In my book on remembering intentions, I spoke of how quickly and easily your thoughts can be derailed, leading to ‘action slips’ and, in the wrong circumstances, catastrophic mistakes. A new study shows how a 3-second interruption while doing a task doubled the rate of sequence errors, while a 4s one tripled it.
The study involved 300 people, who were asked to perform a series of ordered steps on the computer. The steps had to be performed in a specific sequence, mnemonically encapsulated by UNRAVEL, with each letter identifying the step. The task rules for each step differed, requiring the participant to mentally shift gears each time. Moreover, task elements could have multiple elements — for example, the letter U could signal the step, one of two possible responses for that step, or be a stimulus requiring a specific response when the step was N. Each step required the participant to choose between two possible responses based on one stimulus feature — features included whether it was a letter or a digit, whether it was underlined or italic, whether it was red or yellow, whether the character outside the outline box was above or below. There were also more cognitive features, such as whether the letter was near the beginning of the alphabet or not. The identifying mnemonic for the step was linked to the possible responses (e.g., N step – near or far; U step — underline or italic).
At various points, participants were very briefly interrupted. In the first experiment, they were asked to type four characters (letters or digits); in the second experiment, they were asked to type only two (a very brief interruption indeed!).
All of this was designed to set up a situation emulating “train of thought” operations, where correct performance depends on remembering where you are in the sequence, and on producing a situation where performance would have reasonably high proportion of errors — one of the problems with this type of research has been the use of routine tasks that are generally performed with a high degree of accuracy, thus generating only small amounts of error data for analysis.
In both experiments, interruptions significantly increased the rate of sequence errors on the first trial after the interruption (but not on subsequent ones). Nonsequence errors were not affected. In the first experiment (four-character interruption), the sequence error rate on the first trial after the interruption was 5.8%, compared to 1.8% on subsequent trials. In the second experiment (two-character interruption), it was 4.3%.
The four-character interruptions lasted an average of 4.36s, and the two-character interruptions lasted an average of 2.76s.
Whether the characters being typed were letters or digits made no difference, suggesting that the disruptive effects of interruptions are not overly sensitive to what’s being processed during the interruption (although of course these are not wildly different processes!).
The absence of effect on nonsequence errors shows that interruptions aren’t disrupting global attentional resources, but more specifically the placekeeping task.
As I discussed in my book, the step also made a significant difference — for sequence errors, middle steps showed higher error rates than end steps.
All of this confirms and quantifies how little it takes to derail us, and reminds us that, when engaged in tasks involving the precise sequence of sub-tasks (which so many tasks do), we need to be alert to the dangers of interruptions. This is, of course, particularly true for those working in life-critical areas, such as medicine.