📘 Why does switching tasks cost more than it seems?

What you’ll learn

1. A Stop Clock and a Column of SumsDescribe the original 1927 task-switching experiment and the counterintuitive result that easier problems lost the most time.Jersild timed alternating columns of arithmetic and found the switching penalty shrank as the sums got harder.
2. You Cannot Prepare It AwayExplain why advance warning reduces the switch cost only to a floor, and what that floor may be made of.Preparation cuts the cost by about half and then stops; part of the work waits for the stimulus, and part is the effort of taking back a task you just suppressed.
3. The Costs Run BackwardsGive evidence that switch costs do not track task difficulty, and separate the two directions of control involved.Switching to the stronger task costs more, in colour words and in language, and setting up the new task is neurally distinct from suppressing the old.
4. The Cost of Being ReadyDistinguish mixing costs from switch costs and judge the three figures most often quoted about multitasking.Merely keeping a second task loaded is costly, the laboratory cost is a fraction of a second, and the famous 40 per cent figure cannot be traced to the study it is credited to.

Questions this course answers

Grounded in trusted sources

• Jersild, A. T. (1927). Mental set and shift. Archives of Psychology, No. 89 (Columbia University) — https://archive.org/details/mentalsetshift00jers (read in full. The five per cent loss figures 69.5, 54, 34, 21, 22 are quoted from the text's own summary 'going from the easy to the more difficult problem situations', and cross-checked against the Combined row of Table II, which gives 54 / 34 / 21 / 22 for Add-Multiply, +6-3, +14-7 and +17-13; the 69.5 belongs to the +1/-1 test added in the later series of 22 adult students, Table IV. Seventy Barnard students took the first four calculation tests; the timing instrument is described as a large laboratory stop clock. NOTE: Monsell (2003) glosses this study as 'adding 3 to vs. subtracting 3 from numbers resulted in dramatic alternation costs; others (adding 3 to a number vs. writing the antonym of an adjective) did not'. Jersild's actual arithmetic pairs were +6/-3, +14/-7, +17/-13, +1/-1 and add-multiply, and his separate opposites/verb-objects test lost about as much on its first trial as the two hardest arithmetic pairs before nearly vanishing on the second. Every number here is taken from Jersild directly, not from the gloss)
• Monsell, S. (2003). Task switching. Trends in Cognitive Sciences, 7(3), 134-140 — https://doi.org/10.1016/S1364-6613(03)00028-7 (citation verified against Crossref. Source for the typical switch cost of 'e.g. 200 ms relative to a baseline of 500 ms'; the residual cost reaching asymptote after about 600 ms of preparation and persisting when five seconds or more are allowed; the Stroop and bilingual switch-cost asymmetries and Allport's task-set inertia account, together with the note that later work complicated it; the Box 2 warning that Jersild's block method confounds switch costs with mixing costs; and the abstract's own limit, that neuroimaging shows extra activation in numerous regions but 'we cannot yet separate controlling from controlled regions')
• Rogers, R. D., & Monsell, S. (1995). Costs of a predictible switch between simple cognitive tasks. Journal of Experimental Psychology: General, 124(2), 207-231 — https://doi.org/10.1037/0096-3445.124.2.207 (the alternating-runs paradigm; the character pair containing a digit and a letter, classified odd/even or consonant/vowel, with the task changing every second trial. The misspelling 'predictible' is in the published title and is reproduced here rather than silently corrected)
• Monsell, S., Sumner, P., & Waters, H. (2003). Task-set reconfiguration with predictable and unpredictable task switches. Memory & Cognition, 31(3), 327-342 — https://doi.org/10.3758/BF03194391 (cited by Monsell 2003 as reference 42 while in press. Source for response-stimulus intervals of 50, 650 and 1,250 ms, the switch cost falling by about half when preparation was extended to 650 ms with little further benefit beyond it, and the switch cost being limited to the first trial of the changed task under predictable switching)
• Arbuthnott, K., & Frank, J. (2000). Executive control in set switching: residual switch cost and task-set inhibition. Canadian Journal of Experimental Psychology, 54(1), 33-41 — https://pubmed.ncbi.nlm.nih.gov/10721237/ (significant switch cost only for alternating ABA series and not for non-alternating ABC series, in response times and in wrong-task errors. Only surnames are used here: PubMed gives the second author as 'Frank J' and the first name was not obtainable)
• Wylie, G., & Allport, A. (2000). Task switching and the measurement of 'switch costs'. Psychological Research, 63(3-4), 212-233 — https://pubmed.ncbi.nlm.nih.gov/11004877/ (the proactive-interference and task-set-inertia side of the argument, and its own conclusion that task-set inertia was not sufficient to account for the results and that they appeared inconsistent with all previous models — part of why the closing card calls the question open rather than settled)
• Hyafil, A., Summerfield, C., & Koechlin, E. (2009). Two mechanisms for task switching in the prefrontal cortex. Journal of Neuroscience, 29(16), 5135-5142 — https://doi.org/10.1523/JNEUROSCI.2828-08.2009 (hybrid spatial Stroop and task-switching paradigm; the anterior cingulate activated by task switches regardless of trial type including congruent trials, and the caudal dorsolateral prefrontal cortex showing a switch cost unique to switches on the second of two successive incongruent trials)
• Smith, L. L., Banich, M. T., & Friedman, N. P. (2019). Individual differences in mixing costs relate to general executive functioning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 45(4), 606-613 — https://doi.org/10.1037/xlm0000613 (N = 749 young adults from the Colorado Longitudinal Twin Sample, mixing costs across three set-shifting paradigms; Mixing Cost factor with Common EF r = .59, 95% CI .43 to .71, p < .001, with Shifting-Specific r = .11, p = .151 and Updating-Specific r = .11, p = .144. The confidence interval not containing 1.0 is the authors' own evidence for processes unique to mixing)

Every Wunder lesson is built from real, reputable sources — never invented.

Related courses