Every student of introductory psychology knows the pictures: those ambiguous illustrations that demonstrate the flexibility of human perception. Is it a light goblet or two dark profiles? An elegant lady with a feather in her hat, or a vase?
More than amusing visual brainteasers, these figure-ground pictures are important to understanding how people make sense out of their visual environments, and act on what they perceive. Gestalt psychologists, in particular Koffka in 1935 and Metzger in 1953, proposed possible "cues" to guide these perceptual decisions.
Now, three University of Iowa psychologists have serendipitously found and systematically substantiated an important cue: where a stimulus "sits" in a picture, high or low. The result is a rare case of lower trumping higher.
Their study, published in the June issue of the Journal of Experimental Psychology: General (Vol. 131, No. 2), included the results of eight experiments whose results consistently confirmed that people pick stimuli appearing in the lower, not upper, region of a display as the "figure," not the "ground," at rates greater than chance. In fact, says Shaun Vecera, PhD, the study's co-author along with Edward Vogel, PhD, and Geoffrey Woodman, PhD, "Participants found it extraordinarily difficult to try to see the upper region as figure."
The findings reveal the ongoing power of perception to guide us through a world in which everyday visual scenes contain multiple objects that often overlap and partly occlude one another. Knowing how we determine what is figure and what is ground, we can capitalize on the perceptual system's remarkable ability to know what's what.
"Without figure-ground differentiation, one might almost as well be blind," says D. Stephen Lindsay, PhD, of the University of Victoria, British Columbia, Canada, a cognitive psychologist who wrote the introductory text, "Psychology: The Adaptive Mind."
"Figure-ground differentiation is key to the perception of objects--for example, a cup, papers, phone, desk or more dramatically, a tiger, as opposed to a meaningless pattern of colors and edges," he says.
The perceptual system gives figures special treatment: People hold figures in short- and long-term memory longer than grounds; figures seem more salient than grounds; figures have a definite shape, but grounds are shapeless; and figures are perceived as being closer to the viewer.
Although a number of figure-ground cues already were known, the University of Iowa researchers stumbled across the lower-region preference when designing other perceptual studies. When they rotated their left/right figure-ground stimuli, says Vecera, "The lower-region preference really hit us over the head. The sheer salience of this effect drove me to think about its cause." Says Lindsay, "No previous investigator conducted systematic studies documenting and exploring the mechanism."
The finding is "one for the textbooks," concludes Stephen Palmer, PhD, a leader in the study of visual cognition at the University of California, Berkeley. "The human visual system is indeed 'smart' in the sense of being able to come up with the most likely interpretation of visual information."
In their experiments, the authors ran groups of five- to- 12 University of Iowa students through a series of eight experiments that used two-color figure-ground displays to assess participants' figure-ground preferences and investigate their source. The first two experiments empirically established the lower region as a cue to figure-ground assignment, with the lower region perceived as figure on 79 percent and then 71.7 percent of the trials, well above chance. In the third and fourth experiments, Vecera and his colleagues designed displays that didn't resemble any known objects, and demonstrated that the lower-region preference was driven by data and not other aspects of the stimuli. A fifth experiment revealed that people perceived lower regions as figures longer than upper regions, and made significantly fewer reversals for upper/lower displays than left/right displays, again supporting the lower-region cue for figure-ground assignment.
To then inquire about the source of the preference, the researchers' sixth experiment ruled out voluntary spatial attentional differences between the upper and lower visual fields as the source of the results. This important control experiment placed the stimuli entirely within the upper or the lower visual field, and monitored subjects' eye movements to make sure they didn't re-fixate the stimuli. The researchers found that lower-region preference remained for even displays that were entirely in the upper visual field.
The final experiment moved figure-ground displays above or below central fixation points to get at the question of why lower regions appear to be more figure-like than do upper regions. The authors monitored eye movements in order to tabulate data from people who kept their gaze fixed on a central point. Again, placement of the figure-ground display in either the upper or lower visual field did not influence lower-region preference; participants appeared to define "lower" relative to the stimulus configuration--specifically, the horizontal horizon line.
Finally, to rule out any impact of instructing people about figure-ground perception, the researchers did not give participants the usual introductory briefing but rather asked them to determine which of the two shapes in a matching display appeared in the figure-ground stimulus. Participants still matched lower regions faster and more accurately than upper regions.
Cues from the horizon line
Why might lower regions appear to be more figure-like than upper regions? Vecera, Vogel and Woodman favor the idea of a connection between figure-ground assignment and pictorial depth-perception cues.
"The lower-region preference probably reflects a regularity in real-word scenes that our visual systems have become sensitive to through experience," says Vecera. "In real-world scenes, regions that fall below the horizon are physically closer; the perceptual system can use this knowledge to interpret figure-ground displays. The shared contour that separates upper and lower regions acts as a horizon, and we perceive the lower region as the figure because it falls below the horizon line."
Apparently--and not surprisingly--the visual system chooses the most natural interpretation of a scene. Thus, things "lower down" are figures because they are closer to what we have already seen. Vecera says, "Many perceptual phenomena are easily overlooked in everyday life because our visual systems are extraordinarily efficient. The lower-region preference provides a glimpse at one part of this behind-the-scenes work."
Stephen Palmer, PhD, a leader in the study of visual cognition at the University of California, Berkeley, adds, "If we lived in a world in which most objects hung from the ceiling, we would presumably have a bias toward seeing the upper region as closer and figural."
With their findings, Vecera, Vogel and Woodman add a contemporary layer to the Gestalt concept of innate organizing processes, by stressing the flexibility of vision. "There is much information in our visual environments," Vecera observes. "Our visual systems become sensitive to the statistical structure of this information through experience and learning."
Palmer cautions against inferring that the visual system can solve problems or make logical inferences, because, he says, "It may all be done with fairly automatic and 'dumb' mechanisms of statistical inference."
But in any case, it's a solid find for basic science. And as a side benefit, there may be some application to visual design. For example, virtual-reality designers seeking to create the impression of depth should place objects relative to a horizon line. For another, "I'm sure that this principle is used in all kinds of logos," adds Palmer, "if we go back and look."Rachel Adelson is a freelance writer in Raleigh, N.C.