Our visual memory may be short-lived

Look away from this newspaper and your visual system will lose all memory of this page and these words, argues one psychologist.

By Beth Azar
Monitor staff

The human visual system operates in the eternal present?it has to be reintroduced to an object each time it encounters it, according to a controversial theory of visual perception posited by psychologist Jeremy Wolfe, PhD, and his colleagues.

This idea challenges the traditional theory of visual perception, which says that as people attend to a scene, they build internal representations of objects in the world that remain stable even when they look away, said Wolfe, of Harvard Medical School.

In contrast, he and his colleagues argue that before people look at an object it is represented as a shapeless bundle of basic visual features. With attention, those features can be bound together into some recognizable object. When attention is directed elsewhere, the visual system?s representation of the object falls back to the way it was before you attended to it, explained Wolfe.

He discussed this theory at a symposium on perception that was part of the 'Mind, brain and behavior' series coordinated by APA?s divisions 3 (Experimental) and 6 (Behavioral Neuroscience and Comparative Psychology) and funded in part by the National Science Foundation. Wolfe was also the series? co-coordinator along with Lynn Robertson, PhD.

Attention: the tie that binds

Wolfe began building the case for his theory by explaining studies that demonstrate how important attention is for initially recognizing objects. Such studies typically involve visual search tasks in which participants search for a target object among a group of objects. People can easily pick out certain basic search features. So if search items differ only in one of these basic features, such as color, size, curvature or shininess, they can process the entire scene at once and quickly direct attention to the target object. Indeed, increasing the number of items in the scene does not increase a person?s search time to find the target, said Wolfe.

But most visual searches are not for objects defined by a unique feature such as color or orientation. When people search for a face among faces or a word on a page, they must attend to item after item until they find the target. As a result, search time increases as the number of search items increases.

To illustrate the role of attention in recognizing the shape of objects, Wolfe and his colleagues conducted an experiment using a graphic of an odd-looking chicken. The chicken was designed so that it could be cut into quarters that could be recombined to make objects that had the same general features but were clearly not 'chickens.' They might have two back thigh quarters, a head quarter and a back quarter; two back quarters and two front thigh quarters; two head quarters, a back quarter and a front thigh quarter (see picture). Participants had to find the normal chicken among a number of 'abnormal' chickens.

'People find this task difficult,' said Wolfe. 'Without attention, each object is roughly the same bundle of features. We have used many other sorts of stimuli, but we have never found any evidence that subjects can glean the shape of an object before dropping their attention on to that object.' These studies show what little information people have about the visual world without attention, said Wolf.

'What that means is that if you walk into a room, all of the people will initially be some collection of preattentive features in some shapeless sort of a way,' he said. Seeing symposium participant Art Shimamura in the audience, he went on: 'So Art Shimamura would not become Art until I had actually gotten my attention onto him and bound that shapeless lump into something that could then be delivered to my long-term memory and recognized as Art.

'The question then becomes, what happens when I then deploy my attention elsewhere?' asked Wolfe. 'What happens to Art? Does he remain as a fully bound perceptually recognized object, or does he fall apart again into some shapeless blob of features?'

Art vanishes

Wolfe and his colleagues tackled this question with a different version of the standard search task. Instead of seeing a completely new scene for each search, participants saw the same scene, with a new target placed in the center each time?search objects in this experiment were different colored and sized rectangles. If people are able to bind the features together and keep the results in their 'mind?s eye' after they attend to it, they should become more efficient at spotting the target item as time goes on because the target always appears in the same spot. However, people keep searching at the same inefficient rate, even after 300 searches, said Wolfe.

The same thing happened when the researchers used letters as the stimuli. A set of letters, such as 'C-A-T,' sat statically on a computer screen. On each trial, participants would see another letter and had to determine if that letter was on the screen or not: In the 'C-A-T' example, if the probe letter was 'F', the participant would say 'no,' if it was 'C,' the participant would say 'yes,' and so on for hundreds of trials. People never improved their efficiency. In fact, once they had memorized the set of letters, 'subjects did no better looking at the stimulus than they did not looking at the stimulus,' said Wolfe.

The implication of this finding is that 'when you attend away from the object it simply falls apart on you?it?s not represented fully in its attended fashion anymore,' said Wolfe. 'And that suggests that if I changed it when you weren?t attending to it, you wouldn?t notice the change' until you attended to that object again.

In fact, researchers have found that people have a difficult time noticing dramatic changes in scenes unless they are actively attending to the specific item that changes, said Wolfe.

'You simply do not have a representation in your visual system that represents anything other than the preattentive feature information [of objects in the scene] and the current contents of attention,' said Wolfe. 'Everything else seems to be reintroduced to us every time we see it. You can get away with this because you assume that the world is stable over time. And so you can have the illusion that you?re seeing more than you really are.'

At the moment, this theory of visual perception is 'pretty controversial,' admitted Wolfe. A symposium at the Psychonomic Society annual meeting in November, entitled 'The grand illusion,' will feature researchers with different theories. The resulting debate should be 'quite energetic,' Wolfe said.