Multitasking and Human Performance

Types of Processing

There are two general categories of multitasking activity. On the one hand, a person may alternate performance between tasks, switching attention between them in discrete units of time. This creates a form of serial processing in which the performance of one task creates a bottleneck, briefly locking out the processing of other tasks. There are obvious decrements in performance associated with this processing bottleneck. An alarming real-world example of attention switching comes when a driver attempts to text message on his or her cellular phone. Here, visual attention can be allocated to driving, or texting, but not both at the same time. This may be an extreme example of a bottleneck that involves both cognitive and structural limitations; however, one can find these limitations separately as well. Another example based on a large body of laboratory research using the psychological refractory period (PRP) paradigm has found that the processing of one task systematically delays the processing of another concurrent task. The processing delay is thought to stem from a central-processing bottleneck in information processing that is stubbornly resistant to practice, and empirical efforts to bypass the bottleneck have largely been unsuccessful. Together, these findings indicate that a fundamental characteristic of the cognitive architecture is a limited ability to perform more than one attention-demanding task at a time.

On the other hand, some multitask combinations may allow parallel processing, in which attention is shared between two or more concurrent tasks. In this context, attention has been conceptualized as a resource that can be flexibly shared between concurrently performed tasks. However, given the limited capacity characteristics of attention, there is a reciprocal relationship between the tasks such that as one task prospers, because attention is allocated to its processing, performance on the other task suffers. In 1984, Christopher Wickens developed a multiple-resource model, suggesting that some dual-task combinations may draw on separate processing resources. This multidimensional framework conceptualizes modalities of input (auditory vs. visual), codes of operation (verbal vs. spatial), and modalities of response (vocal vs. manual) as separable pools of attentional resources, with performance in dual-task situations varying as a function of the overlap in demand for processing resources; the less overlap there is, the more proficient the multitask performance will be. Other researchers have suggested that it may be better to consider multitask performance with regard to the degree of cross-talk between the cognitive operations and mental representations of the separate tasks. Interestingly, the cell phone/driving dual-task combination would seem to represent a fly-in-the-ointment for this class of theories because driving is primarily a visual/spatial/manual task and conversing on a phone is primarily an auditory/verbal/vocal task. That is, even though these two tasks should, in theory, draw on distinct processing resources, they nevertheless produce considerable dual-task interference. It seems then that bottlenecks in performance may still be present even in circumstances where parallel processing can, at least in principle, take place.

...