Can Exceptional Human States Be Measured?

Imagine a basketball player standing at the free-throw line in a silent arena, and something happens that everyone in the building recognizes and nobody can measure: the shot goes in, repeatedly, as if the body had been replaced with a more reliable version of itself. The player will later describe the moment in the standard vocabulary — time slowing, the crowd receding, the motion happening without deliberation. Coaches call it being locked in. Researchers call it flow. Marketers call it peak state and sell devices that promise to detect it.

The question at the center of this article is whether that episode can be captured as a fingerprint: a signature of physiological and neural activity that identifies the state, distinguishes it from ordinary functioning, and travels with the person from one performance to the next. The answer, after fifty years of research, is instructive rather than satisfying. Some components of the signature are real and reproducible. The composite fingerprint — the thing wearable vendors and coaching apps imply already exists — has not been established, and the studies that come closest to establishing it disagree with one another in ways that matter.

What a state fingerprint would have to be

The phrase “measurable state” hides a specification. Before asking whether a peak performance episode can be measured, it helps to state what a measurement would have to do.

A usable fingerprint needs at least four properties. Specificity: the signature should appear during the exceptional state and not during ordinary alertness, boredom, or anxiety. Sensitivity: it should appear reliably across the episodes a person would call exceptional, not only in the best twenty minutes of a laboratory session. Stability: it should be reproducible in the same person across days and in different people across laboratories, because a marker that only works in the lab that discovered it is a curiosity. Causal relevance: the signature should participate in the state rather than merely accompany it, because otherwise the marker is an epiphenomenon and manipulating it will not produce the state.

Most published work clears one or two of these bars and stops. That is normal science; the difficulty is that the promotional layer around the research reports all four as cleared. The gap between “we found a correlation in sixteen participants during a mental arithmetic task” and “we can tell when you are in flow” is where most of the confusion lives.

Flow, the most-studied candidate

The state that researchers have tried hardest to measure is flow, the term Mihaly Csikszentmihalyi introduced in the 1970s to describe full absorption in a task whose demands match one’s skills. Flow is the right place to start because it has an unusually developed psychometric base, and because it is what the motivational and athletic worlds generally mean by “peak state.”

The best-known instrument is the Flow State Scale, developed by Susan Jackson and Herbert Marsh and published in 1996 in the Journal of Sport and Exercise Psychology. The scale operationalizes flow as nine dimensions: challenge–skill balance, action–awareness merging, clear goals, unambiguous feedback, concentration on the task, a sense of control, loss of self-consciousness, time transformation, and autotelic experience. The original validation used 36 items, four per dimension, across a sample of 394 athletes, and reported acceptable internal consistency. It is a real, thoughtfully constructed instrument, and it did the field a genuine service by making flow concrete enough to study.

It also reveals the first structural problem. The scale measures what people say about a state, not the state itself. When a study reports that flow correlated with some physiological signal, what it usually means is that a self-report score on nine dimensions correlated with the signal. If the self-report is noisy, biased, or influenced by the measurement itself, the physiological result inherits that noise. The reference review by van der Linden, Tops, and Bakker, published in the European Journal of Neuroscience in 2021, makes this point explicitly: the neuroscientific literature on flow leans heavily on self-report, and the field’s own proposed neural model — involving dopaminergic and noradrenergic systems interacting with the default mode, central executive, and salience networks — remains a framework to be tested rather than a settled account.

The EEG evidence contradicts itself

If a state fingerprint exists, the most obvious place to look for it is the electroencephalogram, which is cheap, portable, and sensitive to moment-to-moment changes in cortical activity. Two studies on flow and the same EEG feature point in opposite directions, and comparing them is the quickest way to see how fragile the fingerprint currently is.

In 2018, a team led by Takashi Katahira published results in Frontiers in Psychology from sixteen participants performing a mental arithmetic task. Flow experience was associated with increased frontal-midline theta activity and moderate increases in frontocentral alpha power. Frontal-midline theta is an appealing candidate marker because it has been linked in other work to focused attention and working memory load, so the result fit an intelligible story: flow looks like effortful concentration that feels effortless.

