Residual Switch Cost

What it is

A switch cost is the difference in reaction time between a trial where the task changes and a trial where it repeats. The residual is what is left of that difference when the experimenter has given the participant as long as the design allows to prepare. In the 1995 experiments of Rogers and Monsell, participants judged a digit as odd or even and a letter as consonant or vowel, with the task changing on a fixed schedule. Lengthening the interval between a response and the next stimulus, up to 0.6 seconds, reduced the cost of the predictable switch. With 1.2 seconds of preparation a reaction-time cost still remained on the first trial of the new task.

Size matters for how the idea is used, so the figures deserve to be stated. Kiesel and colleagues report that Schneider and Logan (2005) found a residual switch cost of 58 milliseconds at an 800 millisecond preparation interval, against 46 milliseconds predicted by their model. The survey of earlier work in Rubinstein, Meyer and Evans (2001) notes mean costs of about 180 and 135 milliseconds after response-stimulus intervals of 20 and 1,100 milliseconds. These are fractions of a second, measured on simple tasks, and they say how large the leftover can be under those conditions.

Kiesel and colleagues summarise the literature in a sentence: there are usually substantial true task-switch costs, and they decrease if participants prepare (p. 860).

In effect

The reconfiguration account Two-stage models explain the residual as a second reconfiguration process that can begin only once the stimulus appears (Kiesel et al., 2010, p. 855). On this reading no amount of advance warning completes the switch, because part of the work needs the stimulus to trigger it. Rogers and Monsell took this position in 1995, attributing the residual to a reconfiguration component evoked by the task-relevant stimulus.

The interference account Other accounts attribute the residual to interference from the other task (Kiesel et al., 2010, p. 860). Rogers and Monsell also found evidence for this side: costs rose when the stimulus included a character associated with the currently irrelevant task. The two accounts are therefore not clean rivals, since the same 1995 experiments supply support for both.

Where the residual shrinks or does not Verbruggen and colleagues found that brief cue presentation decreased the residual substantially and in some cases completely. Steinhauser and colleagues obtained substantial residual costs despite brief cues, and Lien and colleagues found large costs for some stimulus-response rules even when participants had more than two seconds to prepare (all as summarised by Kiesel et al., p. 856). Nieuwenhuis and Monsell tested whether the residual comes from preparation sometimes not being completed. Strong incentives raised the share of fully prepared trials only marginally, and one summary of that work reads the result as an intrinsic limit on advance preparation. Monsell and Azuma (2000) list several candidate sources, including occasional failure to prepare and two priming effects from earlier trials, and Monsell and Lavric (2014) propose that attentional inertia, a lingering pull toward the previously relevant location, contributes.

A claim the popular books repeat Gazzaley and Rosen's The Distracted Mind states at p. 79 that task-switching costs persist even when a switch is predictable and grow with complexity. The first half is the residual switch cost in one sentence. Our notes on the book record no named researcher behind that sentence, which is why this page and the article it supports give the papers.

What it does not say

It does not say how much time a working day loses to switching. The residual is measured in milliseconds on tasks that take seconds, and the sources read for this page contain no estimate of hours lost in an office. The interruption studies address that question and use different methods.

It does not settle the mechanism. The reconfiguration and interference accounts both have support, and the experiments above come out differently depending on cue timing and the stimulus-response rules used.

It does not say that preparation is useless. Preparation reduced the cost in every set of experiments summarised here. What it did not do, in most of them, was finish the job.

The evidence for the existence of a residual rests on a review and on summaries of individual studies within it. Where this page gives a figure it names the study it came from and the review that reported it.

Who developed it

No single person developed the residual switch cost as a concept. Rogers and Monsell (1995) reported a reaction-time cost remaining after long preparation, and the term residual is standard in the literature reviewed by Kiesel and colleagues (2010). Our confidence in this attribution is medium: the finding is theirs, and the name is the field's.

---