Researchers Told 164 People They Slept Poorly. Their Cognitive Performance Dropped — Even Though Their Actual Sleep Was Fine.

The Morning Number

Fifty million Americans strap on a sleep tracker before bed each night and check the results before their feet hit the floor, performing a small ritual that feels perfectly rational: if the tracker reports 82% sleep efficiency, you press on with your day; if it says 64%, you reach for more coffee and lower your expectations, taking for granted that the number reflects your biology and your biology determines your capacity. The assumption is so intuitive that questioning it feels silly.

In 2014, two psychologists at Colorado College ran an experiment that dismantled it entirely, finding that what determined cognitive performance was not how participants actually slept but what they were told about how they slept.

The Experiment

Christina Draganich and Kristi Erdal recruited 164 undergraduates across two studies, asking each participant to report their previous night's sleep quality before connecting them to equipment that the researchers described as capable of measuring brainwave data reflecting sleep architecture. The equipment was real, but the readings it produced were entirely fabricated.

Participants were randomly assigned to one of two conditions: half were told their data showed 28.7% REM sleep, which the researchers labeled "above average," while the other half were told they had achieved only 16.2% REM, labeled "below average." Neither number had any relationship to how they had actually slept.

Then everyone took cognitive tests.

On the PASAT, a demanding measure of auditory processing speed and working memory, participants told they slept well scored 34.81 while those told they slept poorly scored 22.13, a 58% performance gap driven entirely by a number somebody made up, with a parallel effect appearing on the COWAT verbal fluency test. Self-reported actual sleep quality showed no significant relationship with any test score, meaning that participants' beliefs about their sleep predicted their cognitive performance while the sleep itself did not.

Draganich and Erdal called it "placebo sleep." Your brain does not merely respond to how you slept; it responds to how you believe you slept, and belief turns out to be remarkably easy to manipulate with a single number on a screen.

It Replicated

In 2018, Dara Gavriloff and colleagues at the University of Oxford tested whether the effect held in people with genuine sleep problems by recruiting 63 adults who met DSM-5 diagnostic criteria for insomnia disorder and giving them sham feedback from actigraphy wristbands, devices that look and function exactly like consumer sleep trackers. The results mirrored Colorado College: negative sleep feedback decreased alert cognition with an effect size of d = 0.79 and increased sleepiness and fatigue at d = 0.55, while positive feedback produced improvements that persisted throughout the day. Even in clinical insomnia patients whose sleep genuinely is disrupted, the number on the screen mattered more than the night behind it.

The 17-Million-Person Morning

In 2017, researchers at Rush University Medical College coined a term for what happens when sleep tracking backfires: orthosomnia. Patients were arriving at sleep clinics anxious not about how they felt but about what their Fitbit said, trapped in a cycle where worry about bad data was degrading their sleep, which generated worse data, which deepened the worry. Baron and colleagues documented the phenomenon in the Journal of Clinical Sleep Medicine and noted that roughly 10% of U.S. adults were using a wearable sleep tracker regularly.

That figure has since grown to an estimated 50 million users, and here is a calculation nobody has published: consumer sleep trackers are known to overestimate wake time and underrate sleep quality relative to clinical polysomnography, and if even a third of users see a discouraging score on a given morning, that is roughly 17 million Americans beginning their day with a number that, based on the Draganich and Gavriloff data, may impair their cognitive performance at an effect size comparable to moderate sleep deprivation. They are not tired. They think they are tired, and the thinking does the damage.

The mechanism has been demonstrated twice, in healthy and clinical populations, with large effect sizes, and the delivery system — a screen showing a sleep score each morning — already exists on millions of wrists.

The Strongest Case Against

The British Psychological Society's review of the original study raised a sharp objection: the "above average" group's PASAT score of 34.81 sat near the established adult norm of 36, suggesting that positive sleep feedback did not boost cognition so much as preserve it while negative feedback dragged performance down, making the finding more of a nocebo effect (bad belief hurts performance) than a true placebo effect (good belief enhances it). If the effect runs in only one direction, the intervention is about removing harmful information rather than providing beneficial information, a distinction that carries real clinical weight.

There is a second structural problem that neither study resolved. Neither the Draganich experiment nor the Gavriloff replication included a baseline control condition in which participants received no sleep information at all, and without that comparison group it is impossible to determine whether "above average" feedback genuinely raised performance above natural levels or simply prevented the decline caused by negative expectations, leaving the core question of directionality experimentally open.

What We Didn't Prove

The original sample consisted entirely of undergraduates at a single liberal arts college, and whether the effect generalizes to older adults, shift workers, or populations with chronic health conditions is unknown. The SDMT (Symbol Digit Modalities Test) showed no significant effect of the sleep manipulation, which was unexpected and suggests that placebo sleep may not touch all cognitive domains equally, while the single-session design of both studies leaves open the question of whether repeated exposure to sham feedback would produce sustained effects, gradual habituation, or something else entirely.

Actual sleep in both studies was self-reported rather than measured with polysomnography, so the research cannot fully separate belief effects from unmeasured real sleep quality differences, and the cognitive tests used are standardized behavioral measures rather than direct indices of brain function — what happens on neuroimaging during placebo sleep remains unknown.

What You Can Do

Consider not checking your sleep tracker first thing in the morning, because the score it displays may cause more cognitive impairment than whatever actually happened during the night. If you use a tracker, focus on trends across weeks and months rather than fixating on any single night's number, and when someone asks how you slept, consider answering optimistically: the Draganich data suggests that the framing shapes your day more than the sleep itself did. For clinicians working with patients who present sleep complaints alongside consumer tracker data, exercise particular caution, because telling a patient they have "poor sleep quality" based on a device reading may be producing the very cognitive impairment you are trying to treat.