Chapter 13 Divided attention

You already know about location selection. You can select an individual location with attention. You can also select multiple locations, and it seems you can select them simultaneously rather than your attention having to switch back and forth between them. However, as we’ll see in the next section, we can only select a very limited number of stimuli at a time, which reflects that we have bottlenecks.

Given how narrow the bottlenecks in our brains are, a critical topic of research has been exactly what kind of processing goes on before signals reach the bottleneck. The study of how we pay attention to moving objects provides some important information about this.

You’ll recall from Duncan’s experiments (3) that higher-level processing is so limited that even identifying a simple feature of two objects yields much worse performance than identifying one. However, if all a person needs to do is keep your attention on multiple locations, without trying to identify things there simultaneously, a person can do pretty well with more than two objects.

13.1 Tracking moving objects

In the real world, objects of interest are often moving,or your eyes are moving, or both. When playing football, for example, while every point in the scene is processed by the first layers of neurons in cortex, certain bits are of special interest to continuously be aware of, like the location of the ball and of a defender. Thus, it pays to keep attention on those bits to ensure they are fully processed.

If all we had was feature selection and location selection, then if we were interested in monitoring a particular object, like the football, or a child playing in the ocean at a crowded beach, every time the object moved, we would have to find it all over again to attend to it. Finding it would likely require a visual search, which as you know from (9) can be time-consuming. Even if it only takes a third of a second, during that time a penalty kick traveling at 30 metres per second will have travelled a full ten metres.

Fortunately, attentional selection is able to follow a moving object (this was mentioned very briefly in 7). This is usually studied with what’s known as the “multiple object tracking” procedure. In multiple object tracking, first a bunch of identical objects appear on the screen. Then, some of them are cued by briefly appearing in a different color or by flashing. The participant’s task is to keep track of the cued objects as they move around. Try it here.

The task feels fairly natural. Indeed, it seems that once one selects an object with attention, if it starts moving, your attentional focus will tend to move along with it. It may be that what we have previously been referring to as location selection may best be thought of as object selection, because if an object in a selected location starts moving, it’s hard to have your attention not follow along and stick to the location instead. Magicians like Apollo Robbins exploit this, moving their hand or another objects of interest with smooth gestures, knowing that if they do that, the viewer’s attention is likely to come along.

The idea that attention selects objects has important implications. For attention to select objects suggests that the visual system must have processed the image preattentively into objects, otherwise attention may not have been able to select them. This is similar to the logic that Treisman used when arguing that processing of individual features occurred prior to the action of attention (preattentively) - otherwise, attention could not go straight to a stimulus with a particular color.

But what does the preattentive visual system consider an object, that attention can then select? This was investigated by VanMarle and Scholl (2003). In what they called a “substance” condition, objects sort of poured from one location to another.

A snapshot of a frame from the ‘substances’ condition of VanMarle and Scholl (2003).

People performed much more poorly when trying to track the pouring substances than they did when tracking the normal objects. Performance was similarly poor in a related ‘slinky’ condition. But the problem was not simply due to the nonrigidity or lack of cohesiveness of the objects, as shown by the fine performance in two control conditions.

Mean tracking accuracy and standard error for each condition. The horizontal line running through the graph represents the performance that would be obtained if one tracked only one object and guessed on the others — 62.5% when tracking four items in a display of eight.

Mean tracking accuracy and standard error for each condition. The horizontal line running through the graph represents the performance that would be obtained if one tracked only one object and guessed on the others — 62.5% when tracking four items in a display of eight.

Experiments like this seem to be revealing how our visual system carves up the world into objects prior to the action of our attention. It also groups moving things based on how they move in relation to one another or are connected to each other. The following animation was taken from here.

You might be able to sometimes see an individual triplet of red, green, and blue as rotating around each other, and at other times see each color as being part of a global circle.

13.2 A temporal limit on tracking

The processes that allow attention to keep up with moving objects are still mysterious. But we do know something about the limits on those processes, as illustrated here. These limits are much slower than those of basic motion perception (Verstraten, Cavanagh, and Labianca 2000; Holcombe and Chen 2012).

13.3 Test yourself

You can test your multiple object tracking more quantitatively at the following site:

Username: education

For the password, see Alex’s Canvas module

The test first tests your forward digit span, then your backwards digit span, tests that you may have heard about in memory and intelligence lectures, and then measures speed thresholds for tracking.

Is there any relation between tracking ability and intelligence? We don’t really know. But in the Testmybrain sample of thousands of people, those who did their degree in a STEM field (science, technology, engineering, or medicine) did better than those who did their degree in another field, such as arts or law. If you took a random STEM person and a random non-STEM person, the STEM person would have a 55% chance of having gotten a higher score on the MOT test. In other words, there are plenty of non-STEM people who have higher tracking performance than the average STEM person, but on average, STEM people do slightly better.

13.4 Exercises

  • What does the fact that attention can select objects have to do with magic tricks?
  • What have you learned about what processing occurs prior to a bottleneck?


Holcombe, Alex O., and Wei-Ying Chen. 2012. “Exhausting Attentional Tracking Resources with a Single Fast-Moving Object.” Cognition 123 (2).

VanMarle, K, and Brian J Scholl. 2003. “Attentive Tracking of Objects Versus Substances.” Psychological Science 14 (5): 498–504.

Verstraten, F A J, P Cavanagh, and A Labianca. 2000. “Limits of Attentive Tracking Reveal Temporal Properties of Attention.” Vision Research 40 (26): 3651–64.