Attentional Disengagement Cost

What it is

Michael Posner and colleagues split an attention shift into three operations in 1984: disengaging from whatever currently holds you, moving, then engaging with something new. Letting go is the slow part, and it is slower when the thing holding you is salient, emotional or rewarding, which is why a glance at a feed so often becomes a session. Resumption lag is the same cost measured at the level of a task rather than a stimulus, the interval between leaving a piece of work and being back inside it. The first look is cheap. The return trip is where the money goes.

In effect

Mark's twenty-four workers resumed about three-quarters of interrupted work on the same day, and nine in ten of those returns were started by the worker rather than prompted by anyone else. Internally interrupted work took longer to come back to than externally interrupted work, twenty-nine minutes and one second against twenty-two minutes and thirty-seven seconds, a difference the authors themselves call marginal. Laura Dabbish's 2011 reanalysis of a thirty-six-person dataset reports resumption lags of about thirty-four minutes after a self-interruption and about thirty-two minutes after an external one, with standard deviations roughly twice each figure. The practical reading is the only one that spread allows: budget for the return as well as the switch, and do not plan around an average.

What it does not say

It does not say recovery takes twenty-three minutes. That number is not in the published record, and the figure it displaced carries a spread more than twice its own size, so most returns are nothing like the average. It does not establish disengagement as a separable stage with a cost of its own, which the orienting literature still argues about, and it does not identify a mechanism. Nor does it generalise past office information work, because every observational dataset behind it is technology, finance or research staff watched over the shoulder by a coder.

---