The mind that is not on the task at all
70 min
Two hosts talk the lesson through. The voices are synthetic; the script was written from this lesson and checked against it, and asserts nothing the lesson does not.
- State what the experience-sampling study measured about mind-wandering, with its sample, its method and its date
- Distinguish an interruption from mind-wandering, and say why the second is harder to measure
- Say what the study's time-lag analysis does and does not establish about causation
Lesson 2 was about somebody else taking your attention. This lesson is about the times nobody did.
They're different problems and only one of them has a tally. An interruption has a phone call attached to it, or a colleague, or a notification, and somebody standing behind you could count them. A mind that has drifted leaves no trace at all, which is why the study this lesson is built on had to do something unusual to measure it. That way of separating the two is this course's own rather than a distinction the study draws,2 and it is the reason these are two lessons rather than one.
How you measure a thing with no witness
You ask, at moments the person didn't choose.
Killingsworth and Gilbert built a web application for the iPhone that contacted people at random during their waking hours and asked three things: what they were doing, how they felt, and one question about attention.1
The question, in full: "Are you thinking about something other than what you're currently doing?" Four answers were available: no; yes, something pleasant; yes, something neutral; yes, something unpleasant.
That is the whole instrument. One question, four options, at a moment nobody chose. It's called experience sampling, and the reason it exists is that asking somebody at the end of the day how often their mind wandered gets you a judgement from memory rather than a count.
The sample, and this is a lesson 1 scope point rather than a footnote: "samples from 2250 adults (58.8% male, 73.9% residing in the United States, mean age of 34 years)".1 They were volunteers who had installed a happiness-tracking application on an iPhone, in 2010.
What it found
"Mind wandering occurred in 46.9% of the samples and in at least 30% of the samples taken during every activity except making love."1
The second half of that sentence is the one to keep. Not that the average is high, but that there's a floor: across almost everything these people were doing, at least three sampled moments in ten were spent somewhere else. Mind-wandering was not concentrated in the dull parts of the day.
And the authors put a caveat next to their own headline, which almost never travels with it: "The frequency of mind wandering in our real-world sample was considerably higher than is typically seen in laboratory experiments."1
And there's a second result in the mood data that almost never travels, and it is the one that makes the first interesting. Minds wandered to pleasant topics in 42.5 percent of samples, to neutral topics in 31 percent and to unpleasant topics in 26.5 percent. People were no happier when thinking about pleasant topics than about their current activity, at a slope of −0.52 that did not reach significance, while neutral topics came in at −7.2 and unpleasant at −23.9, both strongly significant.1
So it is not simply that unhappy thoughts make you unhappy. Wandering to something pleasant did not help, which is what makes the authors' framing about wandering itself rather than about its content.
Read the laboratory caveat as a statement about instruments, which is lesson 1's question arriving early. A lab study of mind-wandering interrupts somebody doing a task the experimenter chose, in a room with nothing else in it. This study asked people in the middle of their lives. The two produce different numbers, and the authors say so rather than leaving you to find out.
Before the next section. The study also asked how happy people were at each moment. What do you expect it found about mood when minds were wandering, and what would you need to know before treating that as a finding about cause?
Show the answer
The direction is the easy half, and nearly everybody gets it.
People were less happy when their minds were wandering, and the paper gives it as a slope: "people were less happy when their minds were wandering than when they were not [slope (b) = −8.79, P < 0.001], and this was true during all activities, including the least enjoyable."1
The harder half is what you'd need, and there are two obvious problems with reading that as "wandering makes you unhappy".
The first is direction. Being in a bad mood is a well-documented cause of mind-wandering, so the association is exactly what you'd expect if the arrow ran the other way entirely.
The second is that both could have a third cause. A tedious meeting produces both.
The paper addresses the first directly and the second only partly, and the next section is about how. On the third cause, it reports that the variance explained by mind-wandering was largely independent of the variance explained by activity, which rules out one candidate and not the class.1
The causal claim, and exactly how strong it is
This is the part of the study that travels furthest and arrives in the worst condition, so the sentence is here in full.
