How the brain limits our ability to multitask

Multitasking refers to the simultaneous performance of two or more tasks, switching back and forth between different tasks, or performing a number of different tasks in quick succession. It consists of two complementary stages: goal-shifting, in which one decides to divert their attention from one task to another, and rule activation, by which the instructions for executing one task are switched off, and those for executing the other are switched on. Multitasking involves dividing one's attention between the tasks, and because each task competes for a limited amount of cognitive resources, the performance of one interfers with that of the other. The greater the similarity of the tasks, the more interference there is, but there is also interference between completely dissimilar tasks.

We know well that it is very difficult to concentrate fully on more than one task; researchers are now beginning to gain an understanding of the neural bases of the limits of multitasking (and some hope to overcome them with augmented cognition). Recent neuroimaging studies in which participants switch between one task and another have implicated several regions of the frontal cortex as bottlenecks to the processing of information. It is emerging that multitasking places excessive demands on executive control centres in the frontal lobe. Hence, multitasking is counterproductive - not only does completion of all the tasks take longer than if they were performed one at a time, but performance on all tasks is also impaired.

Most studies into the neural bases of multitasking limitations have focused on how motor tasks interfere with tasks in the auditory or visual modalities. In a recent study led by René Marois at Vanderbilt University, published in December in the journal Neuron, participants were first asked to perform two different sensorimotor tasks separately. In the auditory-motor task, they were required to press a button on a computer keyboard in response to an auditory stimulus presented to them, and in the visual-vocal task, they were required to respond verbally to a visual stimulus. Using time-resolved fMRI, the researchers identified number of brain regions that were activated during both tasks. These included the anterior cerebellum, the left intraparietal sulcus, and various frontal cortical regions, including the posterior lateral prefrontal cortex (pLPFC, shown as red spots in the image on the right).

The participants were then asked to perform the two tasks simultaneously or in quick succession. When the two tasks were presented within 300 milliseconds of each other, there was long delay in the time taken to respond during the second task. But when the two tasks were separated by periods of 1-2 seconds, their was only a marginal difference in the reaction times. The responses to the second task were also less accurate when a short interval separated the stimuli than when the interval was longer. This occurs because the presentation of two cognitive tasks in quick succession produces what is called a psychological refractory period. This is an increase in the time taken by the brain to process information when one task follows another in quick succession, leading to the delayed response to one of the tasks.

The brain imaging revealed that this information processing bottleneck does not occur at the perceptual stages of information processing, but at a central stage of processing. It was observed that there was prolonged activity in the pLPFC under the condition in which task 2 was presented 300 ms after task 1. There was a queuing of response activity for each of the tasks being performed - the selection of a response for the second task was postponed until the response for the first task has been executed. Thus, even though the two tasks involve processing information of different sensory modalities - one visual, the other auditory - they are apparently processed in series (one after the other) within the same brain region, rather than in parallel within the same or different regions.

The pLPFC fulfills a number of criteria for being a neural substrate of this information processing bottleneck. Firstly, it is activated by activities involving different sensory modalities. It is, therefore, neither a sensory nor a motor area, but is instead amodal. Secondly, the activity in this region was prolonged when two different tasks were performed in quick succession. Furthermore, the pLPFC overlaps extensively with two other regions of the frontal lobe - the periarcuate region and the inferior frontal junction - which are both believed to be vital for decision-making and the selection of multimodal information relevant to the tasks that are being performed.

A more recent study, published in PLoS One earlier this year, examined interference between auditory and visual tasks. In the study, led by Notger G. Müller of the Cognitive Neurology Unit at Frankfurt's Brain Imaging Center, the participants made two simple decisions on the basis of auditory and visual stimuli. The task involved pressing a button to correctly identify two stimuli - one auditory, the other visual - presented in quick succession. The auditory stimulus was presented first, and the visual stimulus was then presented very soon afterwards. As in the study led by marois, the interval between stimulus presentation was varied, so that, in some trials, the participants were required to identity the visual stimulus while the auditory stimulus was still being processed.

Again, it was found that the auditory and visual stimuli produced overlapping activation of regions in the frontal and parietal lobes, and that there was significant interference when the visual stimulus was presented within several hundred milliseconds of the auditory stimulus. But it was also found that, when there was a small interval between the two stimuli, activity in the prefrontal and middle temporal cortices correlated to the visual stimulus was suppressed. Sensory modality-specific activity in the visual cortex was also reduced, and there was a corresponding impairment in awareness of the visual stimulus.

In this study the ventrolateral prefrontal cortex is implicated as one of the brain's bottlenecks. This region is very close to, but distinct from, the pLPFC. And there are different mechanisms at play: whereas Müller's group observed reduced activity in the ventrolateral prefrontal cortex during performance of dual tasks, Marois et al observed prolonged activity in the pLPFC. Müller and his colleagues hypothesize that reduced ventrolateral prefrontal cortex activity is correlated with reduced attentional capacity during dual task performance, and that the limitations of performance on dual tasks may be due to the brain's inability to segregate the encoding of information if different sensory modalities required for each task. Müller and his colleagues refer to this region as a multimodal "global neural workspace", which may represent a working memory of information relevant to the cognitive task or tasks at hand. Because it can execute only one task at a time, multitasking leads to competition for the resources of the workspace.

The neural mechanisms underlying the brain's bottleneck for multitasking are extremely complex. The ventrolateral prefrontal cortex and the pLPFC are bound to interact with each other, but exactly how is unclear. They may form components of the same system, or they may constitute separate systems altogether. And other frontal lobe regions involved in these processes will undoubtedly be discovered. One thing seems quite clear, however: to be more productive and efficient, do one thing at a time.

References:

Hein, G., et al. (2007). Competing neural responses for auditory and visual decisions. PLoS One 2 doi: 10.1371/journal.pone.0000320. [Full text]

Dux, P. E., et al. (2006). Isolation of a central bottleneck of information processing with time-resolved fMRI. Neuron, 52: 1109-1120. [Full text]