When someone’s attention slips away mid-task, the folk explanation arrives fast and unkindly: they weren’t smart enough to hold it. Drifting reads as a deficiency of discipline, focus, or raw mental horsepower. It’s a tidy story, and it maps neatly onto the way people rank minds. The person who stays locked on is sharp. The person whose thoughts wander to dinner plans, an unanswered text, or a half-remembered argument is not.
A preregistered replication and extension published in Psychonomic Bulletin & Review complicates that story in a specific way. Welhaf, Bugg, and Banks tested 269 participants on a task built to detect whether people time their mind-wandering strategically, drifting more during easy stretches and reeling attention back in as demands rise. They replicated that effect. Then they asked which individual differences predicted who did it well. Fluid intelligence didn’t. Trait measures of mind-wandering didn’t. Working memory capacity did.
That dissociation is the interesting part. If mid-task drifting were simply a shortfall of general intelligence, you would expect the standard measure of general intelligence to track it. In this sample, it didn’t track the strategic piece at all.
What “strategic” means when the task is a clock
The paradigm here is a mind-wandering clock task, a design borrowed from Seli and colleagues’ 2018 work. Participants watch a clock hand sweep around a face and respond at a particular position. The task is mostly undemanding, with a predictable spike in demand at the moment the target arrives. Periodically, a thought probe interrupts and asks what the participant was thinking about just then. Were they on task? Were their thoughts elsewhere?
Because the probes are scattered across the task’s cycle, researchers can reconstruct something a single self-report could never capture: not how much a person mind-wandered, but when. Attention is not a fixed quantity dispensed evenly across a task. It has a shape over time. The clock task makes that shape visible.
“Strategic” mind-wandering, in this framing, means the shape is sensible. Drifting clusters in the stretches where the clock hand is nowhere near the target and nothing is required. As the hand approaches the demanding moment, mind-wandering falls off. The mind is not leaking attention at random. It is spending less of it where less is needed and gathering it before it matters.
That behavior was already documented. What the replication adds is the question of who does it, and the answer sits somewhere other than where common assumptions point.
Working memory tracked the timing; fluid intelligence did not
The replication held. Across the 269 participants, the reduction in mind-wandering as task demands rose reproduced the earlier finding, which is worth saying plainly in a field where preregistered replications don’t always land where the original did.
The extension is where the picture shifts. The researchers measured working memory capacity, roughly how much information a person can hold and manage against interference, alongside fluid intelligence, the ability to reason through novel problems without relying on prior knowledge. They also collected trait mind-wandering measures, the dispositional self-reports that ask how often thoughts spontaneously drift or how often people let them.
Only working memory capacity predicted strategic modulation. Participants with higher WMC showed a stronger shift away from mind-wandering in the moments before the clock task’s highest-demand period. Fluid intelligence did not predict who timed their drifting well. Neither did the trait measures. The people who described themselves as frequent mind-wanderers, spontaneously or deliberately, were not distinguishable on this dimension.
There’s a conceptual reason this makes sense once it is considered carefully. Fluid intelligence is largely about solving. Working memory is about holding: maintaining a goal, tracking a position in a sequence, and keeping an upcoming demand in view while thoughts are somewhere else entirely. Timing drifting requires an internal model of the task’s structure that stays live in the background while attention is occupied by something unrelated. That is a maintenance problem, not a reasoning problem.
Note also what the finding is not. This is not a claim that higher WMC means less mind-wandering overall. The novel piece concerns modulation, meaning the when, not the how much.
What a clock task can and can’t show
The caveats are real, and they matter more here than in most single-study writeups.
This is a lab task with an unusually legible structure. The clock’s demand curve is predictable, repeating, and known to the participant. Almost nothing in ordinary life is that cooperative. A meeting doesn’t announce its hard part in advance. A page of reading doesn’t sweep toward a target position at a fixed rate. Whether the same modulation appears when the demand structure is murky, or arrives unannounced, is a separate question this design can’t settle.
The findings also live inside a single sample and a single paradigm. They build on the earlier Seli et al. result, which is what a replication is for. The WMC link and the fluid-intelligence non-link are the new contribution, and new contributions require confirmation. Nothing here supports conclusions about clinical attention conditions or about how any individual should manage their own workday.
What does survive the hedging is a reframe. The intuition that drifting attention signals a weaker mind treats mind-wandering as pure failure, as noise where signal should be. In this sample, the ability to distribute drifting sensibly across a task tracked a capacity for holding goals in mind and did not track the reasoning ability commonly associated with calling someone intelligent. Wandering, at least in this narrow window, looked less like a mind falling short and more like a mind allocating.
