What Can Be "Perceived" In The Absence Of Visual Awareness?*

Randolph Blake
Department of Psychology/Vanderbilt Vision Research Center
Vanderbilt University, Nashville, TN 37240
U.S.A.

At its inception in the latter part of the 19th century, experimental psychology was ordained as the science of the mind, and one of its chief missions was the study of conscious awareness (e.g., James, 1890). During its adolescence in the early part of this century, however, American psychology renounced its mentalistic heritage and focused narrowly on behavior; "consciousness" was consigned to philosophy. During the second half of this century, cognitive psychology reunited the mind and behavior, in part by successfully casting mental operations within the language of information processing. In so doing, cognitive psychology reclaimed consciousness and awareness as legitimate objects of study. In the last decade or so, cognitive psychology has joined forces with neuroscience to launch a concerted attack on the neural concomitants of conscious awareness. To quote from one of cognitive neuroscience's apostles:

"It is hopeless to try to solve the problems of consciousness by general philosophical arguments; what is needed are suggestions for new experiments that might throw light on these problems... One tries to select the most favorable system for the study of consciousness... At any one moment some active neuronal processes in your head correlate with consciousness, while others do not. What are the differences between them?" (Crick, 1994, Pp 19-20)

So how do we figure out which neuronal processes go with consciousness, and which do not? To paraphrase Crick, what we need are innovative strategies for pursing this question. One powerful, albeit indirect, strategy is to ask what perceptual and cognitive operations the brain can accomplish in the absence of visual awareness. Armed with knowledge about the neural bases of those operations, we are then in a position to delimit the necessary conditions for conscious awareness.

To implement a version of this strategy, one can exploit the phenomenon of binocular rivalry as an inferential, "surgical" instrument for dissecting conscious from preconscious stages of visual information processing.

Binocular Rivalry
Stable binocular single vision is easily disrupted by presenting dissimilar monocular images to corresponding retinal locations. When this happens, the eyes send contradictory messages to the brain about the nature of the object located at a given location in visual space. Faced with this physical impossibility, the brain lapses into an unstable state characterized by alternating periods of monocular dominance which continue as long as the eyes view rival stimuli (Figure 1).

During binocular rivalry, a normally visible, potentially interesting monocular stimulus may be suppressed from conscious awareness for several seconds at a time. To what extent does neural information associated with that stimulus continue to be registered within the brain despite the absence of that stimulus from conscious awareness? Answering this question represents a central theme of much of my recent research. In effect, my colleagues and I use rivalry suppression as a neural "reference point" for determining which aspects of visual information processing remain operational during rivalry suppression and which ones do not. After all, neural activity must be fluctuating at some stage of visual information processing, causing this fascinating dissociation between the unchanging physical stimulus and the changing conscious percept. Where is that stage located, relative to other aspects of visual information processing? The following sections illustrate this strategy in action. By way of preview, some aspects of visual information processing proceed unimpaired during suppression phases of rivalry while others are completely disrupted by suppression.

Visual adaptation to invisible targets
Prolonged exposure, or adaptation, to a visual stimulus can briefly alter the appearance of other stimuli viewed immediately following adaptation. Thus, for example, staring at a set of horizontal contours drifting steadily downward subsequently causes a set of stationary horizontal contours to appear to drift upward for a few moments. (You may have experienced a version of this illusory motion after watching movie credits scroll down the screen.) These temporary alterations in stimulus appearance - called visual aftereffects - are commonly attributed to reduced neural responsiveness produced by exposure to the adaptation stimulus. Consistent with this "neural fatigue" idea, the magnitude of an aftereffect increases with the duration of adaptation. But does a person have to be consciously aware of the adaptation stimulus for an aftereffect to develop? Binocular rivalry provides a means for answering this question.

Imagine presenting an adaptation stimulus to the one eye for, say, one minute. At the same time, the other eye views an entirely different stimulus that engages the adaptation stimulus in binocular rivalry. The viewer will be aware of the stimulus for only a fraction of the total, one-minute adaptation period. Will aftereffect magnitude be weakened proportionately?

For some visual aftereffects, the answer is "no" - suppression has no influence whatsoever on their strength. This is true for aftereffects produced by adaptation to tilted lines (e.g., Wade & Wenderoth, 1978), by adaptation to evenly spaced bars of a given width (Blake and Fox, 1974) and by adaptation to an array of contours moving in a given direction (e.g., O'Shea & Crassini, 1981). These aftereffects, in other words, can arise in the absence of conscious visual awareness of the stimuli that induce them. Interestingly, these three forms of adaptation (orientation, bar size and translational motion) are typically attributed to alterations in responsiveness of neurons within visual area V1, a relatively early stage in visual processing.

