For whom the mind wanders, and when, varies across laboratory and daily-life settings

Psychological Science 28(9):1271–1289, doi 10.1177/0956797617706086; author manuscript, PMC5591044 (NIHMS864840). Read for research direction R2, 4 October 2026. Provenance: papers/amadeus/kane2017_mind_wandering_lab_daily.provenance.json.

What was read

• Read in full: the PMC HTML article converted with pandoc: abstract, introduction, methods (lab measures, personality, the experience-sampling questionnaire in Table 1, screening), results, Tables 2 to 8, discussion, conclusion, footnotes, open-practices statement and the reference list.
• Not read: Supplemental Tables S1 to S3 (personality validity checks, personality by context interactions), the open-practices disclosure PDF, the OSF data and scripts, and Figure 1 (image only).

What they did

• Sample. 274 undergraduates (188 women; mean age 18.7; 44% African American, 42% White) from a 3-session lab study of executive control, N = 472.
• Lab measures (latent factor scores):
  • working memory capacity (6 tasks);
  • attention restraint, i.e. overriding a dominant response (5 tasks);
  • attention constraint, i.e. resisting distractors (5 tasks);
  • lab task-unrelated thought (TUT) rate, from 112 thought probes in 5 tasks. Its internal reliability was α .78 to .93, and it correlated .32 to .68 across tasks.
  • Personality from the NEO-FFI-3.
• Daily life. 7 days, 8 random beeps a day between noon and midnight on a handheld device.
  • Each beep first asked: was your mind wandering at the beep? If so: tuning out or zoning out, daydream, worry, things to do, surroundings.
  • Then thought qualities (pleasant, clear, strange, racing, uncontrollable, ...), affect and context, about 35 items.
  • Instructions stressed taking immediate stock of what one was thinking just before the beep.
• Analysis. Multilevel models. Alpha .05 for a few replication and a-priori tests, .005 for everything else. 2.1% of questionnaires were dropped for near-zero variance (careless responding), and 2 participants were removed.

Main results (verified)

• Volume. 38.4 usable questionnaires per person on average (SD 11.6, range 12 to 71). People with higher lab TUT rates answered fewer (r = −0.20), so the missing data are not random.
• Mind wandering in daily life.
  • It occupied 32% of beeps, ranging from 2% to 97% across people.
  • 60% of episodes were "tuned out" (aware) and 40% "zoned out" (unaware).
  • Content was mostly things to do (4.39 on 1 to 7), then daydreams (3.79), surroundings (3.63) and worries (3.20).
• It is a reliable individual difference. The person-level reliability of the single daily mind-wandering item was .77 (Rasch, a lower bound) from about 38 beeps.
• Context drives the moment. Taking all context variables together, people were more on task when trying to concentrate (b = 0.45) and less on task when anxious or bored. Off-task thoughts were rated less pleasant and clear, and more strange, racing and uncontrollable.
• The lab does not predict the life.
  • Lab TUT rate did not significantly predict daily-life mind-wandering rate (b = −0.18, p = .07), with N = 274 against the earlier study's 72. Working memory, restraint and constraint had no main effects on daily-life rates either.
  • What did replicate is a cross-level interaction. People with better executive control were more on task when trying to concentrate and wandered more when not trying (working memory b = 0.17, p = .001). A general executive factor carries this.
• Personality flips between settings.
  • Neuroticism predicted lab TUT rate (r = .18) but not daily-life rate.
  • Openness predicted daily-life rate (p = .013) but not lab rate.
  • For content: neuroticism predicted worry (b = 0.33, p < .001) and openness predicted daydreaming (b = 0.18, p = .014).
  • Executive abilities and traits predicted how controllable and clear people felt their thoughts to be.

Limits

• Undergraduates, one week. Every daily-life variable is a self-report, given just after an interruption.
• "Mind wandering" is a yes/no judgment against the person's own sense of what they were doing. Probe wording and format shape such reports. The study does not test that.
• Effects are modest, and many tests were run (hence the stricter alpha). The cross-level interactions are the robust part.
• No neural or behavioural ground truth: the study shows that lab and life reports dissociate, not which one is right.

What it means for Amadeus (inference)

• Spontaneous thought has a stable person-level signature. Reliability .77 from about 38 beeps; rates from 2% to 97%; content tied to traits (worry with neuroticism, daydreaming with openness). How often and on what a person's mind wanders is a person-specific parameter, measurable by experience sampling alone. A month gives roughly 150 to 250 beeps per person at this rate.
• Lab sessions misrepresent daily life. Lab TUT propensity, executive control and personality relate differently to thought in the lab and in daily life, and lab TUT barely predicted the daily rate. Scanner or lab sessions, including deliberative recall under recording, capture the person in an unusual context. A slot meant to "live" as the person needs daily-life sampling, and lab-derived parameters need checking against it.
• The moment depends on context. Concentration effort, boredom and anxiety predict mind wandering better than traits do. A slot that reproduces a person's spontaneous thought needs the context model (what they were doing, how they felt), not just a trait rate. The experience-sampling questionnaire here is a ready template.
• Missing data carry information. People who wander more answer fewer beeps. In a month of sampling, non-response must be modelled, not ignored.