Light flicker and temporal light modulation: the invisible strobe in LEDs, dimmers and screens

Summary

Much modern lighting and most OLED phones at low brightness flicker faster than you can consciously see, your retina registers that modulation anyway, the EU has regulated it in lamps by law since 2021 while explicitly exempting screens — and the one double-blind field study of removing invisible flicker roughly halved office headaches and eyestrain, with the benefit concentrated in people who were headache-prone to begin with.

Why Moderate

This entry sits at Moderate.

Why not the tier above (Strong). The headline health claim — invisible flicker produces headaches and eyestrain in susceptible people — rests principally on one double-blind field study from 1989, conducted in fluorescent lighting, never replicated as a controlled trial in the LED era, with the benefit concentrated in a subgroup. One good study is Moderate by this platform's rules regardless of how well it was run. The screen-PWM arm has no clinical evidence at all, and the risk bands the practical advice leans on are engineering consensus rather than outcome-derived thresholds. An entry whose strongest symptom receipt is singular cannot carry Strong.

Why not the tier below (Emerging). The exposure layer is as hard as evidence gets — binding EU law with numeric limits, an international engineering standard, and direct physical measurement. The biological-registration layer is replicated human electrophysiology and psychophysics. And the one symptom study is a double-blind controlled field experiment, not an observational signal — the design quality that usually separates Moderate from Emerging is present; only the count is missing. The entry is a solid exposure with thin-but-real outcome evidence, not a speculation.

Per-sub-area split. Standards and regulation: Foundational as facts about what the documents say, with the consensus-not-outcome caveat mandatory whenever the IEEE bands are cited. Retinal and cortical registration of invisible flicker: Strong as measurement. Phantom array: Strong as psychophysics. Wilkins symptom evidence: Moderate. Screen-PWM: Emerging for the exposure-to-risk-band mapping, and explicit absence for symptoms and prevalence. Photosensitive-epilepsy risk from visible flicker: established clinical fact, owned by clinical guidance rather than this entry.

Practical takeaway

Who should act: people with frequent headaches, migraine, or persistent eyestrain that tracks specific rooms, specific lamps, or evening phone use — the population in which the only controlled trial found its benefit. Who need not act: people with no symptoms. Nothing in the evidence supports pre-emptive flicker-proofing of an asymptomatic life, and this entry sells no alarm.
• Buy declared-compliant lighting. In the EU, any LED lamp placed on the market since September 2024 must meet SVM ≤ 0.4 by law, so recent reputable products are largely handled; the practical risk is old stock, no-name imports, and other jurisdictions. Anywhere, "flicker-free" claims that cite IEEE 1789 or state SVM/PstLM values are more checkable than bare marketing.
• Stop deep-dimming cheap hardware. A compliant bulb on a cheap wall dimmer at 10% brightness can flicker severely — the dimmer, not the bulb, becomes the exposure. If a dimmed circuit is suspect, either run it brighter, pair dimmer and bulb models the manufacturer certifies together, or swap the lamp for one with continuous (constant-current) dimming.
• The glance test costs nothing. Flick your eyes quickly across a small bright source (a dimmed bulb, an LED tail-light, a string of festoon lights) at night. A trail of discrete dots — the phantom array — means deep modulation at up to several kilohertz. It is a crude detector, not a meter, but it is free and surprisingly sensitive.
• For suspected screen sensitivity: check your phone model's measured PWM behaviour on display-measurement databases (measurements, not the vendor prevalence claims); if the phone offers a DC-dimming or flicker-reduction mode, enable it; failing that, keep OLED brightness high (modulation depth is worst at low brightness) and dim the room instead — which also serves the evening-light guidance that lives in the light-at-night entry; LCD screens with DC-dimmed backlights are an alternative form factor. Then treat it as an experiment: two weeks on, note symptom frequency, two weeks off. No lab required.
• Keep the cost shape in view. Every action above is free or the price of a light bulb. Across the whole light-and-health literature, the interventions with the strongest evidence cost nothing, and spending scales with evidential weakness — flicker mitigation belongs firmly on the cheap, defensible end of that curve.