Other visual aftereffects, however, are susceptible to suppression - their growth is disabled by rivalry. These include aftereffects produced by adaptation to complex optic flow involving looming and rotation (Wiesenfelder & Blake, 1990; Blake, 1995). Evidently the neural events underlying adaptation to these complex forms of motion transpire at neural sites subsequent to that at which processing is intermittently disrupted during suppression. It may be noteworthy that complex optical flow entails global integration of local motion information. Thus, for example, "looming" motion consists of many different local motion vectors present simultaneously in the visual field (see Figure 2). Rivalry suppression seems to disrupt the "binding" of those local features into coherent, global representations. This binding operation is thought to transpire within visual areas forming a pathway into the parietal lobe, several stages removed from area V1 where local motion is neurally registered.

This strategy of inducing aftereffects during rivalry can be modified to allow presentation of an adaptation stimulus only during periods of suppression: an observer is adapted to a stimulus that is never consciously experienced. Blake (1995) accomplished this by replacing the suppressed stimulus with the desired adaptation pattern for as long as suppression lasts; once the eye returns to dominance, the adaptation pattern is switched back to the original rival target. This "replacement" procedure is repeated for a duration sufficient to produce an adaptation aftereffect. Using this strategy, I have found robust motion aftereffects produced by adaptation during suppression to translational motion (i.e., local motion vectors in a single direction) but not by adaptation to rotational motion (i.e., complex optic flow) presented only during suppression. These results, consistent with earlier findings, underscore the potency of this revised strategy involving adaptation to stimuli presented entirely outside of conscious awareness. In principle, the strategy can be applied to any visual phenomena whose strength or character depends on exposure duration.

Visual priming and binocular suppression
Binocular rivalry can also be fruitfully exploited in experiments involving visual priming. The idea behind priming is simple: exposure to one visual stimulus makes it easier to see other, related stimuli - the initial stimulus "primes" information processing for subsequently presented items.

One potent form of priming involves motion perception. The upper panel in Figure 3 illustrates the two frames of a simple animation sequence. When shown rapidly one after the other, the two frames produce a vivid impression of motion: the two vertically aligned balls shown in frame 1 appear to jump to the positions occupied by the two horizontally aligned balls in frame two. However, observers can see this apparent motion in one of two ways: on some presentations, the two balls appear to move in a "clockwise" direction and on other presentations they appear to move in a
counter-clockwise direction. The motion, in other words, is bistable, which is not surprising as the sequence of events is ambiguous with respect to direction of motion.

As Anstis and Ramachandran (1987) have shown, it is possible to bias motion perception in favor of one particular direction, by adding another animation frame at the beginning of the sequence (lower panel, Figure 3). This new frame 1 contains a "priming" ball whose sudden disappearance following the offset of frame 1 generates clear motion in a direction toward the upper ball in frame 2. Moreover, this unambiguous motion now causes people to see the ambiguous sequence as "clockwise" motion essentially all the time - priming motion influences the perceived direction of ambiguous motion.

Does priming motion remain effective even when the priming ball itself is suppressed from awareness by binocular rivalry? To learn the answer, David Alais and I performed an experiment in which the priming ball was viewed by one eye while a completely different visual stimulus was continuously imaged on the corresponding location of the other eye. Because of binocular rivalry, the apparent motion associated with the removal of the priming ball was sometimes visible and was sometimes invisible. But the visibility of the ball was irrelevant: the invisible apparent motion associated with the priming ball always biased perception of motion for the rest of the animation sequence. Evidently neural processing associated with the priming motion transpires at a preconscious level (Blake et al, submitted).

Another form of priming involves picture naming. People are faster and more accurate at naming objects portrayed in pictures if they have seen those objects previously, even if those prior exposures were incidental (e.g., Tulving & Schacter, 1990). The first exposure, in other words, "primes" processing of the same objects on subsequent encounters. But do observers evidence priming for normally visible pictures erased from conscious awareness by binocular suppression?

Cave, Blake and McNamara (in press) recently found the answer to be "no" - successful visual priming requires that observers actually have conscious awareness of a stimulus at the time of its presentation for that stimulus subsequently to facilitate performance on a speeded naming task. Pictures rendered invisible by rivalry suppression behave like new ones never seen before.

Zimba and Blake (1983) found the same result using a word-priming procedure (Meyer & Schvaneveldt, 1971). In that study, observers judged whether or not a short string of letters spelled a real word. The speed with which this word/nonword decision can be made is shortened (i.e., reaction times are faster) if the observer has previously seen a "priming" word that is semantically related to a word spelled by the letter string. Thus the prime-word "fruit" makes it easier to judge that "apple" is a real word, compared to a condition where the prime-word was, say, "fish." Zimba and I found that this semantic priming effect was measurable only if the observer consciously saw the prime word - words erased from consciousness by rivalry suppression failed to speed performance on the subsequent word/nonword task.

