Chapter 4 Which aspect(s) of tracking determine performance?
The number of objects one can track is highly dependent on display characteristics, which hints that the underlying process may be continuous and flexible rather than determined by the fixed, discrete set of pointers hypothesized by Pylyshyn. If so, a person might be able to apply more resource to particular targets to reduce the deleterious influence of a particularly high speed for those targets. There is good evidence for this (e.g. Chen, Howe, and Holcombe 2013), which will be discussed later, but here I would like to explain the resource concept more, and make an important distinction.
To understand why we can track several objects in some circumstances, but only a few in others, we must distinguish between display factors that impose data limitations on tracking, and display factors that impose resource limitations.
The “data” of data limitation refers to sensory data (Norman and Bobrow 1975). If a target moving on an unpredictable trajectory moves outside the edge of our visual field, it is the absence of sensory data that prevents tracking. No amount of mental resources can overcome this for an unpredictable stimulus. Data limitations may also occur when sensory signals are impoverished rather than entirely absent. For example, it is a data limitation that prevents tracking when an object travels at such a fast rate that our neurons hardly register it.
People with poor visual acuity perform less well on many visual tasks than people with high visual acuity, due to differences in the sensory data that they have to work with. Thus, some individual differences are almost certainly due to data limitations rather than variation in tracking processes between people. When performance is data-limited, bringing more mental resources to bear provides little to no benefit. The most popular way of investigating this is by varying the number of stimuli one needs to process, as in visual search studies. If the number of stimuli one must evaluate does not affect how well a person can perform a task, this suggests that the task is data-limited rather than resource-limited, because performance is the same regardless of the proportion of the putative resource can be devoted to it.
Resource-limited processing more interesting for those interested in attention and the capacity limits on mental processing. A classic example is from visual search: if response time or error rate increases with the number of distractors presented, a resource-limited process may be required for success at the task. However, science is hard - an elevation in e.g. error rate can also occur even if there is no resource limitation, if each additional distractor has a non-zero probability of being mistaken for a distractor, yielding more errors with more distractors even if the probability of successfully evaluating each individual stimulus remains unchanged (Palmer 1995).
Even in ideal conditions, where data limitations are avoided, it will be clear (see Section 5) that the number of objects that can be tracked is much less than the number of objects that are simultaneously processed by early visual areas. In other words, there is some sort of resource limitation.
We’d like to know what factors consume the resource. I’ll also be using the term “resource-intensive”, meaning a deleterious stimulus factor that can be compensated for by increasing the amount of resource available. One example is the speed of the moving targets. An increase in target speed can hinder performance, but reducing the number of targets can make up for that because it provides more resource to the remaining targets. Moreover, if one object moves faster than another, it consumes more resource. The evidence for that is that the addition of a fast-moving target hurts tracking performance for a first target more than does the addition of a slow-moving target (Chen, Howe, and Holcombe 2013).
Speed, then, appears to be resource-intensive. Speed also can result in a data limitation, at very high speeds, but long before such speeds are reached, speed is resource-intensive. One should not assume, however, that when one manipulates something about a display, that something is the only thing that changes. Increasing the speed of the objects in a display can also result in more close encounters between targets and distractors, unless one shortens the duration of the trial to equate the total distance the objects travel. Thus, it could be that addressing spatial interference is what consumes resource, rather than speed per se. This brings us to the next Section, which is all about spatial interference.