"Although negative moods are known to cause mind wandering, time-lag analyses strongly suggested that mind wandering in our sample was generally the cause, and not merely the consequence, of unhappiness."1
Three things in that sentence are doing work.
"Time-lag analyses" means they used the order of the samples: does wandering at one moment predict lower mood at the next, more than low mood at one moment predicts wandering at the next? In observational data, that is the strongest move available.
"Strongly suggested" is the authors' own hedge, and it isn't decoration. A time-lag analysis in observational data does not rule out a third cause that moves first.
"Generally" admits the other direction exists in the data.
So what can you say? That in this sample, the order of events fits wandering leading unhappiness better than the reverse. What you can't say is that mind-wandering causes unhappiness, full stop, which is how the finding is almost always reported. The gap between those two sentences is the whole lesson.
And one more figure, because it's the study's most quoted comparison. What people were thinking predicted their happiness better than what they were doing: activities explained 4.6 percent of the within-person variance in happiness, mind-wandering 10.8 percent.1 That is a comparison of explained variance, not a claim that what you do doesn't matter, and the difference between those two readings is the same kind of gap.
Somebody says this is a study of two thousand people who downloaded a happiness app, so it tells you nothing about anybody else. Are they right?
Show the answer
They have identified the right problem and drawn the wrong conclusion from it.
The problem is real and it's self-selection, not the sample size. Two thousand two hundred and fifty people is a large sample for this kind of work. What shapes it is who installs a happiness-tracking application: people who own an iPhone in 2010, who are interested enough in their own mood to sign up, and who keep answering. That's a specific group and the study says so, which is why the figures are reported with the sample attached.
What it rules out is the sentence "people spend 46.9 percent of their time thinking about something else". That sentence is everywhere and it isn't what the study supports.
What survives is more interesting than what is lost. The floor across activities, at least 30 percent everywhere, is a finding about a large and varied group doing ordinary things, and it would be remarkable in almost any sample. The mood association held during every activity measured. And the instrument itself, which is the thing worth taking away: one question, at moments the person did not choose, is a better measurement of this than anything you could ask at the end of a day.
So the honest form is narrower and still worth having: in a large self-selected sample of adults in 2010, minds were elsewhere in about half the moments sampled, and in at least three in ten during every activity but one. Whether your day looks like that isn't something this study was built to tell you, and the exercise below is how you find out.
Worked: one hour, sampled twice
This case is constructed.3 It's built to show what the instrument does and isn't taken from the paper.
Nine in the morning. Marcus is writing a report. The prompt arrives. What was he doing? Writing. Was he thinking about something other than what he was doing? He was, in fact, thinking about a conversation from yesterday. He answers yes, something unpleasant.
Half past nine. He is still writing the report, further down the page. The prompt arrives again. This time he was reading a sentence he had just written, and his answer is no.
Now notice three things about those two data points.
- Nothing external distinguished them. A colleague watching Marcus would have seen a man typing in both cases. There's no tally anybody but Marcus could keep, which is the difference between this lesson and lesson 2.
- The first answer was available for about two seconds. Ask him at lunchtime how much of the morning his mind wandered and you'll get an estimate built from memory, which is a different instrument giving a different number. This is lesson 1's point and it's why the study prompts at random.
- Both moments were "writing the report". If you sorted Marcus's morning by activity, the two are the same row. The study's finding is that this sorting is the less informative one.
What the case cannot show you is what Marcus's proportion was, because he's made up.3 The proportion is the one thing in this lesson you can only get by doing it.
Three things people get wrong about this
"Half your life is wasted." The study counted moments when attention was elsewhere, not moments that were wasted. Planning, remembering and anticipating are all mind-wandering by this definition, and the paper calls the ability "a remarkable evolutionary achievement" in its own opening.
"Mind-wandering causes unhappiness." Stated flatly, that is stronger than the paper, which says time-lag analyses "strongly suggested" it was "generally" the cause.
