Two kinds of cost
Switching tasks costs something in two different ways, and they are studied by different people.
The first is the switch cost measured in cognitive psychology. In laboratory experiments people switch frequently among a small set of simple tasks. Monsell of the University of Exeter, reviewing this work in 2003, summarised the result: responses are substantially slower, and usually more error-prone, immediately after a switch. A chance to prepare reduces the cost but does not remove it. Part of it seems to come from the settings of the previous task carrying over, and part from the time needed to reconfigure.
The second cost is slower and closer to office life. The organisational researcher Sophie Leroy gave it a name in 2009: attention residue.
Attention residue
Leroy’s paper, “Why is it so hard to do my work?”, reported two experiments. Its central claim is that to perform well on a new task, people need to stop thinking about the old one, and that this is hard when the old one is unfinished. Participants who switched away from an unfinished task carried thoughts of it into the next, and their performance on the next task suffered.
A second finding concerned finished tasks. Completing the first task was not, on its own, enough to free attention. What helped was finishing it under time pressure. The interpretation is that a deadline pushes people to close the task in their minds as well as on paper, while a leisurely finish leaves them still turning it over.
Planning the return
Interruptions add a twist. In 2018 Leroy and Theresa Glomb reported four studies on what happens to the interrupting task, the one you are pulled into, rather than the one you leave. When people expected to return to their interrupted work under time pressure, they found it hard to give their attention to the interruption, showed more residue, and did worse on it.
They also tested a remedy, which they called a ready-to-resume plan. Before turning to the interruption, people briefly reflected on the interrupted task and planned how they would return to it. With that plan, residue was reduced and performance on the interrupting task no longer suffered. The plan did not finish anything. It gave the unfinished work a place to wait.
Limits
This is a small body of evidence. Leroy’s studies are controlled experiments, the attention-residue papers come largely from one research programme, and they measure performance on particular tasks over short periods rather than a working day. They did not test email, messaging or any particular kind of break.
What they support is modest and fairly consistent with the older task-switching work. Moving between tasks has a cost; unfinished work tends to keep a share of attention; and a brief plan for how to resume appears to reduce the share it takes.