The Science of Flow

On flow: the state cannot be ordered; the conditions can be prepared.

The piece you cannot order

A constructed scene, not a reported event: a conference slide promises flow at five times your productivity. Under the number is a photo of a surfer deep inside a wave. Half the room photographs the slide with their phones.

One true line from my own history sits beside that slide. In the 1990s lab I saw results move with nothing in the training to explain them (see No. 09). That is where this work began for me, and it left me wary of any slide with a multiplier on it.

I have rowed in some capacity for forty years. The best kind of piece does not come on order. It shows up, or it does not, and the harder anyone reaches for it the more politely it declines.

Flow has become one of the most marketed ideas in performance. You can read that people in flow are five times more productive. You can read that the brain in flow releases a cocktail of six chemicals at once. Some of that is science, some is hypothesis, and some is a number that grew in the retelling.

This article sorts them into three bins: well supported, promising, and overstated. I am the reader here too, and fallible on all of it. For the record, I would like the cocktail to be real. It would make Saturday mornings much easier to plan.

What the research holds

Where the word came from. Mihaly Csikszentmihalyi's first flow work, published as Beyond Boredom and Anxiety in 1975, studied chess players, rock climbers, rock dancers and surgeons. It began with interviews, people recalling their own peak experiences, so it carries every weakness of memory. UChicago Magazine

Nine dimensions. The description that came out of that work is usually given as nine parts. Three are conditions: a balance between challenge and skill, clear goals, and unambiguous feedback. Six are the experience itself: action and awareness merge, concentration narrows onto the task, a sense of control, loss of self-consciousness, altered time, and an activity that becomes rewarding in itself (autotelic). Józefowicz et al. 2022 Some parts, time transformation especially, hold together poorly when the questionnaires are tested statistically.

Catching it in the act. To get past memory, Csikszentmihalyi and colleagues built the experience sampling method: a pager went off at random moments, about eight times a day, and people wrote down what they were doing and how they felt. UChicago Magazine Csikszentmihalyi & Larson 1987

The balance, revised. The first model drew flow as a channel running between anxiety (too hard) and boredom (too easy). Sampling data broke it: a beginner doing an easy task is balanced and not in flow. Fausto Massimini and Massimo Carli, of the Milan group, set the bar higher in 1988. Flow is likely when challenge and skill are both above that person's own average, and the group's later model splits the space around it into eight experiences, which the map below lays out. Bassi & Delle Fave 2016 Critics note the quadrant version can label a quarter or more of daily life "flow" from one condition alone. Abuhamdeh 2020

How strong is the balance? Fong, Zaleski and Leach pooled 28 studies (about 9,600 people). The balance-flow link was moderate, about 0.52 in their random-effects model (the paper leaves unclear whether that figure is a correlation or a Fisher z). It was weaker at work and school (0.32) than in leisure (0.73), and weaker when challenge and skill were measured separately (0.30). Clear goals and a sense of control predicted flow at least as well. Fong, Zaleski & Leach 2015 So challenge-skill balance is an influential model, not a guaranteed recipe. It matters, and it is one condition among several, not the master switch.

Attention is the best-supported mechanism. A 2017 review by Harris, Vine and Wilson argued that flow is best explained in attentional terms: higher-order attention networks engaged, and reduced activity in regions tied to thinking about yourself. They also named two tensions honestly. Physiological and self-report measures often disagree, and brain-imaging evidence for the popular hypofrontality account is equivocal. Harris, Vine & Wilson 2017

Hypofrontality is a hypothesis. Arne Dietrich proposed in 2004 that flow requires a temporary down-regulation of the explicit, analytic system, so that a well-practiced skill can run on the implicit one. Dietrich 2004 The paper is theoretical and reports no new data.

The best imaging study is small and specific. Ulrich and colleagues put 27 people in a scanner doing mental arithmetic, with difficulty set to bore them, overload them, or adapt to their skill. In the adaptive (flow) condition, activity rose in some frontal regions and fell in the medial prefrontal cortex and amygdala, though the medial prefrontal effect was only a trend. Ulrich et al. 2014 So the frontal lobe did not go quiet. One part of it worked harder, one part tied to self-reference eased off.

Dopamine is a proposal. Reviews by van der Linden and colleagues propose that dopamine and noradrenaline systems support the motivation and focus of flow. van der Linden, Tops & Bakker 2020 Their own locus coeruleus paper calls it a hypothesis and notes that no published study has explicitly tested the link. van der Linden, Tops & Bakker 2021 The popular "flow cocktail" goes much further. Shortform on Kotler I could find no study that measured that set of chemicals in people during flow.

