Method at a glance
- Modality: self-report (retrospective, laboratory; multi-item thought-content probe)
- Measures: The content and form of ongoing self-generated thought, decomposed into simultaneous dimensions — temporal orientation, self- and social reference, affective tone, modality (words/images), specificity, deliberateness — rather than treated as a single on-task/off-task binary. Extended by Banellis et al. (2026) with eight interoceptive and somatomotor items to measure Body-wandering.
- Procedure: Participants rate a battery of statements about the thoughts they have just had (or are having), each on a 0–100 visual analogue scale with labelled poles (‘Not at all’ → ‘Completely’). Item presentation order is randomized. Probes may be delivered during a task at quasi-random intervals (online sampling) or, as in Banellis et al., once retrospectively immediately after a fixed period — there, a 14-min resting-state fMRI scan, with participants kept unaware beforehand that they would be asked, and instructions kept deliberately broad (‘Please answer a few questions about how you felt during the scan’). The resulting item matrix is reduced by factor analysis (principal axis, oblimin rotation) into latent thought dimensions, which are then related to brain, physiology or trait measures across individuals.
- Also called: MDES, multidimensional experience sampling, thought probes, multidimensional interoceptive experience sampling
Strengths
- Multidimensional rather than binary. The predecessor question — ‘were you on task?’ — collapses a rich state into one bit. MDES recovers separable axes that dissociate: in Banellis et al. (2026), a somato-affective factor and a mental-time-travel factor loaded on different items, correlated with opposite affect, opposite autonomic arousal, and different symptom profiles, and had almost non-overlapping connectivity maps (r = 0.061).
- Extensible. Adding items adds dimensions. The entire body-wandering result exists because eight probes were written for a battery that had 14. This is the method’s chief virtue and the source of the literature’s blind spots — nobody had asked about the stomach, so nobody had found it.
- Cheap enough to scale. 22 VAS items after a scan is a negligible burden, which is what makes N = 536 individual-differences designs feasible where true momentary sampling would not be.
- Reliable at the individual-differences level. Split-half ICCs of 0.94 (item prevalence), 0.91 (cognitive–affect associations) and 0.88 (body–affect associations) in Banellis et al. — the dispositional signal is reproducible even though single items are noisy.
Limitations
- Retrospective, in the resting-state application. One probe set after 14 minutes measures the relative salience of thought contents in memory, not their moment-to-moment dynamics — and cannot say whether bodily state preceded, followed, or merely accompanied the thought. Online probing avoids this at the cost of interrupting the state being measured.
- Items are not equated for grain. ‘My thoughts were about my body’ and ‘about my bladder’ differ in specificity, which shapes rating distributions and therefore correlations. Banellis et al. name this as a limitation of their own extension.
- Reactivity and demand. Asking about the body may recruit attention to it. Banellis et al. mitigate by not warning participants in advance and by keeping instructions broad, but a body item still teaches the participant that bodies were relevant.
- Context-bound. The MRI environment — motionless, supine, low sensory input, bladder recently emptied — is a specific bodily and cognitive situation, and MDES in it does not obviously generalize to daily life. Contrast EMA, which buys ecological validity at the cost of control.
- Latent factors explain little variance. 9% and 8% for the two factors in Banellis et al. Reliable structure, small share of the whole.
Connections
The Mind-wandering field’s instrument for asking what the mind was doing, not merely whether it was off task. Developed largely in Jonathan Smallwood’s programme (Smallwood et al. 2016, 2021; Ruby et al. 2013; Wang et al. 2018; Martinon et al. 2019), it presents a fixed battery of statements about ongoing thought and treats the resulting profile as a multidimensional description to be factor-analysed rather than a score.
The wiki meets it through Banellis et al. (2026), where it arrives already modified: eight interoceptive and somatomotor probes bolted onto the standard fourteen.
The battery, and what the extension added
The 14 classical items cover temporal orientation (past, future, temporal distance), reference (self, other), affect (positive, negative), form (words, images, vivid, vague/detailed, repetitive), and control (spontaneous–deliberate, focus).
The eight added items are: body, breath, heart, movement, bladder, skin, stomach, plus arousal (sleepy–alert). Each identically phrased — “My thoughts were about my ___” — and identically scaled.
That symmetry is the design’s argument. Because the new items sit in the same battery on the same scale in randomized order, the body probes and the cognitive probes are directly comparable, and questions like “is body content more or less prevalent than temporal content?” become answerable arithmetic (median 36.3 vs 47) rather than a comparison across instruments.
Why it matters to this wiki
MDES is the wiki’s first method that measures spontaneous interoceptive attention. Every other self-report instrument here asks about interoception in a framed way: MAIA asks how you generally relate to bodily sensation, the Property association task asks what you take an emotion to involve, EMA asks what you are feeling right now. MDES asks what the mind drifted to when nothing was asked of it — which is a different quantity, and the only one of the four that can find out how often the body shows up unbidden.
It is also, straightforwardly, a sensibility instrument in taxonomy terms. Nothing is detected; nothing is scored against ground truth. What it measures well is disposition, and the wiki should not let its impressive reliability be read as accuracy.
Relation to EMA
ESM and MDES are cousins that trade in opposite directions. EMA signals people in ordinary life and buys ecological validity at the cost of experimental control and of any concurrent brain measure. MDES probes in the lab — often inside a scanner — and buys a controlled, instrumented, physiologically-monitored state at the cost of that ecological validity, which in Banellis et al.’s case is a peculiar bodily state indeed (supine, motionless, in a tube). Neither is more real; they answer different questions, and the body-wandering result is a lab finding awaiting an EMA replication the authors explicitly call for.