Evidence detail

Why This Entry Exists

Almost every discussion of "screens and lights causing headaches" collapses into a blue-light argument, and the blue-light argument is largely lost — blue-blocking lenses probably do nothing for eyestrain on the best systematic evidence. Meanwhile a physically distinct property of light sits mostly undiscussed: temporal light modulation, the rapid fluctuation of light output over time. Mains electricity alternates, so a lamp with a cheap driver ripples at twice the mains frequency — 100 Hz in Europe, 120 Hz in North America. Dimming is often done by pulse-width modulation, switching the light fully on and off hundreds or thousands of times a second. None of this is visible as flicker to conscious perception, which fuses somewhere around 60 to 90 Hz for most people. All of it is measurable, some of it is registered by the retina, and a specific slice of it has controlled human symptom evidence.

This exposure crossed a threshold that most wellness topics never reach: it became law. Since September 2021, LED and OLED light sources placed on the EU market must meet numeric flicker limits, tightened further in September 2024. A regulator does not impose compliance costs on an entire industry for a phantom concern. At the same time, the regulation contains a gap that matters for the modern exposure profile: displays are exempt. The screen you stare at for hours — which on most OLED phones dims by strobing at 240 to 480 Hz at essentially full modulation depth — is legally unregulated for flicker, in the EU and everywhere else.

The entry's job is to hold both halves at their honest size. The exposure is real, universal, measured and regulated. The symptom evidence is real but modest: one good double-blind field study, a coherent measurement literature showing the retina follows "invisible" flicker, and a screen-sensitivity story whose exposure side is verified physics but whose symptom side has never been quantified in a controlled study.

What bad advice this protects against, in all directions:
• "If you can't see the flicker, it can't be affecting you." Wrong on direct measurement. Electroretinography shows the human retina responding to flicker up to roughly 145 to 200 Hz — two to three times above conscious fusion — and cortical responses have been recorded to 100 Hz. Invisible is a statement about perception, not about biology. The double-blind field evidence agrees: office workers could not see the difference between 100 Hz-modulated and effectively steady fluorescent light, and their headache rates roughly halved anyway when the modulation was removed.
• "PWM flicker is why you feel bad — 20% of people are PWM-sensitive." The circulating prevalence figure is unverifiable vendor marketing from companies selling flicker-free displays. No controlled clinical study of screen-PWM symptoms exists at all, so nobody knows what fraction of people are affected, and any confident percentage is a tell that you are reading an advert. The honest statement is: the exposure is real and sits inside the risk bands of the relevant engineering standard; the symptom prevalence is unknown.
• "Modern LED lighting is all flicker-free now — regulation solved this." Three gaps. The EU limits bind only products placed on the EU market — there is no equivalent binding limit in most of the world, including the US. Displays are exempt everywhere. And dimming breaks compliant products in practice: a lamp that measures fine at full output can flicker severely when deep-dimmed on a cheap phase-cut wall dimmer, because modulation depth explodes at low duty cycles.
• "Your headaches are from light flicker — fix your bulbs and they'll stop." The overclaim in the sympathetic direction. Headache is one of the most multi-causal symptoms in medicine, the flicker benefit in the one field trial was concentrated in a susceptible minority rather than universal, and letting flicker become the explanation can delay the differential a recurring headache actually needs. Flicker is a cheap, testable candidate among many — the diagnostic route lives in the headache entry, not here.
• "You need a flicker meter and a whole-house lighting overhaul." The mitigation is among the cheapest in the light family: choose products declaring flicker compliance, stop deep-dimming cheap bulbs, and — if you suspect screen sensitivity — test a DC-dimming mode or a high brightness setting for a fortnight, which costs nothing. Nothing in this entry requires buying a device or a service, which is consistent with a pattern that runs across the whole light literature: the well-evidenced interventions are free or near-free, and the money concentrates where the evidence is weakest.

