"I'm doing several things at once"

The word multitasking came from computing, where it described a processor handling several programs. People adopted it for themselves, and it carried a false picture with it. It suggests that several streams are running through your mind side by side.

For tasks that need your attention, that isn't what happens. You can't compose a sentence for the report and a sentence for the supplier at the same moment. Try it and you'll find yourself doing one, then the other. What feels like doing two things at once is rapid alternation. You're on the report, then the email, then the report. At any instant you're doing one thing. The question is how often you change which one, and what each change costs.

Psychologists have studied this for decades under the name task switching. Stephen Monsell summed up the field in a review in 2003. In the experiments, people switch between very simple tasks, such as judging whether a number is odd or even and then whether a letter is a vowel or a consonant. The consistent finding is that responses are substantially slower, and usually more error-prone, immediately after a switch. Researchers call this the switch cost. Giving people warning of the switch reduces it but doesn't remove it. Monsell's explanation is that the mind has to reconfigure itself for the new task, and that some of the settings from the old task linger and interfere.

Rubinstein, Meyer and Evans looked at what makes the cost bigger. In four experiments people either repeated one task or alternated between two, classifying shapes or doing arithmetic. Alternating always cost time. The cost grew as the rules of the tasks became more complex. They describe two steps in a switch. There's a shift of goal, where you decide that you're now doing the other thing. Then there's what they call rule activation, which means loading up the rules for the new task and dropping the old ones.

Those tasks took a second or two each. Yours are far more complex. The rules for the report include its argument so far, who it's for, the point you were building towards, and the three figures you had in mind. Loading those takes minutes, which is the rereading you did four times today.

The clearest demonstration that attention can't be shared comes from driving. David Strayer and colleagues put people in a driving simulator, sometimes while holding a phone conversation. When people were talking on the phone their braking reactions were slower, and they had more collisions. This held whether the phone was hand-held or hands-free.

The study also tested drivers who were drunk at the American legal limit, and on the researchers' measures the phone users were the more impaired of the two. Driving feels automatic and so does talking. Put together, each takes attention from the other. In the UK it's illegal to hold and use a phone while driving. Hands-free use is legal, but in the simulator it was no safer.