Copyright (c) 1998, John Michael Williams. Permission of the author is granted
to make (download) a single copy of this explanation
for personal, noncommercial use. All other rights are reserved.
Modification or redistribution of all or part of this work in any form is forbidden: Please contact the author for permission to post or copy this work for any purpose.
This explanation is adapted from J. M. Williams's 1980 doctoral
dissertation on the Hess effect (UMI #82-13538).
Several illusions in vision depend on dichoptic presentation of stimuli such that the time- course of neural activity in one retina (and associated central projections) is made different from the corresponding time-course in the other. The response to stimulation thus is made sooner or more rapidly for one eye than for the other; this difference in response may be considered to involve both a binocular retinal latency difference and a difference in rise- and decay-time of events at some central perceptual site(s) at which the different responses of the two eyes initiate whatever behavioral or (electro)physiological change is being used to study the illusion in question.
Two illusions of the type just described are the Pulfrich effect and the Mach-Dvorak effect. In the context of object-perception, both of these effects involve a single target-object which moves back-and-forth, usually along a straight path, in a horizontal direction parallel to the observer's frontal plane.
With no illusion present, the target is seen, as might be expected, to be moving back-and- forth at a single, well-defined distance; this distance may be estimated as seen by asking the observer to set a comparison object to the same distance in depth as the nearest point on the apparent path of the moving target. With no illusion present, such distance-settings measure the localization acuity of the observer under the given experimental conditions (Lit, 1960c, 1964; Lit & Hamm, 1966; Lit & Vicars, 1970; Lit & Finn, 1976).
In the Pulfrich effect (Pulfrich, 1922; Lit, 1949), as measured by distance settings, a filter placed before one eye makes the target appear to be displaced in depth either toward or away from the observer: While the two images of the target move across their respective retinas, the latency at any given point in the unfiltered eye evidently remains shorter than that of the corresponding retinal point in the filtered eye; this latency difference advances the neural effect in the unfiltered eye relative to that in the filtered eye.
At any given instant, the same, unique target thus corresponds to two (retinotopically) spatially displaced neural responses (e.g., at the ganglion cell layer of the retina), which responses determine all subsequent responses of the observer, including distance-settings.
The retinotopically advanced displacement of the neural effect in the unfiltered eye evidently is processed by the visual system as though it originated from a retinal image disparity. The illusion thus occurs because the stereoscopically fused images of the moving target cannot be discriminated from the images of an object actually displaced in depth from the real path of target motion. The obtained direction of apparent displacement of the target from the real path at any given moment, is found to depend on which eye is filtered and on the direction in which the target is moving.
Julesz and White (1969) have shown that a filter over one eye can facilitate stereopsis under certain circumstances just as would a delay in stimulus presentation to that eye.
In explaining precisely the obtained magnitude and direction of the Pulfrich effect (e.g., Lit, 1949), it may be assumed that, at any given instant, the two eyes are converged to some arbitrary fixation point and that the two images of that fixated point fall directly on the two foveas.
A visible object farther away than the fixated point will be imaged on the two retinas also--but the images of the more distant object will not fall directly upon their respective foveas: Instead, there will be a retinal disparity in the locations of the images; if the disparity be great enough, the more distant object will be seen "double." If the eyes then be converged to the distant object just mentioned, the previous fixation point, in turn, itself may be seen "double," but the newly-fixated object, will be seen "single," there being no retinal disparity.
So, by this purely geometrical approach, it can be understood that the retinal disparities of nearby objects must differ (for any given fixation distance) systematically from those of distant objects. It may be assumed that persons with normal binocular vision learn early in life to interpret these different disparities as discriminative cues to differences in distance of the objects to which the images correspond.
Normally, for a given convergence of the eyes, a moving object, at a well-defined, constant distance will be associated with a well-defined, constant, retinal disparity of the object's dual images on the two retinas. This disparity will remain about constant although both images of the object be in motion across the two retinas. But if a filter be placed before one eye, then, among other things, the response of that eye will become slower than that of the unfiltered eye (evidence for increased visual latency as a function of decreased retinal illumination is given elsewhere).
Thus, an additional disparity in the two retinal images will be created, as it were, in terms
of the neural responses to the two retinal images. For constant effect of the filter, and for
constant target speed, the effectively added disparity caused by the filter will be constant,
too: So, we have
For example, in the Pulfrich effect, suppose an object happened to be moving from left to right, at a constant distance from the observer. Then no matter what the (fixed) distance to which the eyes happened to be converged, any given instant, the dual images of the object, reversed by the optics of the eyes, would be moving right-to-left with some particular retinal disparity proper to the real distance between object and observer.
Now, if a filter happened to be before the left eye, the response in the right eye effectively would be advanced somewhat. The advance would change the percept so that, for judgement of depth, the image in the right eye were advanced farther to the left than otherwise it would. have been. This change in disparity would be the same as the change caused by removing the object to a distance farther away from the observer. Thus, by the geometry, placing a filter before the left eye would be expected to increase the binocularly perceived distance of an object moving left-to-right, and this is just what is observed. The magnitude of the change in disparity caused by a filter is found (Lit, 1960a, 1960b) approximately to be proportional to the speed of the moving object.
