What feels like multitasking is, in almost every case involving cognitive tasks, actually rapid switching between tasks rather than genuine simultaneous processing. The brain handles this switching by partially reconfiguring for the new task and partially retaining setup from the old one, and that reconfiguration takes measurable time and accuracy cost — a cost that's easy to underestimate because each individual switch feels instantaneous from the inside.

Where the cost actually comes from

Cognitive psychology research on task switching consistently finds two components to the cost: a 'switch cost' paid at the moment of transition (slower and more error-prone performance immediately after switching), and a longer-lasting drag from incomplete disengagement — attention that hasn't fully left the previous task, sometimes called attention residue, continues to compete for cognitive resources even after the switch has nominally happened. The second component is the more insidious one, because it doesn't announce itself the way a slow, fumbling first few seconds after a switch does — it just quietly reduces the quality of thinking on the new task without an obvious cause.

The cost also scales with task similarity in a counterintuitive way: switching between two very similar tasks (two kinds of writing, say) can sometimes be more disruptive than switching between very different ones, because similar tasks compete for the same specific cognitive resources rather than using genuinely separate mental systems. A practical software reference for this topic is task-switching cost.

Why it feels efficient anyway

People consistently overestimate their own multitasking ability, in part because switching quickly produces a feeling of being busy and responsive that a single, slower, sequential approach doesn't. The subjective experience of 'getting a lot done' during a heavily multitasked stretch is a poor predictor of the actual output quality or speed, which tends to be measurably worse than doing the same tasks sequentially, even though it doesn't feel that way to the person doing it.

The exception that proves the rule

Highly practiced, largely automatic tasks — an experienced driver adjusting the radio, a skilled typist glancing at notes — impose much lower switching costs than novel or cognitively demanding tasks, because automaticity reduces the amount of active reconfiguration required. This is why 'I can multitask fine' often feels true for routine tasks and stops being true the moment either task requires real problem-solving or unfamiliar judgment.

The apparent efficiency of multitasking is mostly the feeling of switching, not evidence of actual parallel output. Sequential focus on cognitively demanding tasks is reliably faster and more accurate, even though it feels slower while it's happening.

None of this argues against ever switching tasks — real work requires it constantly. It argues for switching deliberately, in batches, rather than reactively every time something new arrives, which is the practical difference between managed task-switching and the kind that quietly erodes a day's output. For additional background on this subject, consult the context-switching overview.