Flow and performance travel together, direction unknown. A 2021 meta-analysis of 22 studies (2,462 people in sport and gaming) found a medium correlation, r = 0.31, with moderate to high variation between studies. None of the studies could establish direction. Harris et al. 2021 Flow may help you perform, or performing well may produce flow, or both. Engeser and Rheinberg found flow added 3 to 7% of explained variance in exam and course results beyond ability, and was only marginal for a video game. Engeser & Rheinberg 2008

Measuring it is still hard. The Flow State Scale-2 (Jackson and Eklund) uses 36 items across the nine dimensions; its partner, the Dispositional Flow Scale-2, asks how often you typically get there. Józefowicz et al. 2022 A ten-athlete study found the questionnaires could not reliably separate flow from "clutch," a related but more effortful state that athletes described in interviews. Jackman, Crust & Swann 2017; Swann et al. 2017

Every one of these scales is self-report, and flow is a hard thing to report. Ask in the middle of a piece and you break the absorption you wanted to measure. Ask afterward and the job goes to memory, which tidies and flattens. People also read a word like "control" in different ways. Body measures do not settle it, because they often disagree with what people say. So a flow score is a careful account of what someone noticed, not a reading of the state itself. Measuring a feeling does not show that the feeling improved the work.

What the science cannot yet establish. Most of the evidence is correlation, so we do not know whether flow lifts performance or good performance produces flow. The samples are small: 27 people in a scanner, 22 men under stress, ten athletes in interviews. Many of the tasks are lab tasks, and mental arithmetic is not a 2,000-meter race. No study has measured the brain chemistry the popular accounts describe. The effects that hold up are real and modest. That is enough to work with, and not enough to sell.

The map of eight channels

You do not need all eight names. Start at the center, which is your own average, not anyone else's. Flow sits up and to the right, where challenge and skill both run high for you. Its neighbors are near misses: arousal when challenge runs a little ahead of skill, control when skill runs a little ahead. Around the left, challenge outruns skill, and you reach anxiety and worry. Around the bottom, too little is asked, and you reach relaxation, boredom and apathy. The farther a moment sits from the center, the stronger the experience. High challenge and high skill together still do not guarantee flow. Goals, feedback, attention and the person all matter.

Sorting the claims

On the last row: the McKinsey article asked executives how much more productive they felt at their peak, and the most common senior answer was five times. That is a recalled self-estimate, not a measure, and it was about "peak" experience generally. In retellings it became a finding about flow, then "500%," which is arithmetically a different number: five times baseline is 400% more. Thought Economics on Kotler A hint, not a law.

That is the gap between feeling productive and producing better work. Flow can feel like your best hour of the year. Whether the work was better is a separate question, and the work answers it: the split time, the error count, the draft read cold the next morning. A person remembering that they felt five times more productive is not the same as demonstrating five times the output. The distinction is not a technicality. It is the difference between an experience and a measurement.

Not calm, not happy, not winning

The most useful thing the research does is take three words away from flow.

Flow is not calm. The physiology points to moderate activation, not rest, and on the map above relaxation sits in a different wedge. In my own language, calm is the sky, not the weather, and flow is a kind of weather: it moves, it has heat in it.

Flow is not happiness. Csikszentmihalyi wrote that flow may not feel especially pleasurable while it is happening. Abuhamdeh argues that reading is too narrow: you can enjoy something in the moment without stopping to name the feeling. Abuhamdeh 2020

Flow is not high performance. They overlap, at r = 0.31, which means plenty of good performances happen without it and plenty of flow happens in a mediocre result. Athletes describe a second route to excellence, clutch, which they describe as effortful and deliberate, not effortless (UOW). Swann et al. 2017 The days you fought your way to a good piece count too.

The three may overlap. They are not interchangeable. Chase calm and you may turn the heat down too far. Chase happiness and you may avoid the hard thing flow needs.

The wisdom traditions point the same way, without needing a scanner. They promised no state on demand. They taught a practice and let the state come or not. The research and the old texts rhyme here: you prepare the ground, and you hold the outcome loosely.

That is the principle I build on: the state cannot be ordered; the conditions can be prepared. The conditions are the part you control. The experience is the part you notice and record.

I hold that as my reading, not a rule. Your conditions may look different from mine. Mine includes a decent breakfast, which no meta-analysis has yet tested.

Where SportsFlow fits

SportsFlow is where I measure this, and it follows the same sorting. Its Flow Readiness module reads the conditions you can arrange, your attention, and a short report after a session, prompted rather than scheduled. Together they make a Flow Readiness Index.

One rule matters more than the arithmetic. When a part is missing, the index is shown as partial, not filled in. A Setup Score with no Zone report tells you about the conditions and nothing about the experience, and the screen says so. That is the discipline of the table above: report what was measured, at its real size.

The measures are still in development. Over weeks, sessions with and without a condition you set may show you your own recipe, not someone else's average.

Still not enlightened, Orion