Grayscale Phone: Does It Actually Reduce Screen Time?

In three short-term studies with different designs and comparisons, daily screen time was lower with grayscale on. There is no pooled estimate, so the separate figures of 20, 38 and 50 minutes should not be averaged. The one study that logged pickups found no change, and sleep and productivity outcomes were null in the studies that measured them.

Turning it on

On iPhone, Apple’s support page gives the path verbatim: “Open the Settings app, then tap Accessibility > Display & Text Size > Color Filters.” Grayscale is one of the presets there. Apple’s page does not state an iOS version; it scopes the feature by hardware, iPhone 5s and later.

On Android there is no single path to print. Under Accessibility, Google documents colour correction and colour inversion, and neither of those is grayscale. The only setting Google documents by the name Grayscale sits inside Digital Wellbeing, under Bedtime mode and its Bedtime routine, which is the Pixel route. Google’s own pages carry two warnings on these steps: some of them vary by device, and some work only on Android 13 and up. On a non-Pixel phone, check your manufacturer’s support pages rather than a generic instruction.

That is the whole of the how-to. The rest of this article is the part the settings guides skip.

The studies, and what each one actually measured

Seven studies are set out below. They ran between one and three weeks, most used students, only two have a genuine randomised control arm, and no meta-analysis of them was located. Every screen-time figure comes from a single study, and they cannot be pooled.

Read the design column before the results column. One preregistered randomised field experiment, one three-arm field experiment, one within-subjects device-logged study, and four weaker designs. That is the entire base.

The screen-time reduction is the most consistent signal in the small retrieved literature

Three studies produced a screen-time number, measured three different ways against three different comparators.

About 38 minutes a day (Holte and Ferraro 2023). A longitudinal quasi-experiment in a single college sample of 161, of whom 73 used grayscale, over 8 to 10 days. The figure is 37.90 minutes and it is a within-group change, not a difference against a control arm. The capture method is not confirmed, so treat objective measurement as probable rather than established.

About 50 minutes a day against control (Zimmermann and Sobolev 2023). A preregistered three-arm randomised field experiment, 112 participants with 40 in the grayscale arm, run over three weeks. The grayscale arm averaged 233.19 minutes a day against 283.20 in the control condition. This is the strongest design retrieved for this article, and it carries the only hard adherence figure in the set. That figure is poor: grayscale participants reported having it switched on only about half the time, M=46.5%, SD=32.79. A standard deviation of nearly 33 points means the arm contains people who used it almost always and people who barely used it.

About 20 minutes a day (Dekker and Baumgartner 2024). A within-subjects study of 84 people, six days of baseline followed by six days of grayscale, measured by device logging rather than self-report. Screen time fell by roughly 20 minutes a day, p=.022, d=0.51. Self-reported adherence here was 89.73%, much higher than Zimmermann and Sobolev’s, and self-reported rather than hard.

Three studies, three designs, one direction. That consistency is the strongest thing this literature has. What it does not give you is a size. No pooled estimate for grayscale was located, and 37.90, 50 and 20 minutes are not three estimates of the same quantity, so averaging them would produce a number with no meaning.

Grayscale changed how long people stayed, not how often they reached

The most useful finding here is the one left out of every recommendation.

Dekker and Baumgartner (2024) is the only study retrieved that logged unlocks on the device. Screen time fell, and unlocks did not change, p=.860. People reached for the phone as often as before and stayed on it less.

That is a specific and limited claim about what grayscale did in that study. If your problem is total minutes, the mechanism has some support. If your problem is the frequency of reaching for the phone, the one study that measured it found no change, and no other study in this set logged unlocks at all. The reaching itself is a cue problem rather than a display problem, which is the argument about defaults rather than about colour.

Sleep and productivity were null in the studies that measured them

Three studies measured a defined sleep or productivity outcome, and all three found nothing.

Dekker and Baumgartner (2024) reported productivity at p=.686 and sleep at p=.529. Zimmermann and Sobolev (2023) found no effect on sleep quality or GPA, and none on wellbeing, stress or happiness either. Hagerty et al. (2026), a cross-over feasibility trial with one week per condition, 61 enrolled and 51 completing, found no consistent effect. Three different designs, one consistent null across them. Hagerty’s measurement was self-reported with no screenshot verification, which is a further reason not to lean on it in either direction.

The one positive signal on sleep in this area comes from Myers, Drees and Cain (2022), and it is a multi-component intervention with pharmacy students built on the salience principle, which also included keeping the phone away from the bed. Nothing in it isolates the display setting.

So there is no basis in the retrieved studies for a sleep claim or a productivity claim. If a page tells you grayscale improves your sleep, ask which study, then check whether that study changed anything else at the same time.

Two other circulating claims do not survive contact with the sources either. A grayscale anxiety benefit is unreplicated and was contradicted within the same research group, whose own authors note it may not be clinically significant (Holte, Giesen and Ferraro 2023). And the idea that grayscale specifically suppresses social media conflicts by app category: an earlier paper from that lab found a reduction, the same lab’s later paper did not, and Dekker and Baumgartner’s image-based against text-based app interaction was not significant at p=.356. Their study was not powered to detect it, and a non-significant trend is not a finding.

In one of these studies, the control group declined too

Ochs and Sauer (2022) ran a three-arm field experiment with 97 participants over one week. One arm moved problematic apps to a different page, one arm used grayscale, and one was a control group. Smartphone use fell during the intervention, and it fell in the control group as well.

Within that study, this shows that observation or self-monitoring can contribute to a measured reduction. It is a finding about Ochs and Sauer, not a discount to subtract automatically from every other result. Whether it applies elsewhere depends on the design: a within-group change against a participant’s own baseline, such as Holte and Ferraro’s 37.90 minutes, cannot separate the two, while Zimmermann and Sobolev’s comparison against a randomised control arm can. That is a reason to prefer the controlled comparison, not a reason to treat the other figures as inflated by a known amount.

Nothing here speaks to what happens after week three. The documented retrieval for this dossier, dated 4 September 2026, did not locate a grayscale study running longer than three weeks. That absence cuts both ways. It is not evidence that the effect persists, and it is not evidence that it decays. The question has not been measured.

So is it worth doing

The honest summary is short.

It costs nothing, it takes under a minute, and it reverses instantly, which puts it in a different category from interventions you have to buy, install or commit to. The evidence that it reduces daily screen time points consistently in one direction across three short-term studies, none of which measured the same quantity in the same way.

The retrieved studies do not support a sleep benefit, a productivity benefit, or a reduction in how often you pick the phone up. In the best-designed study, people kept it on less than half the time. And what happens in month two has not been tested.

Where a single device setting sits in the wider evidence on attention and distraction is set out in the digital distraction and focus guide.

HOS answers does-it-work questions by reading the studies and reporting what they actually measured. The Weekly System has not launched yet. If you want it when it does, join the launch list.