and may not be produced can be adequately explained in terms of the 60-70 msec neural processing period. As was pointed out at the beginning of this section, there is a strong similarity between those spatial and temporal conditions necessary for the perception of beta movement and those necessary for the perception of apparent ra- dial movement. Both effects are produced when geometrically similar stimuli are presented to adjacent or overlapping retinal regions and are separated by an interval great enough for the perception of asynchrony but too small for the perception of successiveness. The perception of apparent movement reaches a maximum when successive stimuli of equal {39} intensity are separated by 100-120 msec. But the experience of apparent movement is often used by Ss as a cue for the presence of the TS when the perceptual task is one of detection or perceived order rather than recognition (Fehrer and Smith, 1962; Thor, 1968; Hirsch and Sherrick, 1961). Thus the same perceptual effect that makes a stimulus difficult to recognize may, nevertheless, serve as a cue for its presence, e.g. in a yes-no or forced choice situation.

In his parametric study of disinhibition, Robinson (1968) found that the phenomenon (1) could not be obtained when MS1 and MS2 are separated by more than 100 msec, (2) could only be obtained over a 100 msec TS-MS1 interval range, and (3) could be consistently obtained over this entire latter range only if MS1 and MS2 are separated by no more than 20 msec. Since all stimulus durations in this study were maintained at 5 msec, SOA and ISI may be regarded as essentially equal for the purpose of the present analysis.

The general and specific findings of Robinson’s study are precisely what would be predicted from our earlier discussion of the relationship between apparent movement and the 60-70 msec neural processing period. When the TS-MS1 interval was exceedingly brief (less than 20 msec) the subsequent introduction of MS2 would have yielded one of the two following perceptions, depending on the MS1-MS2 interval: If the latter interval was small Ss would be expected to see a single disk corresponding in size to that of the MS2. If the intermask interval was large, Ss would be expected to see either {40} apparent radial movement of a disk corresponding in size MS1 , or, at still larger intermask intervals, the perception of a disk corresponding in size to MS1 followed by a disk corresponding in size to MS2. In any case Ss should no able to identify TS when its processing period greatly laps that of the second larger stimulus, MS1 .

When the TS-MS1 interval gradually increased, the interference of MS1 information with TS information would decreased, and the subsequent presentation of MS2 within 100 msec of MS1 and within 120 msec of TS would be expected produce the perception of apparent radial movement of T thereby supplying a cue for the latter’s presence. It possible, therefore, that apparent movement underlies t called disinhibition phenomenon.

Consistent with this assumption are Uttal’s (1970 findings that whereas figural masking by overlapping stimulation can be readily produced using dotted visual noise, neither metacontrast masking nor disinhibition can be produced by this type of stimulation. If it is true that apparent movement accounts for both phenomenon, as we have assumed in this paper, then any set of stimuli that cannot be used to produce metacontrast (type B) should also be inadequate produce “disinhibition.”

It should be pointed out here that recent experin purporting to demonstrate the causal independence of b: apparent movement and metacontrast (Stoper and Banffy, 1977) do not properly take into account the facts (1) that two {41} point resolution is poor in peripheral vision, (2) that two-point resolution is the capacity to spatially differentiate two stimuli, (3) that where spatial differentiation is poor, so is local displacement, and (4) that where the detection of spatial displacement is poor, so is the experience of apparent movement.

These facts suggest that if a briefly presented vertical line is viewed peripherally, its location will be difficult to pinpoint perceptually, and the experience of the same line flashed off and on would be difficult to differentiate from the experience of two adjacent lines flashed on and off alternately. Positional perceptual ambiguity would tend to reduce the illusion of apparent movement in the latter case. Nevertheless, the S will find it difficult to perceive the lines as two discriminable existents.

It is not apparent movement per se that this is the cause of metacontrast masking, but, rather, the fact that the S attributes the movement to one of the other stimuli.   In peripheral metacontrast, it is not movement per se that causes masking of the TS, but the fact that the S attributes the ambiguity of position to the other stimuli.

Latest revision: April 8, 2014 6:29 pm

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.