That story took a serious hit in 2024, when Naoki Sugino and colleagues at Keio University posted a preprint reporting an EEG investigation of more than 700 video gameplay sessions across seven participants. They found no relationship between frontal-midline theta and flow in six of the seven participants. The study is a preprint, meaning it has not completed peer review, and its participants played video games rather than solving arithmetic, so it is not a direct refutation. It is, however, a strong counterexample to the claim that frontal-midline theta is a general neural signature of flow. A marker that fails to replicate in 700 sessions of a naturalistic task is not yet a fingerprint.

The responsible reading of this pair is not that one study is right and the other wrong. It is that flow is probably not a single neural state. The arithmetic version and the video-game version may recruit different processes, and self-reported flow may be a family of experiences rather than one. If that is true, the search for a single signature is aimed at the wrong target, and the better question becomes whether there are several distinguishable flow-like states, each with its own profile.

What the body does during flow

Peripheral physiology has produced the most consistent — and the most modest — findings. In a 2010 study in Emotion, Örjan de Manzano and colleagues had professional classical pianists perform the same piece five times while recording a battery of autonomic measures. Flow was significantly related to heart period, blood pressure, heart-rate variability, activity of the zygomaticus major muscle in the face, and respiratory depth. The authors interpreted the pattern as consistent with “effortless attention”: a state in which engagement is high but the usual signs of strain are absent.

This is genuinely informative, and it is also a long way from a fingerprint. Autonomic measures shift with physical exertion, posture, temperature, caffeine, and emotion of any kind. The pianists were playing an instrument, so the signal is inseparable from the activity producing it. Nothing in the result says the pattern would identify flow in a person sitting still.

A 2023 study in Scientific Reports by Lu, van der Linden, and Bakker tried to isolate a more specific mechanism by looking at the locus coeruleus–norepinephrine system, using pupil dilation and the P300 component of the event-related potential as indirect indicators while participants performed a gamified working-memory task. Both flow and the noradrenergic indicators followed an inverted-U pattern relative to subjective task difficulty: the state peaked when the task was neither too easy nor too hard. Pupil dilation showed a positive linear relationship with flow, but P300 amplitude did not. The authors read this as partial support for a role of the noradrenergic system in flow, with the caveat that neither pupil size nor P300 is a direct measure of locus coeruleus activity.

The inverted-U result is worth pausing on, because it is one of the few places where the physiology and the theory converge. Challenge–skill balance is the first dimension of the Flow State Scale, and it is also the shape the psychophysiology describes. That convergence is real progress. It is also the kind of finding that makes a state fingerprint hard, because an inverted-U marker is ambiguous by construction: the same pupil diameter can mean “under-challenged,” “in flow,” or “overwhelmed” depending on context. A measurement that returns the same value for three different states is not yet a measurement of any one of them.

Why self-report is both necessary and dangerous

Every route into the exceptional state runs through the person’s own account of it. There is no behavioral test for flow, the way there is for a knee reflex, and there cannot be, because the state is defined partly by how it feels. That makes self-report indispensable — and it makes the whole enterprise vulnerable in a specific way.

The vulnerability is not that people lie. It is that the act of reporting changes what is reported. Asking someone to rate nine dimensions of absorption during a performance interrupts the absorption. Asking afterward invites reconstruction: memory smooths the episode, and the person’s theory of what flow should feel like shapes the recollection. Both the timing and the framing of the question therefore enter the data. Longitudinal designs compound this, because the standard against which new episodes are judged drifts as the person learns the vocabulary.

There is a second issue, mostly statistical. Most flow studies compare people to each other at a single time point. But the claim that matters — “this person is in flow right now” — is a within-person claim, and within-person relationships between a physiological signal and a psychological state are frequently weaker, or even reversed, relative to the between-person relationship. A marker that separates high-flow people from low-flow people may have no ability to tell whether any individual is in flow at this moment. Much of the applied enthusiasm about detecting peak states rests on exactly that conflation.

A much narrower marker that does hold up

If the composite fingerprint is unsettled, at least one narrow marker has survived decades of scrutiny, and it shows what a mature result looks like. The “quiet eye,” studied extensively by Joan Vickers and others, is the final steady fixation on a target immediately before a movement — operationally, a fixation of at least about a hundred milliseconds within a few degrees of visual angle, terminated by the initiation of the movement itself (Vickers, 2016). Across basketball free throws, golf putting, surgical knot-tying, and many other aiming and interceptive tasks, elite performers show longer quiet-eye durations than less skilled ones, and within an individual, quiet-eye duration tends to be longer on successful attempts than on misses.