What this entry OWNS: temporal light modulation as a named exposure — the physics (mains ripple, PWM), the engineering standard (IEEE 1789), the EU regulation and its display exemption, the retinal and cortical measurement evidence, the phantom-array effect, the Wilkins symptom trial, the screen-PWM question with its prevalence firewall, and the practical mitigation ladder.

What this entry DEFERS: headache and migraine differential diagnosis and the photophobia spectral layer (headache_migraine_cross_pillar_diagnostic); blue light, circadian timing, and the broader "junk light" firewall (artificial_light_vs_sunlight and the light family hub); dry eye and blink-rate mechanics of screen strain (dry_eye_and_screen_eye_strain); cognitive-symptom differentials (brain_fog_diagnostic_framework); light-at-night disease epidemiology (light_at_night_and_metabolic_health); and light-emitting therapy devices and their dosing (red_light_therapy_devices_evidence_and_dose).

Evidence

Read the tiers, not the thesis. This entry's layers carry very different kinds of certainty and they must not be averaged: the existence and magnitude of the exposure is measurement and law (as solid as facts get), the retinal registration of invisible flicker is replicated electrophysiology, the symptom claim rests principally on one double-blind field study, and the screen-PWM symptom question has no controlled evidence at all. Every claim below is labelled by what kind of evidence it is.

Sub-area A: The exposure is defined, banded, and regulated (Foundational as fact — but the bands are consensus, not outcome trials)

1. [Foundational — engineering-consensus standard, not an outcome trial] IEEE Standard 1789-2015 defines health-risk bands for LED flicker as a function of frequency and modulation depth: at any given frequency, deeper modulation is higher risk, and risk falls as frequency rises, with modulation above roughly 3 kHz treated as having no observable effect. The standard's recommended practice is what "flicker-free" marketing usually points at. Critical honesty point: the bands were assembled by an engineering committee from the existing literature on visible flicker, photosensitive epilepsy, stroboscopic effects and the limited invisible-flicker symptom data, with safety margins — they are expert consensus about where caution is warranted, not thresholds derived from controlled outcome trials. IEEE Std 1789-2015. Cui bono: flicker-meter vendors and premium LED-driver manufacturers cite the standard to sell measurement gear and better electronics; parts of the lighting industry, including US Department of Energy-adjacent critiques, have argued the bands are over-strict — an argument that happens to align with cheaper driver designs.

2. [Foundational — regulation as fact] Commission Regulation (EU) 2019/2020 makes flicker limits binding law for LED and OLED light sources placed on the EU market: a short-term flicker metric (PstLM) of at most 1.0 and a stroboscopic-effect metric (SVM) of at most 0.9 from September 2021, tightened to SVM at most 0.4 from September 2024. Two facts about its scope carry most of this entry's practical weight. First, this is the EU — most other jurisdictions, including the US, have no equivalent binding limit, so "it's regulated" is a European statement. Second, displays are exempt from the regulation entirely: phone, monitor and television flicker — including OLED pulse-width-modulation dimming — is legally unregulated everywhere. Commission Regulation (EU) 2019/2020, EUR-Lex CELEX 32019R2020. Cui bono: the regulation imposed real compliance costs on lamp makers, which is a point in its favour as a signal; the display exemption benefits every major phone and screen manufacturer, who face no flicker requirement at all.

