The dots
The standard test is called multiple object tracking, and it began with a 1988 paper by Pylyshyn and Storm at the University of Western Ontario. Participants watched a screen of ten identical objects. A few were briefly marked as targets, then all ten moved around at random, independently of each other. The task was to keep track of the targets and to tell whether a change happened to a target or to one of the others.
People managed this with up to five targets among the ten. The paper’s sharper result concerned how they did it. The researchers set the speeds and distances so that a strategy of moving attention from target to target, checking each in turn, would have produced no better than about 40 per cent accuracy, even on deliberately cautious assumptions about how fast attention can move. Participants were 87 per cent accurate.
Indexes
Pylyshyn’s explanation was that tracking is done in parallel, not by scanning. He had already proposed a model, called FINST, in which the visual system has a small number of indexes it can attach to objects. Each index stays attached as its object moves, so the system can keep referring to that object without working out again which one it is. The number of indexes is limited.
By the 2000s, experiments of this kind were usually described as showing a maximum of about four. That figure, and the idea of a fixed set of slots behind it, became one of the standard descriptions of visual attention.
Not a fixed number
In 2007 Alvarez, at MIT, and Franconeri tested whether the limit really was fixed. In their first experiment they varied how fast the objects moved. At slow speeds, people could track up to eight. At the fastest speeds, they could track only one. A second experiment suggested the limit was tied to the spatial resolution of attention, how finely it can separate objects that come close together.
Their conclusion was that the number of objects you can track is set mainly by a resource that can be divided up flexibly, not by a fixed number of slots. On this view, track fewer objects and each gets more of the resource, which allows them to move faster or come closer together before you lose them.
The specific claim that people can track exactly four, regardless of conditions, does not survive the 2007 result.
What the figure means
“About four” is a fair summary of performance with identical dots at moderate speeds on a laboratory screen. It describes that task well. It is not a measured limit for driving, sport or any other real setting, where objects differ from each other, move predictably and can be followed with the eyes. The studies here did not test them.