How Missing Information Keeps Us Sane

You, as a reasonably typical human, miss more things visually than you would likely care to admit. How long has that chip been on your windshield anyway? When did the ad on this site change? Was it while you were looking at this post? (Disclaimer: I have no idea.) Sensory input is strange in this way: the brain protects us from pure data streams, so missing the initial windshield chip or the changing ad are not flaws, but shrewd adaptations. This upside, or the identification of it as such, comes courtesy of new research from teams at the University of California, Berkeley, and the Massachusetts Institute of Technology, published today (but not yet online) in the journal [" target="_blank">Nature Neuroscience](http://

What the researchers discovered is the existence of a sensory "continuity field" that merges similar objects viewed within a 15 second timeframe. "The continuity field smoothes what would otherwise be a jittery perception of object features over time," said Berkley's David Whitney, the study's senior author. "Essentially, it pulls together physically but not radically different objects to appear more similar to each other. This is surprising because it means the visual system sacrifices accuracy for the sake of the continuous, stable perception of objects."

What happens without that is an effective overloading of the brain's processing ability, a hypersensitivity to changing shadows and sudden movements. We'd see a world continually morphing, and the effect would be akin to being on hallucinigenic drugs, according to the authors. The world would swim. Sort of like this:

How the research team figured this out is interesting in itself. The researchers had participants look at bars on a computer screen. Every so often, the bars would change their orientation by some angle. This would happen several times over the course of a sitting, and after each change, the participants would estimate the angle by which the bars reoriented. The result was that instead of aligning the bars with the most recent shift as instructed, the participants estimated the change to instead be an average of all of the changes they'd witnessed. The brain, instead of storing a discrete flipbook of image memory, merged them together, creating a sort of meta-vision resistant to information overload.

"Even though the sequence of images was random, participants' perception of any given image was biased strongly toward the past several images that came before it," said MIT's Jason Fischer, the study's lead author. The term for it is "perceptual serial dependence."

To test things further, the team repeated the experiment, but with the changes in angles appearing at either end of a computer screen. The effect was not repeated, suggesting that continuity fields are limited to changes that appear in the same or very close to the same place.