Does GPS Make You Worse at Navigating? The Largest Study Found Nothing

This is one of those claims that everybody has heard and almost nobody has checked. The version in circulation runs: satellite navigation is eroding your sense of direction, shrinking your hippocampus, and the London taxi drivers prove it.

One third of that is roughly true, one third is small and complicated, and one third is a fabrication assembled from two unrelated bodies of research.

What the meta-analysis found

Miola and colleagues (2024) ran the only systematic review and meta-analysis on this question - preregistered, PRISMA-compliant, 23 studies reviewed and 18 meta-analysed, with individual samples ranging from 13 to 636 participants.

Outcome	Effect	95% CI	Significant?
Environmental knowledge	r = -.18	[-.28, -.08]	Yes, p < .001
Self-reported sense of direction	r = -.25	[-.39, -.12]	Yes, p < .001
Wayfinding performance	r = .07	[-.28, .41]	No

Read the third row twice. On whether you can actually get where you are going, the pooled effect is not significantly different from zero, and the confidence interval spans both directions. The studies do not even agree on the sign.

Heterogeneity is severe throughout - Q(11) = 200.44 for sense of direction, Q(6) = 77.55 for wayfinding. The authors note there is no standardised navigation task in the field: “each study used a different task.” Risk of bias was low in only 30% of studies, no study blinded participants or experimenters, and 91% of the experimental studies reported no reliability or validity information for their measures.

And crucially, the authors refuse the one-way causal story themselves. They describe the relationship as bidirectional: people with weaker spatial ability use assisted navigation more, which may then weaken it further.

The study everybody cites

Dahmani and Bohbot (2020), in Scientific Reports, is the source of most GPS coverage. It has two parts.

Cross-sectional: 50 regular drivers. Lifetime GPS use correlated negatively with spatial memory on a virtual maze (r = -0.22), with map-drawing accuracy (r = -0.22), with landmarks noticed (r = -0.26) and with spatial navigation strategy (r = -0.20). All small.

Longitudinal: 13 of those drivers, retested after about 3.2 years. Here the correlations are large - GPS hours since baseline against decline in spatial memory, r = -0.68 - and this is the number that travels.

It rests on thirteen people, and its confidence interval runs from -0.91 to -0.10. Quoting -0.68 without that sample size is the single most common error in coverage of this topic.

The paper contains nulls that are rarely mentioned. GPS use showed no relationship with self-reported sense of direction (r = 0.07). Lifetime driving experience correlated with nothing, which is a useful specificity control. And most maze performance measures were null against every GPS measure. The authors themselves “caution against any strong conclusions as spurious correlations are possible.”

The largest study found nothing

Yavuz and colleagues (2024) tested 822 participants - roughly sixteen times Dahmani and Bohbot’s cross-sectional sample and sixty-three times their longitudinal one - using a validated wayfinding task.

GPS reliance was not significantly associated with wayfinding performance: β = -0.03, p = 0.350, 95% CI [-0.08, 0.03], f² < 0.001.

Weekly video-gaming hours, by contrast, did predict better navigation (β = -0.12, p < 0.001), and controlling for gaming eliminated the male advantage in the sample.

This is not a formal failed replication - different task, different construct - but it is a large, well-powered null on the outcome the meta-analysis also found null. Two independent lines converge on the same place: GPS reliance does not predict whether you can find your way.

Other nulls sit inside the meta-analysis’s own study table: several studies found no effect on environmental knowledge or wayfinding, and one of them is by the author of the 2008 field study that started this literature. On wayfinding specifically, six of nine experimental studies favoured GPS. Satnav makes you better at arriving. That has never been in dispute.

The mechanism is probably just attention

The strongest causal work here is not about long-term damage at all.

Gardony and colleagues (2013) had people navigate virtual environments with a verbal aid, a tonal aid, or no aid. Both aid types produced modest navigation benefits and substantial spatial memory costs - and the impairment was equivalent for verbal and tonal aids, which rules out a specific verbal-interference account and points to plain divided attention.

Their 2015 follow-up made it sharper. Divided attention alone impaired spatial memory. Aids reproduced that impairment when attention was undivided - but aids did not impair memory further when attention was already divided.

In other words: the navigation-aid deficit may be nothing more exotic than not looking at where you are. That is a claim about momentary attention at encoding, not about a degraded brain - and it has an obvious implication.

The effect is modifiable by design

Nobody has run a “stop using GPS and recover” study. But researchers have removed the deficit by changing the interface, which tells you something about its nature.

Gramann, Hoepner and Karrer-Gauss (2017) compared standard turn-by-turn instructions against instructions augmented with landmark information - some irrelevant, some tied to participants’ own interests. Driving performance was equivalent across conditions, but both augmented conditions significantly improved three measures of spatial learning: unaided route navigation afterwards, landmark recognition, and sketch-map quality.

Liu and colleagues (2022) redesigned navigation aids around virtual global landmarks with 27 participants and found 31.8% lower angular errors in pointing, 16.5% faster responses, and 43.0% shorter distances travelled in wayfinding.

A deficit that disappears when you change the wording of the instructions is an attention-allocation effect, not damage.

The taxi drivers are not about GPS

Maguire and colleagues (2000) scanned 16 licensed London taxi drivers against 50 controls and found larger posterior hippocampi, correlating with years spent driving a cab (r = 0.6).

That study contains no navigation aid, no satnav, and no technology condition of any kind. The independent variable is years of experience holding a London taxi licence, ranging from 1.5 to 42 years. It is a study of structural plasticity following intensive traditional route learning - “The Knowledge” - and it is cross-sectional, so it does not even establish that navigating enlarges the hippocampus.

The chain “navigation builds the hippocampus, therefore GPS shrinks it” is an inference nobody has tested. No study has shown that GPS use reduces hippocampal volume. If you have read that it does, you have read two unrelated literatures welded together.

What this does not establish

Nothing causal about long-term use. There is no randomised trial. Two of the 23 studies in the meta-analysis had any follow-up, and the multi-year one has n = 13.

Not a one-way arrow. The meta-analysts state the relationship is bidirectional; people who navigate poorly plausibly lean on GPS more.

Not a robust effect. The largest study ever run found nothing, and heterogeneity across the field is severe.

Nothing about the brain. No neuroimaging study links GPS use to hippocampal structure.

We are also not printing sample sizes for four of the older experimental studies in this area, including the 2008 field study most often cited, because their full texts sit behind publishers we could not reach. Their designs and directions are reported; their numbers are not.

One more caveat worth stating: the foundational experiments used unfamiliar devices with very small screens on participants who had essentially never used satnav. Their external validity to habitual smartphone navigation in 2026 is questionable.

What follows

The defensible version of this story is smaller and more interesting than the popular one.

Using a navigation aid means you attend less to the environment, and you remember less of it. That is well supported and mechanistically plain.

Whether that accumulates into a lasting deficit is unknown. The evidence is small, correlational, bidirectional and contradicted by the largest study in the field.

If you want to keep the map in your head, the fix follows from the mechanism. Look at the route before you set off, notice landmarks rather than instructions, and - the finding with actual experimental support - use whatever prompts your attention outward rather than down. Researchers achieved this by rewriting the instructions. You can achieve it by occasionally putting the phone away for a route you half know.

And the next time you see the taxi-driver study invoked to explain what your phone is doing to your brain, you are watching a science story mutate in real time.

Navigation is one of several places where the popular claim outruns the measurement - see the cognitive offloading guide for the wider picture, digital amnesia for the memory version of the same worry, and second screening for another case where the cost turned out to be attention rather than the device.