Sub-area B: Invisible flicker is not biologically silent (Strong — replicated measurement)

3. [Strong — human electrophysiology measurement] The human retina responds to light modulation far above the frequency at which conscious perception fuses it into steady light. Electroretinogram recordings demonstrate retinal responses to flicker up to roughly 145 Hz, with the literature's upper bound for measurable retinal following in the 145 to 200 Hz range, and cortical (EEG) responses have been recorded to modulation at 100 Hz. Conscious flicker fusion for most people under most conditions sits around 60 to 90 Hz. The gap between those numbers is the entry's central physiological fact: a 100 or 120 Hz-rippling lamp looks steady and is nonetheless delivering a modulated signal into the visual system. This is measurement, not a symptom claim — it establishes that a pathway exists, not that harm travels down it. Berman et al., Optometry and Vision Science 1991, PMID 1923343; Herrmann 2001 for cortical following. Cui bono: minimal on the measurement itself — laboratory electrophysiology from academic and national-laboratory lighting research, selling nothing; it is quoted eagerly by flicker-concern marketers, which is a use the data permits but does not require.

4. [Strong — replicated psychophysics] The phantom array effect gives high-frequency flicker a route back into visibility: during a rapid eye movement (a saccade) across a small, bright, flickering source, the flicker is smeared across the retina as a trail of discrete dots. Mean detection thresholds sit around 2 kHz, and some individuals detect phantom arrays above 11 kHz — two orders of magnitude beyond conscious flicker fusion. This is why PWM-dimmed LED car tail-lights break into a string of lights when you glance across them at night, and why a "flickers too fast to ever see" claim needs a frequency far higher than intuition suggests. Roberts & Wilkins, Lighting Research and Technology 2013;45:124; Brown E et al. 2020 for the individual upper range. Cui bono: premium-driver makers benefit from a high bar ("you need multi-kilohertz or DC to be truly clean"); volume lamp makers benefit from the opposite reading. The psychophysics itself is academic and sells nothing.

Sub-area C: The symptom evidence (Moderate — one double-blind field study, done well, done once)

5. [Moderate — double-blind human field study] The load-bearing symptom receipt: in a double-blind field study in offices, replacing conventional fluorescent lighting (which modulates at 100 Hz on European mains, invisibly) with high-frequency ballast lighting (effectively steady to the visual system) roughly halved the incidence of headaches and eyestrain. Workers could not consciously distinguish the conditions — that is what makes the design informative. The benefit was concentrated in people with pre-existing high headache frequency rather than spread evenly, which is the shape you would expect if flicker acts as a load on already-susceptible visual systems rather than as a universal toxin. Honest limits, which must travel with the finding: this is one study, from 1989, in fluorescent lighting rather than LED, and it has not been repeated as a controlled trial in the LED era. It is the best evidence in the field and it is also nearly all of it. Wilkins et al., Lighting Research and Technology 1989;21:11, doi:10.1177/096032718902100102. Cui bono: conducted in the pre-LED era by academic vision researchers with no flicker-product market to serve — one of the cleaner funding pictures in the light literature. Its modern beneficiaries (driver makers, meter vendors) did not exist when it was run.

Sub-area D: Screen PWM (Emerging — the exposure is verified physics; the symptoms are unquantified)

6. [Emerging — physics-measurement for the exposure; absence of clinical evidence for the symptoms] Most OLED phones dim by pulse-width modulation, and display-measurement testing consistently places common flagship devices at PWM frequencies of roughly 240 to 480 Hz with modulation depth approaching 100% at low brightness settings — the screen is fully off between pulses. Mapped onto the IEEE 1789 bands, that combination of frequency and depth sits inside the standard's higher-risk region, and the evening phone-in-a-dark-room scenario is precisely the low-brightness condition where the modulation is deepest. That is the verified half. The unverified half must be stated as absence: no controlled clinical study of PWM-related symptoms from screens exists — no trial, no objective prevalence estimate, nothing beyond user reports and forum self-diagnosis. The widely circulated claim that around one in five people is PWM-sensitive traces to marketing by flicker-free display vendors and cannot be verified against any study; this entry firewalls it. A user who suspects screen sensitivity has a real, physically-grounded hypothesis and a free way to test it — not a diagnosis with a known prevalence. Display-measurement grey literature applied to IEEE Std 1789-2015; symptom-evidence absence confirmed by search. Cui bono, unusually sharp both ways: niche flicker-free display companies inflate prevalence to create a market; the major phone manufacturers profit from the regulatory exemption and have no obligation to publish flicker specifications at all. Neither side's number should be trusted, because only one side has a number and it is marketing.