Thus some forms of priming succumb to suppression while others can survive outside of awareness. There are no obvious differences in task demands between these two categories of priming. Rather, the crucial distinction between tasks that prime during suppression and those that do not probably turns on the level of processing required by the prime stimulus. Information about local translational motion is registered relatively early in processing, perhaps as early as primary visual cortex. Form information specifying objects and words, however, is only registered to that level of abstraction within higher brain areas including, possibly, the temporal lobes (e.g., Sheinberg & Logothetis, 1997).

Awareness and Cued Visual Attention
Imagine trying to maintain fixation on a central spot but attentively looking for the abrupt presentation of a target that could appear either to the left or to the right of the central fixation spot - your task is to hit a computer key as soon as this target is seen. Being uncertain about the target's location, your reaction times are relatively slow - uncertainty requires you to distribute attention to both locations. But if a brief cue presented at the point of fixation forewarns the observer about where the RT target will appear, RTs are significantly faster - the cue allows the observer covertly to shift attention to one of the two locations (Posner, 1980). Several of my colleagues and I (Schall et al., 1993) found that this attentional cue is effective only when observers actually see the cue. When presented during dominance phases of rivalry, the cue can successfully direct attention to the correct location and, therefore, speed RT; the same cue presented during suppression is useless, with observers performing as if the cue were never presented.

Now it would be an overstatement to assert that a visual cue must be consciously perceived in order to guide attention. For one thing, our results apply to what is termed "endogenous" attention that is intentionally directed to a location in visual space, not to "exogenous" attention automatically captured by abrupt stimulus onset. Moreover, our results merely show that removal of the cue from awareness using rivalry strips the cue of its effectiveness. It is conceivable that other means of manipulating awareness (e.g., visual masking) might produce different results. At the least, however, our findings demonstrate that the critical neural events mediating cued, endogenous attention occur after binocular suppression.

Final Comment
As Williams James (1890) put it, "We know what consciousness is as long as no one asks us to define it." There's no doubt that conscious awareness is exactly what fluctuates during binocular rivalry: A normally visible, potentially interesting stimulus completely disappears from conscious awareness for several seconds at a time. This intermittent loss of awareness is easy to induce and encompasses a wide range of stimulus conditions, making it uniquely suited for the study of conscious awareness. Rivalry can be used to identify aspects of visual processing that transpire outside of awareness and those that do not. This information, in turn, can guide the search for neural concomitants of binocular suppression, using single-cell recording from alert, behaving animals (Logothetis & Schall, 1989; Sheinberg & Logothetis, 1997). Moreover, rivalry can be incorporated into neural imaging studies, in the attempt to identify brain areas where activity is modulated coincident with fluctuations in dominance and suppression. This is a strategy I am currently employing in collaboration with David Heeger and Alexander Polonsky working at the Stanford University Imaging Center.

Of course, rivalry's usefulness as a "dissecting tool" increases as we learn more about the neural events underlying binocular suppression itself. Phenomenologically, suppression involves the complete erasure of an otherwise complex, interesting stimulus from visual awareness for several seconds at a time. But the remarkable invisibility of a stimulus does not necessarily imply that the underlying neural events are equally remarkable. Indeed, modest shifts in the balance of activity among competing neural representations may be sufficient to trigger alternations in suppression. For that matter, it is conceivable that suppression is occasioned by a temporary disruption in the temporal patterning of activity within populations of neurons, not by reductions in activity level. This second possibility takes its inspiration from the controversial idea that synchronized action potentials among neurons may serve to bind those neural "messages" into a coherent, global representation of visual objects (e.g., Engel et al., 1997). Rivalry could be construed as a failure to bind left- and right-eye views.

For decades binocular rivalry was construed as a curious, laboratory artifact. Ironically, it has evolved into an effective psychophysical tool for dissecting visual information processing. Based on the kinds of evidence summarized in this paper, one can construct a functional diagram summarizing the locus of rivalry within heirarchically arranged processing stages, with both feedforward and feedback connections among stages (Blake, 1995). Because rivalry - by definition - involves fluctuations in conscious awareness, this model and the evidence upon which it is based provide potentially powerful clues in the search for the neural concomitants of conscious visual awareness.

Acknowledgments
Supported by grants from the National Institutes of Health (EY07760 and EY08126). Robert Fox and Emanuel Donchin provided helpful comments on this paper.

Recommended Readings
Blake, R. (2001) Primer on binocular rivalry, including controversial issues. Brain and Mind, 2, 5-38.
Blake, R. & Logothetis, N. (2002) Visual competition. Nature Reviews Neuroscience, 3, 13-23.

