Storage Limits In The Brain’s Visual Hard Drive

Todd and Marois used functional magnetic resonance imaging (fMRI), a technique that reveals the brain regions active in a given mental task by registering changes in blood flow and oxygenation in these regions, to identify where the capacity limit of visual short-term memory occurs.

The brains of research participants were scanned with fMRI while they were shown scenes containing one to eight colored objects. After a delay of just over a second, the subjects were queried about the scene they had just viewed.

While the subjects were good at remembering all of the objects in scenes containing four or fewer objects, they frequently made mistakes describing displays containing a larger number of objects, indicating that the storage capacity of visual short-term memory is about four.

The fMRI results revealed that activity in the posterior parietal cortex strongly correlated with the number of objects the subjects were able to remember. The magnitude of the neural response in this brain area increased with the number of objects viewed up to about four and leveled off after that, even when additional objects were presented.

Importantly, the posterior parietal cortex did not respond differently to the number of objects presented in a scene if the participants were not asked to remember what they had seen. In contrast, regions of the visual cortex in the occipital lobe did respond differently to the number of objects even in the absence of the memory task. Thus, it appears that while some regions of the visual cortex register how much visual information we see, the posterior parietal cortex represents how much of that information we can hold in mind.

“The results suggest that the posterior parietal cortex is a key brain area associated with our impoverished ability to hold a representation of a visual scene in our mind once the scene is out of sight,” Marois said.

Marois is a member of the Kennedy Center for Research on Human Development, Vanderbilt Vision Research Center and the Center for Integrative & Cognitive Neuroscience. The work was supported by a grant from the National Science Foundation.

3 thoughts on “Storage Limits In The Brain’s Visual Hard Drive”

  1. … A farmer is having problems with crows eating his seeds. He takes a gun out to the barn to shoot them, but the crows are smart enough to recognize someone with a Gun as a threat, and all fly out of range until the farmer returns to the house with his gun.

    He calls upon a couple of neighbors, and together they go out to the barn, and the neighbors subequently go back to the house, but the crows don’t return until the farmer goes back to the house.

    He calls upon some more neighbors, who repeat the process.

    It turns out that once he gets over 4 or 5 people out to the barn, and all but one return to the house, the crows have lost count, and become “easy pickings”.

    I think this concept of only being able to distinctly recall four objects in short term memory is a mental function from ages ago, and we see the end results in a variety of forms.

    Phone numbers are usually 3 or four collections of 1, 3 or 4 digits. SSN’s are 3 collections of 2, 3 and 4 digits. IPv4 addresses are 4 8bit numbers, represented as 4 collections of 1-3 digits.

    Even a credit card number is a collection of 4 four digit numbers. Ever notice how hard it is to find where you made a typing error in it when making an online order that asks you not to put spaces or dashes in the number?

    If you look at form ID’s, they can broadly be cattagorized into either short ID strings, (form 1040 for taxes, form 8332 for taxes, etc.) or by “name” (e.g. “Ten Fourty” same form, you can probably come up with others.)

    Then again, probably reminding the choir…


  2. In the novel “Calculating God“, by Robert Sawyer, the alien visitor recalls a show he saw on the Discovery Channel, which claims that humans can perceive the cardinality (i.e. without counting) of up to 5 objects. But after that, we have to count. They determined this with tests to see how fast humans could count objects. If shown one, two, three, four or five objects, we can tell how many are present in roughly the same amount of time. But for six or more objects, the amount of time taken to count varies linearly with the number of objects.

    Anyone know if this show on Discovery Channel actually existed? I suspect it did, as Rob typically does a lot of research to be as factually correct as he can in his novels.

  3. The technical name for perceiving cardinality is “subitizing”. As a quick google search will reveal this constitutes a vast literature (practically all subject groups and stimuli manipulations have been tried; from pigeons through to schizophrenics, from 3D VR stimuli to visual illusions etc.) The subitizing literature then dovetails into the visual search literature (e.g., what radar operaters do all day).

    It seems reasonable this did make the Discovery Channel at some point.

    The other thing is, I often believe that imaging results should be treated with some circumspection. They measure spikes in activation level. What they don’t do so well with is subtle changes and especially subtle inhibitory changes. Which is to say, the logic is actually: big change in a given area of the brain != this is the only part of the brain responsible.

Comments are closed.