Working memory capacity not 4 but 2+2

October, 2011

A monkey study finds that our very limited working memory capacity of around 4 items reflects two capacities of two items. The finding has practical implications for information presentation.

In the study, two rhesus monkeys were given a standard human test of working memory capacity: an array of colored squares, varying from two to five squares, was shown for 800 msec on a screen. After a delay, varying from 800 to 1000 msec, a second array was presented. This array was identical to the first except for a change in color of one item. The monkey was rewarded if its eyes went directly to this changed square (an infra-red eye-tracking system was used to determine this). During all this, activity from single neurons in the lateral prefrontal cortex and the lateral intraparietal area — areas critical for short-term memory and implicated in human capacity limitations — was recorded.

As with humans, the more squares in the array, the worse the performance (from 85% correct for two squares to 66.5% for 5). Their working memory capacity was calculated at 3.88 objects — i.e. the same as that of humans.

That in itself is interesting, speaking as it does to the question of how human intelligence differs from other animals. But the real point of the exercise was to watch what is happening at the single neuron level. And here a surprise occurred.

That total capacity of around 4 items was composed of two independent, smaller capacities in the right and left halves of the visual space. What matters is how many objects are in the hemifield an eye is covering. Each hemifield can only handle two objects. Thus, if the left side of the visual space contains three items, and the right side only one, information about the three items from the left side will be degraded. If the left side contains four items and the right side two, those two on the right side will be fine, but information from the four items on the left will be degraded.

Notice that the effect of more items than two in a hemifield is to decrease the total information from all the items in the hemifield — not to simply lose the additional items.

The behavioral evidence correlated with brain activity, with object information in LPFC neurons decreasing with increasing number of items in the same hemifield, but not the opposite hemifield, and the same for the intraparietal neurons (the latter are active during the delay; the former during the presentation).

The findings resolve a long-standing debate: does working memory function like slots, which we fill one by one with items until all are full, or as a pool that fills with information about each object, with some information being lost as the number of items increases? And now we know why there is evidence for both views, because both contain truth. Each hemisphere might be considered a slot, but each slot is a pool.

Another long-standing question is whether the capacity limit is a failure of perception or  memory. These findings indicate that the problem is one of perception. The neural recordings showed information about the objects being lost even as the monkeys were viewing them, not later as they were remembering what they had seen.

All of this is important theoretically, but there are also immediate practical applications. The work suggests that information should be presented in such a way that it’s spread across the visual space — for example, dashboard displays should spread the displays evenly on both sides of the visual field; medical monitors that currently have one column of information should balance it in right and left columns; security personnel should see displays scrolled vertically rather than horizontally; working memory training should present information in a way that trains each hemisphere separately. The researchers are forming collaborations to develop these ideas.

Reference: 

[2335] Buschman TJ, Siegel M, Roy JE, Miller EK. Neural substrates of cognitive capacity limitations. Proceedings of the National Academy of Sciences [Internet]. 2011 . Available from: http://www.pnas.org/content/early/2011/06/13/1104666108.abstract

Related News

Here’s an intriguing approach to the long-standing debate about gender differences in spatial thinking. The study involved 1,279 adults from two cultural groups in India. One of these groups was patrilineal, the other matrilineal.

Previous research has found practice improves your ability at distinguishing visual images that vary along one dimension, and that this learning is specific to the visual images you train on and quite durable.

Here’s a perception study with an intriguing twist. In my recent round-up of perception news I spoke of how images with people in them were more memorable, and of how some images ‘jump out’ at you.

Memory begins with perception. We can’t remember what we don’t perceive, and our memory of things is influenced by how we perceive them.

In the first of three experiments, 132 students were found to gesture more often when they had difficulties solving mental rotation problems.

Two experiments involving a total of 191 volunteers have investigated the parameters of sleep’s effect on learning.

Contrary to previous laboratory studies showing that children with autism often demonstrate outstanding visual search skills, new research indicates that in real-life situations, children with autism are unable to search effectively for objects.

When stroke or brain injury damages a part of the brain controlling movement or sensation or language, other parts of the brain can learn to compensate for this damage. It’s been thought that this is a case of one region taking over the lost function.

An imaging study of 10 illiterates, 22 people who learned to read as adults and 31 who did so as children, has confirmed that the visual word form area (involved in linking sounds with written symbols) showed more activation in better readers, although everyone had similar levels of activation i

In a study in which 14 volunteers were trained to recognize a faint pattern of bars on a computer screen that continuously decreased in faintness, the volunteers became able to recognize fainter and fainter patterns over some 24 days of training, and this correlated with stronger EEG signals fro

Pages

Subscribe to Latest newsSubscribe to Latest newsSubscribe to Latest health news