A brief mention of an interesting study on multitasking from Cornell University. This is only a press release (the article was apparently published in Psychological Science in October 2006):

A new Cornell study shows that people are pretty good at perceptual multitasking -- except when multiple sources of incoming stimuli are of the same type. Morten Christiansen, associate professor of psychology, co-authored the study with Christopher Conway, a National Institutes of Health research fellow at Indiana University.

Note that this sort of multitasking—simultaneous streams of information—differs from the widely (and mistakenly, I think) criticized sequential-task work that people commonly do on computers—moving back and forth between applications but only working in one at a time.

All of these, of course, beg the question of quality. Work in such studies is constructed as an assembly line activity with the simple goal of processing information as efficiently as possible. One factory worker trying to put a nut on a gizmo on conveyor belt a and trying to put a washer on a widget on conveyor belt b spends a lot of time running back and forth between the belts.

But that sort of metaphor doesn't do "work" justice (any more than it does justice to the intelligence of a factory worker). Most people who do a lot of knowledge work are familiar with the feeling of being in the zone (or "flow," to use Csikszentmihalyi's term). Dealing with a complex problem often requires moving across applications and information spaces, the display of multiple fields of information, and even different modes of work. So there may be some cognitive load to switching among tasks or applications, but it's an "opportunity cost": the quality of work may benefit enough from the task switching and multitasking that the added cognitive load are worth it.

[via Project NML]