References

Anstis, S. & Ramachandran, V.S. (1987) Visual inertia in apparent motion.Vision Research 27, 755-764.

Blake, R. (1997) Binocular rivalry and conscious visual awareness. In Proceedings of the International Conference on Neural Information Processing and Intelligent Information Systems. N. Kasabov, R. Kozma, K. Ko, R. O'Shea, G. Coghill, T. Gedeon (Eds), Springer-Verlag: Singapore, pp 47-50.

Blake, R. (1995) Psychoanatomical strategies for studying human vision. In Early vision and beyond, T. Papathomas, C. Chubb, E. Kowler, A. Gorea (Eds), MIT Press: Cambridge.

Blake, R. & Fox, R. (1974) Adaptation to invisible gratings and the site of binocular rivalry suppression. Nature, 249, 488-490.

Blake, R., Ahlstrom, U. & Alais, D. (Submitted) Perceptual priming by invisible motion.

Cave, C.B., Blake, R. & McNamara, T. (in press) Binocular rivalry disrupts visual priming, Psychological Science.

Crick, F. (1994) The astonishing hypothesis. Charles Scribner's Sons: New York.

Engel, A.K., Roelfsema, P.R., Fries, P., Brecht, M. & Singer, W. (1997) Role of the temporal domain for response selection and perceptual binding. Cerebral Cortex, 7, 571 582.

James, W. (1890) The principles of psychology. Holt & Co: New York.

Logothetis, N. & Schall, J.D. (1989) Neuronal correlates of subjective visual perception. Science, 245, 761-763.

Meyer, D.E. & Schvaneveldt, R.W. (1971) Facilitation in recognizing pairs of words: Evidence of a dependence between retrieval operations. Journal of Experimental Psychology, 90, 227-234.

O'Shea, R.P. & Crassini, B. (1981). Interocular transfer of the motion aftereffect is not reduced by binocular rivalry. Vision Research, 21, 801-804.

Posner, M. (1980) Orienting of attention. Quarterly Journal of Experimental Psychology, 32, 3-25.

Schall, J.D., Nawrot, M., Blake, R. & Yu, K. (1993) Visually guided attention is neutralized when informative cues are visible but unperceived. Vision Research, 33, 2057-2064.

Sheinberg, D.L. & Logothetis (1997) The role of temporal cortical areas in perceptual organization. Proceedings of the National Academy of Science, U.S.A. 94, 3408-3413.

Tulving, E. & Schacter, D.L. (1990) Priming and human memory systems. Science, 247, 310-306.

Wade, N.J. & Wenderoth, P. (1978) The influence of colour and contrast rivalry on the magnitude of the tilt aftereffect. Vision Research, 18, 827-835.

Wiesenfelder, H. & Blake, R. (1990) The neural site of binocular rivalry relative to the analysis of motion in the human visual system. Journal of Neuroscience. 10. 3880-3888.

Zimba, L. & Blake, R (1983) Binocular rivalry and semantic processing: Out of sight, out of mind. Journal of Experimental Psychology: Human Perception & Performance, 9, 807-815.

Figure captions

Figure 1. Example of left- and right-eye stimuli that instigate binocular rivalry. Interocular differences along any number of stimulus dimensions can produce rivalry, including orientation, spatial frequency, direction of motion, color and combinations of these.

Figure 2. Global "looming"comprises multiple local motion vectors. A. Imagine facing a surface textured with black dots. While fixating the "plus" mark in the center of the array, you begin walking toward the surface. B. The arrows denote the motion vectors associated with the "looming" surface. By virtue of their small receptive fields and sensitivity to a given direction of motion, neurons in primary visual cortex (area V1) respond to individual, local motion vectors. In contrast, neurons in higher visual areas (e.g., the medial temporal superior area) receive inputs from an array of direction selective neurons distributed over large regions of the visual field - these neurons thus respond selectively to complex optic flow such as expansion, or "looming." Binocular suppression has no effect on adaptation to individual motion vectors but blocks adaptation to complex fields of motion vectors.

Figure 3. Schematic of apparent motion sequences. A. Two-frame sequence in which the upper ball in frame 1 can appear to appear down and to the right (and the lower ball appears to move up and to the left), or vice versa. With repeated presentations, both alternatives may be seen; motion is ambiguous. B. Three-frame sequence in which the ball in the upper left-hand corner appears unambiguously to move to toward the upper ball in frame 2, causing this upper ball, in turn, to appear to move down and to the right (and the lower ball to move up and to the left); motion is unambiguous because of the priming motion produced by the transition from frame 1 to frame 2.