Mechanism

The physical chain is short and fully verified. Alternating-current mains passes through a lamp's driver electronics; a cheap driver lets the supply ripple through as light modulation at twice mains frequency (100 Hz in 50 Hz countries, 120 Hz in 60 Hz countries), while a well-designed driver smooths it to near-steady output. Separately, dimming by pulse-width modulation switches the source hard on and off at a fixed frequency, and the fraction of time spent on sets the perceived brightness — meaning the deeper you dim, the larger the proportional dark interval, and on cheap phase-cut dimmer circuits the waveform degrades further. Incandescent bulbs modulate only mildly because the filament's thermal inertia smooths the ripple; LEDs respond essentially instantaneously, which is why the LED transition made driver quality a health-relevant variable.

The biological chain is verified at its front end and inferred at its back end. Front end, measured: retinal photoreceptors and downstream retinal circuits follow modulation to roughly 145 to 200 Hz by electroretinogram, and cortical following has been recorded at 100 Hz — the visual system receives and transmits a signal that perception has already discarded. During saccades, high-frequency modulation additionally produces the phantom-array percept, a genuinely visible artifact at kilohertz frequencies. Back end, inferred: the proposed symptom pathway is that this continuous, unregistered neural drive acts as a load on visual systems that are already hyperexcitable — which fits the observation that the Wilkins benefit concentrated in the headache-prone, and fits the broader finding that migraine involves cortical hyperexcitability to visual stimulation. But the full causal chain from invisible modulation to headache has been demonstrated only at the level of the field-trial outcome, not traced step by step. The mechanism section of this entry is therefore honest scaffolding: a measured input, a plausible susceptible-population model, and one controlled outcome result connecting them.

Risks And Contraindications

• Photosensitive epilepsy is the one established serious flicker harm, and it is a different exposure. Visible flicker in roughly the 3 to 70 Hz range can trigger seizures in photosensitive individuals — this is long-established, it is part of why the IEEE bands exist, and it concerns flicker you can see (strobes, malfunctioning lights, screen content), not the invisible high-frequency modulation this entry mostly discusses. Anyone with photosensitive epilepsy already has clinical guidance that supersedes anything here.
• No disease claim. There is no evidence that invisible flicker causes disease — no cancer, no retinal damage, no neurological deterioration. The evidenced harms are symptomatic: headache, eyestrain, and (in the visible range, in susceptible people) seizure risk. Any source claiming flicker "damages" you has left the evidence.
• Over-attribution is the main practical risk. Recurring headaches have a long differential, some of it consequential. Flicker is a cheap candidate to test precisely because testing it takes two weeks and no money — but a negative test, or headaches that persist after mitigation, should route to the proper diagnostic pathway (headache_migraine_cross_pillar_diagnostic), not to escalating environmental purification.
• Nocebo is live. Screen-PWM sensitivity is largely learned about through forums and marketing before it is experienced, and symptom expectation demonstrably generates symptoms. This is a second reason the two-weeks-on, two-weeks-off structure matters more here than in most self-experiments — and a reason to hold the unquantified-prevalence line firmly.

Controversy

Nature of the dispute. Two live arguments, usually fused. The narrow one is regulatory calibration: are the IEEE 1789 bands and the EU limits set at the right strictness, given that they derive from consensus and safety margins rather than outcome trials? The broad one is the screen question: does OLED PWM at 240 to 480 Hz produce symptoms in a meaningful fraction of users, and should displays lose their regulatory exemption? The first argument has good-faith engineering positions on both sides; the second is currently conducted almost entirely between marketing departments, because the clinical evidence that would settle it has never been generated.

