A Functional Account of Motion-Induced Blindness

In motion-induced blindness (MIB), salient objects in full view can repeatedly fluctuate into and out of conscious awareness when superimposed onto certain global moving patterns. Here we suggest a new account of this striking phenomenon: Rather than being a failure of visual processing, MIB may be a functional product of the visual system’s attempt to separate distal stimuli from artifacts of damage to the visual system itself. When a small object is invariant despite changes that are occurring to a global region of the surrounding visual field, the visual system may discount that stimulus as akin to a scotoma, and may thus expunge it from awareness. We describe three experiments demonstrating new phenomena predicted by this account and discuss how it can also explain several previous results. Many of the most exciting phenomena in the study of perception arise because of a disconnect between distal stimuli and their associated visual percepts. Perhaps nowhere are such disconnects greater than in the study of visual awareness itself, as observers can completely fail to be consciously aware of objects and events that are right in front of them. Vision scientists have now uncovered many ways to make clearly visible objects and events ‘‘invisible’’ in this way (Kim & Blake, 2005). However, most such methods rely on weakening observers’ visual processing from the outset—by displaying stimuli especially rapidly (e.g., in the attentional blink or repetition blindness), to each eye in a different manner (e.g., in binocular rivalry), in the periphery (e.g., Troxler fading), or in an unexpected manner while attention is otherwise engaged (e.g., in inattentional blindness). Recently, however, vision scientists have also uncovered a striking case of stimuli disappearing from awareness in what seems to be a more pedestrian context. MOTION-INDUCED BLINDNESS In motion-induced blindness (MIB), a target stimulus may disappear and reappear from conscious awareness when it is presented along with a global motion pattern (Bonneh, Cooperman, & Sagi, 2001). This disappearance can occur repeatedly, for surprisingly salient objects, and even when observers are fully knowledgeable about the relevant manipulations. Figure 1 illustrates a dynamic display that can produce MIB. While the observer fixates the concentric circles at the center of the display, the target disc in the upper left corner might repeatedly disappear from awareness for several seconds as the grid of crosses rotates. (In the actual color displays, this effect is even more striking, because the disappearing target disc is bright yellow and thus even more easily distinguished from the white fixation circles and dark-blue crosses. Animations of this phenomenon and of the other manipulations reported in this article can be viewed on-line at http://www.yale.edu/perception/MIB/.) The initial studies of MIB (Bonneh et al., 2001) found that the frequency or duration of the disappearances could be increased by manipulating a number of characteristics of both the target (e.g., higher contrast, smaller size, or less motion) and the mask (higher contrast, more mask elements, or greater speed). Since the initial report of this phenomenon, further research has characterized its dependence on additional types of lowerlevel visual properties—for example, demonstrating that MIB is enhanced when the target is placed stereoscopically behind the mask (Graf, Adams, & Lages, 2002), is interrupted by nearby transients (Kawabe, Yamada, & Miura, 2007), and is affected by factors such as target size, boundary length, and target-mask similarity (Hsu, Yeh, & Kramer, 2004, 2006). At the same time, other results have characterized MIB as a more central process. For example, the targets rendered invisible by MIB can still be processed in various ways (Mitroff & Scholl, 2004)—fueling negative afterimages (Hofstoetter, Koch, & Kiper, 2004) and orientation adaptation (Montaser-Kouhsari, Moradi, Zandvakili, & Esteky, 2004), undergoing grouping (Bonneh et al., 2001), and contributing to continually updated representations of objects persisting over time (Mitroff & Scholl, 2005). MIB cannot be Address correspondence to Joshua New or Brian Scholl, Department of Psychology, Yale University, Box 208205, New Haven, CT 065208205, e-mail: [email protected] or [email protected]. PSYCHOLOGICAL SCIENCE Volume 19—Number 7 653 Copyright r 2008 Association for Psychological Science fully explained by appeal to sensory suppression or adaptation, because it occurs even with slowly moving targets (Bonneh et al., 2001), more readily for higher-contrast targets than for lowercontrast targets (Bonneh et al., 2001), and within seconds of display onset (Hsu et al., 2004). In addition, signal detection studies involving stimuli rendered invisible by MIB suggest a