"Your mind wanders because you are undisciplined." At least 30 percent of samples during every activity the study measured except one, in 2,250 adults. Whatever explains that, individual failure is not the obvious candidate.
Practice
Take one day, and about 20 minutes of writing at the end of it.
Set something to prompt you at random, not on the hour. A random-interval timer, or a friend, or an app. Eight to twelve prompts across your waking hours is enough.
At each prompt, answer two questions in five seconds, before you've thought about it.
- What are you doing? Three words.
- Are you thinking about something other than what you're currently doing? No; yes, pleasant; yes, neutral; yes, unpleasant. Use the study's four options rather than your own.
At the end of the day, work out your proportion: how many of the prompts got a yes.
Then write three lines.
- Your number, and how many prompts it came from.
- Whether your yeses clustered in particular activities, and whether that surprised you.
- Why your number and the study's 46.9 percent are not comparable. At least three reasons, and this is the line the exercise exists for.
Take 20 minutes. This one is about reading rather than about you.
Find a claim in the wild of the form "X causes Y", about attention, mood, sleep, phones, or anything else you have met this week. Write it down word for word, with where you found it.
Then answer four questions.
- What was measured, and was it measured at one moment or over time?
- Could Y cause X? Write the most plausible version of that story you can.
- Could something else cause both? Again, the most plausible version.
- What would rule those out? An experiment, a time-lag analysis, something else, or nothing available.
If the answer to four is "an experiment nobody has run", that's a result, and it is the commonest one. The point isn't that causal claims are always wrong. It is that most of them are carrying more weight than their design can bear.
Connections
Back. Lesson 1's instrument question is what makes the 46.9 percent readable: it's a self-report, at a moment the person didn't choose, which is a different instrument from both the questionnaire and the task. Lesson 2 measured somebody else taking your attention, and this lesson is the half with no external cause. Logic and Argument lesson 5 is where the readings of a correlation were taught, and the causal exercise here is that lesson applied to one sentence. Habits and Self-Discipline lesson 2 is the nearest relative in structure: a diary study, a criterion, and a figure that travels without its scope.
Forward. Lessons 4 and 5 are the two device claims, and both are about pooled evidence rather than single studies. Lesson 6 asks whether attention can be trained, which is the natural next question once you know how often it is elsewhere. Lesson 7 comes back to causal language.
Go deeper
- A Wandering Mind Is an Unhappy Mind (Science, 2010). One page, free, read in full by this course. Read the first paragraph as carefully as the results: it states the case for mind-wandering as an achievement before the paper measures its cost, which is the kind of balance a finding this quotable rarely keeps.
- The Cost of Interrupted Work: More Speed and Stress (CHI 2008), read in full by this course and the subject of lesson 2. Worth reading beside this one: the two papers measure the two halves of the same problem, one where somebody else takes your attention and one where nobody does, and the instruments they reach for are completely different.
Sources
- Matthew A. Killingsworth and Daniel T. Gilbert, "A Wandering Mind Is an Unhappy Mind", Science 330, 2010, page 932. Read in full; it is a one-page Brevia. Supports: the quoted sampling question and its four options; the quoted sample description; the quoted 46.9 percent finding with its floor across activities; the quoted caveat about laboratory rates; the quoted mood slope; the quoted time-lag sentence; the variance figures of 4.6 and 10.8 percent; and the description of mind-wandering as an evolutionary achievement, which is in the paper's opening paragraph.
- The distinction between an interruption and mind-wandering, and the claim that the second leaves no external trace, is this course's own framing rather than a claim from source 1, and the body states it as the lesson's own in the opening section. What it rests on is a property of the two phenomena rather than a study.
- Marcus is constructed, and so is every detail attached to him; the lesson says so where he appears and again where it says his proportion is unavailable. The three observations drawn from his two data points are this course's own reading of what the instrument does, marked inline, and none of them is a finding from the paper.
Check your understanding
This lesson has a 6-question quiz. Pass it and the questions come back on a schedule in Review, so what you learned stays learned. Your progress is saved in your browser; no account needed.