This finding suggests, as above by purely geometrical argument, that the given filter must in fact induce a given constant retardation (latency difference); alternative explanations such as constant spatial displacement of retinal response thus are ruled out, although refinements based on target spatial-frequency spectrum (Wist, Brandt, Diener, & Dichgans, 1977) are not. The validity of the geometrical explanation is discussed elsewhere, but as an approximation under very general conditions it is reasonably accurate.
Applying the geometrical argument to the classical Pulfrich simple pendulum, it is shown that the pendulum bob must appear to describe an almost-elliptical trajectory in an approximately horizontal plane, the perceived apparent path enclosing the real path and intersecting the latter near the extremes of the target's horizontal traverse; at these extremes, the target speed goes to zero and the direction of target motion is reversed (Pulfrich, 1922). Distance-settings have been used to map the apparent path and have confirmed this geometry to be very closely correct (Weale, 1954; cf. Rogers, Steinbach, & Ono, 1974).
It should be noted that Lit's apparatus is designed to present a CONSTANT-SPEED target; so, the Pulfrich apparent displacement is a constant over the entire visible traverse of Lit's target. Thus, the ellipse reported for the historical Pulfrich effect, caused by a dangling pendulum bob, can not be seen by Lit's experimental observers. Lit's apparatus permits study of the latency of the human visual system unconfounded by the phenomenal shape of the classical elliptical path.
In the Mach-Dvorak effect (Lit, 1978; Burr & Ross, 1979), no filter need be used. The binocular difference in time-course of central processing is induced directly by presenting the target intermittently, but with stimulus onset delayed slightly for one eye relative to the other: A temporal phase-difference in stimulation thereby has the same effect as an interocular latency difference, either of which may alter perception just as would a binocular spatial disparity.
No latency of response is needed to explain the Mach-Dvorak effect if exposure duration for each eye be brief (Lit, 1978). In the Mach-Dvorak effect, as in the Pulfrich effect, above, if the target should happen to be in simple harmonic motion, it will be seen to move approximately in a horizontal, apparently elliptical path, the departure in depth from the real path at any instant being governed by the direction of target motion and by the eye chosen to receive each stimulus onset first (Dvorak, 1872).
|Burr, D. C. & Ross, J. (1979)||How does binocular delay give information about depth? Vision Research, 1979, 19, 523 - 532.|
|Dvorak, V. (1872)||Uber Analoga der personlichen Differenz zwischen beiden Augen und den Netzhautstellen desselben Auges. Sitzungsberichte der Bohmischen Gesellschaft der Wissenschaften in Prag, 1872, Jan - Jun, 65 - 74 (Presented by Prof. Mach).|
|Julesz, B. & White, B. (1969)||Short-term visual memory and the Pulfrich phenomenon. Nature, 1969, 222, 639 - 641.|
|Lit, A. (1949)||The magnitude of the Pulfrich stereophenomenon as a function of binocular differences of intensity at various levels of illumination. American Journal of Psychology, 1949, 62, 159 - 181.|
|Lit, A. (1960a)||Magnitude of the Pulfrich stereophenomenon as a function of target thickness. Journal of the Optical Society of America, 1960, 50, 321 - 327.|
|Lit, A. (1960b)||The magnitude of the Pulfrich stereophenomenon as a function of target velocity. Journal of Experimental Psychology, 1960, 59, 165 - 175.|
|Lit, A. (1960c)||Effect of target velocity in a frontal plane on binocular spatial localization at photopic retinal illuminance levels. Journal of the Optical Society of America, 1960, 50, 970 - 973.|
|Lit, A. (1964)||Equidistance settings at photopic retinal-illuminance levels as a function of target velocity in a frontal plane. Journal of the Optical Society of America, 1964, 54, 83 - 88.|
|Lit, A. (1978)||Spatio-temporal aspects of binocular depth discrimination. In S. J. Cool & E. L. Smith (Eds.), Frontiers in visual Science. New York: Springer-Verlag, 1978, 396 - 424.|
|Lit, A. & Finn, J. P. (1976)||Variability of depth-discrimination thresholds as a function of observation distance. Journal of the Optical Society of America, 1976, 66, 740 - 742.|
|Lit A. & Hamm, H. D. (1966)||Depth-discrimination thresholds for stationary and oscillating targets at various levels of retinal illuminance. Journal of the Optical Society of America, 1966, 56, 510 - 516.|
|Lit, A. & Vicars, W. M. (1970)||Stereoacuity for oscillating targets exposed through apertures of various horizontal extents. Perception and Psychophysics, 1970, 8, 348 - 352.|
|Pulfrich, C. (1922)||Die Stereoskopie im Dienste der isochromen und heterochromen Photometrie. Die Naturwissenschaften, 1922, 10, 553 - 564; 569 - 574; 596 - 601; 714 - 722; 735 - 743; 751 - 761.|
|Rogers, B. J.,
Steinbach, M. J. & Ono, H. (1974)
|Eye movements and the Pulfrich phenomenon. Vision Research, 1974, 14, 181 - 185.|
|Weale, R. A. (1954)||Theory of the Pulfrich effect. Ophthalmologica, 1954, Separate #128(6), 380 - 388.|
|Wist, E. R., Brandt, T.,
Diener, H. C., & Dichgans, J. (1977)
|Spatial frequency effect on the Pulfrich stereophenomenon. Vision Research, 1977, 17, 391 - 397.|
The Pulfrich Effect, SIU-C. Last updated 2003-12-18