Body doubling is working on your own task while another person is present. They are not helping you. They are not reviewing your work. They are simply there — in the room, on a video call, sometimes only as a recording — while you do the thing you have been putting off.
The other person is the body double. Their entire job is to exist nearby while you work.
That is the whole method. Its plainness is the main reason people dismiss it before trying it once.
Body doubling did not come out of a laboratory. It came out of ADHD coaching and community spaces, spread by word of mouth for years, and only reached researchers afterwards.
The first peer-reviewed study on it makes that point directly. Eagle, Baltaxe-Admony and Ringland surveyed 220 neurodivergent people in 2024 and found the practice “remains nameless to many” — people described a strategy they had been using for years, and only recognised it as a named thing once the researchers gave them the term.
That study also found the community’s definition is wider than the popular one. A body double can be:
If you have ever left a “study with me” video playing while you worked, you have body doubled.
This is where most articles on body doubling overstate their case, so here is the honest version.
What exists. One peer-reviewed study specifically on body doubling (Eagle et al., 2024), which is descriptive: it documents how, when and why people use it. Follow-on work began in 2025: an EEG study of 26 people measuring brain activity while working alone versus with a double, and a 12-person VR experiment (a preprint, not yet peer reviewed) in which people worked faster with either a human or an AI double than alone.
What does not exist. A randomised controlled trial. Nobody has published a controlled effect size for body doubling. The 2024 study is a survey; participants report that it helps, and self-report from people who have already adopted a practice is the weakest form of evidence for whether the practice works.
What that means for you. Body doubling is a well-documented practice with a plausible mechanism and no proven effect size. That is not a reason to skip it — it costs nothing to test — but it is a reason to run it as an experiment on yourself for two weeks and judge from your own completion data rather than from someone’s blog post.
Social facilitation — with a detour worth taking. The standard story is that Norman Triplett settled this in 1898: cyclists rode faster when paced than alone, children wound fishing reels faster in pairs. It is called the first experiment in social psychology and it is in every textbook.
The standard story is wrong in an interesting way. When Michael Strube reanalysed Triplett’s original data in 2005, the competition effect was not statistically significant. His conclusion was that evidence “can be coaxed from the data” — roughly the politest available way of saying it was not really there. For a century, the field’s founding result had been oversold.
Then in 2025 a team re-ran it properly, with 445 children. This time the effect was unmistakable: 55.0 seconds working together against 58.4 seconds alone (F(1,444) = 133.73, p < .001). Triplett’s conclusion was right all along; only his evidence had been thin.
So the mechanism is real. The part that matters to you is the asymmetry Zajonc described in 1965: other people raise your arousal, arousal strengthens whatever your dominant response already is, and that helps on well-learned tasks and hurts on unfamiliar ones. Hold onto that — it predicts exactly when body doubling will fail you.
The cost of starting. Most avoided tasks are not hard. The transition into them is hard. A body doubling session has a start time and another human attached to it, which converts an open-ended decision (“at some point I should do my taxes”) into a scheduled one (“9:00, with Ayşe”).
Cheap externalised commitment. You told someone you would be there. That is a very small commitment device, and small is the point — it costs nothing to set up, so you actually use it.
Every body doubling session is a choice on four dials. Getting them wrong is why people try it once and conclude it does not work.
Four formats, in rough order of setup cost:
| Format | Setup | Best for |
|---|---|---|
| Recorded “work with me” video | None | Testing whether presence helps you at all |
| Scheduled call with a friend | One message | Weekly recurring blocks |
| Drop-in co-working room | Sign-up | Irregular schedules, no coordination |
| Same room, same table | Travel | The hardest tasks, the ones you have avoided for months |
Body doubling has four reliable failure modes, and knowing them is worth more than another paragraph on why it is great.
The work is genuinely new. This is Zajonc’s asymmetry, and it is the best-supported thing on this page. If you are thinking your way into an unfamiliar problem, being watched can make it worse. Use the session to start, then end it and keep going alone.
Your double is a friend who wants to talk. The best body doubles are pleasant strangers or friends with strong meeting discipline. A close friend on a video call is a social event with a spreadsheet open.
You are using it to avoid deciding. If you arrive at three sessions in a row without a named task, the problem is not starting. It is that your task list has no next action in it.
You cannot start without it any more. A useful tool has become the only route in. Deliberately run one solo session a week to keep the other route open.
There is also a fifth case that is not a failure of the method: sometimes the task is stalled because it is undefined, blocked on someone else, or something you have decided not to want. No amount of company fixes that.