Position A: flicker is an under-recognised exposure and the standards are, if anything, lenient. Best evidence: the retina demonstrably registers modulation the standards would permit; the one double-blind field trial found a large symptom reduction from removing modulation nobody could see; the phantom array makes multi-kilohertz frequencies perceptually relevant; and the EU regulator was persuaded to legislate. Where it overreaches: it routinely presents the IEEE bands as if they were outcome-derived safety thresholds (they are consensus with margins); it extrapolates the Wilkins result — one fluorescent-era study — to every LED and screen context; and its commercial wing quotes a prevalence figure that does not exist in any study.

Position B: the risk bands are over-strict and the health case for invisible flicker is thin. Best evidence: the symptom literature genuinely is one field trial deep; US Department of Energy-adjacent analyses have argued the IEEE low-risk recommendations exceed what the evidence supports and impose real cost; most people display no detectable response to compliant-range modulation; and decades of universal 100/120 Hz fluorescent and magnetic-ballast exposure did not produce a legible population-level harm signal beyond the susceptible-minority symptoms Wilkins measured. Where it overreaches: it slides from "the evidence is thin" to "the effect is absent," which the double-blind data contradicts for the susceptible subgroup; it leans on population averages exactly where the evidence says effects concentrate in a minority; and its industrial wing has a direct cost interest in cheap drivers that it rarely declares when arguing the science.

The funding/bias dimension — cui bono, both ways. Strictness has customers: flicker-meter vendors, premium-driver and ballast manufacturers, lighting consultants, and flicker-free display startups all monetise concern, and the unverifiable "20% sensitive" figure is their tell. Laxity has bigger customers: driver electronics are a real cost line in volume lamp manufacture, and the display exemption is worth a great deal to phone manufacturers who would otherwise face redesign of their dimming architecture — the quietest lobby in this space is the one that succeeded, because an exemption achieved is invisible. The academic evidence base — Wilkins, the electroretinography, the phantom-array psychophysics — predates or stands apart from both markets, which is exactly why this entry rests its weight there.

Realised Position: We treat the exposure layer as settled fact: temporal light modulation is universal in cheap LED products and PWM-dimmed screens, it is registered by the retina well beyond conscious fusion, and the EU regulates it in lamps while exempting displays. We hold the symptom claim at Moderate: real, demonstrated double-blind, concentrated in headache-prone people, and resting on a single study that the field has never repeated in the LED era. We hold screen-PWM symptoms at Emerging: verified exposure, absent clinical evidence, and a prevalence figure we refuse to repeat because it is marketing. We recommend the cheap mitigations to the symptomatic and nothing to the asymptomatic, and we sell neither meters, nor bulbs, nor phones — which is what allows this entry to state both that the concern is physically real and that its loudest advocates are advertising.

Cross-Pillar Connections

• light_environment_and_health — the family hub; owns the tier map across all light exposures, the cost-inversion headline, and the canonical statement of the split firewall. This entry is the hub's flicker spoke and defers all family-level framing there.
• headache_migraine_cross_pillar_diagnostic — owns headache differential diagnosis, migraine photophobia and its spectral layer. This entry supplies flicker as one cheap, testable environmental candidate — and notes the Wilkins benefit concentrated in the headache-prone — then routes every persistent or diagnostic question there.
• artificial_light_vs_sunlight — the one-stop light entry; owns the blue-light and screen-spectrum claims, the retinal-safety firewall, and the daylight framing. Flicker is the property of artificial light that entry demotes to a pointer here; spectral claims travel the other way.
• dry_eye_and_screen_eye_strain — owns the blink-rate and evaporative mechanism of screen discomfort, which explains most screen eyestrain without any flicker involvement. This entry adds the modulation lane for the subset whose symptoms track dimming and specific sources, and defers the ocular-surface story entirely.
• brain_fog_diagnostic_framework — owns the differential for nonspecific cognitive symptoms. Flicker belongs on its environmental-candidates list at low prior weight; this entry supplies the honest evidence size so it enters that differential no larger than it is.
• light_at_night_and_metabolic_health — owns the light-timing disease evidence. Its practical advice (dim, dark evenings) interacts with this entry's warning that dimming is exactly where PWM modulation deepens — dim the room with quality sources rather than deep-dimming cheap ones.
• red_light_therapy_devices_evidence_and_dose — owns light-emitting therapy devices, including their driver quality and dosing. Panel and mask products are LED devices like any other and inherit the same flicker considerations; device evidence and dose arithmetic live there.