role for both decisional processes and sensitivity (Caetta, Gorea, & Bonneh, 2007). Explaining Motion-Induced Blindness Much of the interest generated by MIB is due to its mystery: Why does it occur at all? Other examples of perceptual disappearances can now be largely explained in terms of factors such as interocular suppression (in binocular rivalry; e.g., Blake, 1989), boundary adaptation (in Troxler fading; e.g., Krauskopf, 1963), lack of attention (in change blindness and inattentional blindness; e.g., Mack & Rock, 1998; Rensink, O’Regan, & Clark, 1997), or delayed attentional engagement (in the attentional blink; e.g., Nieuwenstein, Chun, van der Lubbe, & Hooge, 2005). None of these factors, though, seems to account for MIB—in which even moving, attended, and fully visible objects may frequently disappear. To date, we can distinguish four approaches that address various aspects of MIB: Attentional competition. Inspired by similarities between MIB and impairments of visual awareness such as simultanagnosia (e.g., Rafal, 1997), Bonneh et al. (2001) initially suggested that MIB may reflect a disruption of the attentional competition that normally determines what observers are and are not aware of. In MIB, this disruption may be due to the salience of the mask, which slows attentional shifts and produces a winner-take-all competition for awareness (Bonneh et al., 2001; Keysers & Perrett, 2002). Interhemispheric rivalry. Because MIB shares features such as oscillation dynamics with binocular rivalry (Bonneh et al., 2001; Carter & Pettigrew, 2003), the same underlying interhemispheric competition may contribute to both phenomena (Carter & Pettigrew, 2003; Funk & Pettigrew, 2003). This possibility is supported by evidence that transcranial magnetic stimulation can enhance or disrupt MIB depending on the hemisphere to which it is applied (Funk & Pettigrew, 2003). Boundary adaptation. Analogies between MIB and perceptual filling in have suggested to some researchers that the same underlying mechanism of boundary adaptation may be involved in both phenomena (Hsu et al., 2004, 2006). Surface completion. Inspired by the demonstration that the relative stereoscopic depth of the mask and target can affect MIB, other researchers have suggested a link to visual surface processing: Perhaps the mask elements are integrated into a single visual surface, which is then taken to occlude the static target ‘‘underneath’’ (Graf et al., 2002). Each of these accounts has been supported by impressive experimental demonstrations, and the perspective we offer in this article is not meant to compete with them. However, we suggest that these accounts do a better job of explaining the details of how MIB works than of why it occurs in the first place, perhaps because many of them are closely tailored to specific experimental results. Of course, these accounts are not mutually exclusive, and several of them may be combined, but even collectively they still do not give a clear sense of why MIB occurs. One reason for this may be that some of these accounts treat MIB at least implicitly as a failure of visual processing— due to overloaded attention, sensory overstimulation, or interhemispheric competition, for example. In contrast, we suggest that MIB could represent a functional response in visual processing; that is, it may be an example of the implicit ‘‘logic of perception’’ (Rock, 1983), rather than a failure to cope with the visual input in some way. Time Fig. 1. Still frames of a typical motion-induced blindness display (not to scale). Observers fixate the center circles while attending to the target disc in the upper left quadrant. The mask made of crosses rotates smoothly counterclockwise. In the typical display, the center circles are white, the target is bright yellow, and the crosses are blue. It remains an open question, however, whether the sort of functional account offered here could be applicable to other forms of perceptual disappearance, beyond MIB. Disappearances can arise from many different factors (Kim & Blake, 2005), and in some situations there may simply not be any additional explanation for why the stimuli disappear. Some other forms of perceptual disappearance, for example, are less temporally stochastic than MIB, being more refractory, time-locked responses to various kinds of transients (e.g., Kanai & Kamitani, 2003; May, Tsiappoutas, & Flanagan, 2003; Wilke, Logothetis, & Leopold, 2003); these phenomena may simply be a result of ‘‘low-level manipulations, directly impacting the early sensory representations’’ (Wilke et al., 2003, p. 1051). In other words, such disappearances may simply be an artifact, or inadvertent consequence, of the mechanics of how sensory representations are formed and maintained, rather than the result of a type of perceptual ‘‘logic.’’ 654 Volume 19—Number 7 Motion-Induced Blindness