This satisfies several of the fingerprint criteria in a way flow does not. It is specific to tasks with a clear target. It is stable in the sense that different laboratories using different equipment recover the same ordering of skill groups. It is causally suggestive, because training studies in which performers are taught to hold the fixation longer show improvements in accuracy. It is also honest about its limits: the mechanism behind the effect remains debated, and later work has shown that quiet-eye durations can rise under training and then drift back toward baseline.

The quiet-eye case is a useful contrast because it isolates why flow is so hard to pin down. Quiet eye is narrow, task-bounded, and observable without asking the performer anything. Flow is broad, domain-general, and defined by subjective experience. The narrower the state, the more measurable it becomes — which suggests that the productive research program is not a single fingerprint for peak performance but a library of small, well-specified markers for particular states inside particular tasks.

What would count as success

It is worth stating, in plain terms, what evidence would move a claim like “this device can detect your peak state” from speculation into engineering and then into science.

Speculation, in this context, is a marker reported in one small study, measured through a proxy whose link to the construct is assumed rather than demonstrated, and never tested on an individual. A great deal of commercial “state detection” sits here, whether or not the underlying paper was sound.

Plausible engineering is a marker with replicated group-level effects, a stated task and population, and a clear indication of what it cannot do. The inverted-U relationship between challenge, arousal, and flow is a reasonable example: useful for designing training progressions, not yet capable of identifying a state in a single person in real time.

Demonstrated science would require the marker to classify an individual’s state above chance in a held-out dataset, to do so in more than one laboratory, to survive pre-registered replication in a naturalistic task, and to respond in the predicted direction to an experimental manipulation of the state. Quiet eye meets most of these tests for the specific motor tasks it covers. No physiological marker of flow meets them.

Why the stakes are not only academic

The reason to be careful here is that the market has already decided the answer. Consumer devices routinely claim to measure “stress,” “focus,” “readiness,” and increasingly “flow,” usually from heart-rate variability, skin conductance, or a single EEG channel. Some of these signals are legitimate in a limited sense. Heart-rate variability has accepted measurement standards and established relationships to autonomic function, so a device can honestly report a number derived from it. Whether that number tracks the psychological state in the label is a separate claim, and it is frequently unsupported.

Two consequences follow. The first is that feedback changes behavior: tell someone their state score has dropped and you have intervened, so the device becomes part of the system it claims to observe. The second is subtler. If a coaching system is optimized to raise a proxy — a smoother heart-rate trace, a lower stress index — it will find ways to raise the proxy that need not raise the underlying state, and the person may end up managing the metric rather than the skill. This is the same failure mode that runs through this series wherever measurement and intervention get tangled: the map becomes a target, and the target starts bending the territory.

None of that argues against measuring. It argues for saying which of the four fingerprint properties has actually been demonstrated, for each marker, in each task. The honest state of the field is that we have solid psychometrics for describing subjective flow, suggestive autonomic and pupillary correlates, contested EEG findings, and at least one narrow motor marker with real predictive power. We do not have a state fingerprint in the general sense, and the studies that come closest are the ones most careful about how little they can claim.

Where this leaves us

An exceptional performance episode is real. It is reported consistently across cultures, sports, arts, and professions, and it is not an illusion produced by retrospection, even granting that retrospection distorts it. What is not yet real is the fingerprint: a portable, specific, stable signature that identifies the state as it happens and generalizes beyond the task and the laboratory that produced it.

The more interesting possibility is that this is not a temporary limitation of instrumentation but a fact about the phenomenon. Peak performance may be a family of states rather than a single one — absorptive in arithmetic, embodied in piano playing, target-directed in free-throw shooting — and the search for one signature may be a category error rather than an unfinished project. If that is right, the future of measurement is not one fingerprint but many small ones, each narrow enough to be verified and each honest about the task it belongs to. That is a less thrilling prospect than the wristband that tells you when you are great. It is also the version that could survive being tested.