What would change our mind

Industry bias note

Sources (6)

• IEEE Std 1789-2015. IEEE Recommended Practices for Modulating Current in High-Brightness LEDs for Mitigating Health Risks to Viewers. IEEE Power Electronics Society. (Engineering-consensus body; committee membership spans academia and industry, including parties who sell against its recommendations — meter vendors and premium-driver makers benefit from strictness, volume lamp makers from laxity.) Defines flicker health-risk bands by frequency and modulation depth, with modulation above roughly 3 kHz treated as no-observable-effect; consensus with safety margins, not outcome-trial thresholds.↗
• Commission Regulation (EU) 2019/2020 of 1 October 2019, laying down ecodesign requirements for light sources and separate control gears. EUR-Lex CELEX 32019R2020. (Regulator; imposed compliance costs on the lighting industry — the display exemption benefits screen manufacturers.) Binding limits for LED/OLED light sources on the EU market from September 2021: PstLM ≤ 1.0 and SVM ≤ 0.9, tightening to SVM ≤ 0.4 from September 2024; displays exempt.↗
• Wilkins AJ, Nimmo-Smith I, Slater AI, Bedocs L (1989). Fluorescent lighting, headaches and eyestrain. Lighting Research and Technology 21(1):11-18. doi:10.1177/096032718902100102.↗ (Academic vision research, pre-LED era; no flicker-product market existed to fund or benefit from it at the time — among the cleanest funding pictures in the light literature.) Double-blind field study in offices: replacing 100 Hz-modulated fluorescent lighting with high-frequency (effectively steady) lighting roughly halved headache and eyestrain incidence, with the benefit concentrated in participants with high pre-existing headache frequency; participants could not consciously distinguish the conditions.
• Berman SM, Greenhouse DS, Bailey IL, Clear RD, Raasch TW (1991). Human electroretinogram responses to video displays, fluorescent lighting, and other high frequency sources. Optometry and Vision Science 68(8):645-662. pubmed.ncbi.nlm.nih.gov/1923343↗/" target="_blank" rel="noopener">PMID 1923343↗. (Academic/national-laboratory lighting research; sells nothing — subsequently quoted by flicker-concern marketing, a use the data permits but does not require.) Electroretinogram evidence that the human retina responds to light modulation up to roughly 145 Hz — well above conscious flicker fusion; with later work (Herrmann 2001, cortical following to 100 Hz) establishing that invisible modulation is registered by the visual system.
• Roberts JE, Wilkins AJ (2013). Flicker can be perceived during saccades at frequencies in excess of 1 kHz. Lighting Research and Technology 45(1):124-132. (Academic psychophysics; no commercial sponsor identified.) Phantom-array effect: during saccades across a small bright flickering source, modulation becomes visible as a trail of dots at frequencies around 2 kHz on average — with subsequent work (Brown E et al. 2020) finding some individuals detecting phantom arrays above 11 kHz.↗
• Display-measurement grey literature on OLED PWM (multiple independent display-measurement databases, 2020s). (Grey literature; some measurement sites are vendor-adjacent, and the derived "prevalence" claims in circulation are marketing from flicker-free display vendors — the measurements are checkable physics, the prevalence figures are not.) Common flagship OLED phones dim by pulse-width modulation at roughly 240 to 480 Hz with modulation depth approaching 100% at low brightness — inside the higher-risk region of the IEEE 1789 bands. No controlled clinical study of screen-PWM symptoms exists; the "roughly 20% sensitive" figure is unverifiable against any study and is firewalled by this entry.↗

Related entries