← Blog

Focus, explained

Does music help you focus?

This is the most confidently argued question in the whole subject and one of the least settled. Half the internet insists that lo-fi beats unlocked their concentration; the other half insists that silence is the only serious option. The research is not on either side. It says the average effect of background music on task performance is close to zero — and that "close to zero" is what a real benefit and a real cost look like when they cancel each other out.

By ComunAI · 24 August 2026 · ~6 min read

Start with the famous result, because it is not what you were told

In 1993 Frances Rauscher, Gordon Shaw and Katherine Ky reported that college students who listened to ten minutes of Mozart before a spatial-reasoning test scored higher than after silence or a relaxation tape.¹ The effect lasted about ten to fifteen minutes. It was one short paragraph in Nature, on one narrow task, in adults.

What happened next is the useful part. Replications failed — Kenneth Steele and colleagues published a direct attempt under the title "The mystery of the Mozart effect: failure to replicate."² Evidence: Strong A later meta-analysis by Jakob Pietschnig and colleagues pooled the accumulated studies and concluded there was little support for a specific effect of Mozart's music on spatial ability.³ Evidence: Strong

And then the mechanism was found, which is the bit almost nobody repeats. William Forde Thompson, Glenn Schellenberg and Gabriela Husain showed that the advantage tracked arousal and mood, not Mozart: a piece that made listeners more alert and more cheerful produced the bump, and an equivalently pleasant non-musical stimulus could do the same job.⁴ Evidence: Moderate

The cost side: it is the words, not the volume

The clearest negative evidence comes from a different literature entirely — short-term memory research. Pierre Salamé and Alan Baddeley found that background music impaired immediate serial recall, and that vocal music impaired it more than instrumental music.⁵ Evidence: Strong This is the same phenomenon we went through in the open-plan office post: the disruptive ingredient in background sound is not loudness but structured, intelligible speech, which gets into the verbal rehearsal loop whether you attend to it or not.

Nick Perham and Joanne Vizard then closed the door most people leave open. Participants did worse on serial recall with background music than in quiet — and it made no difference whether the music was music they liked or music they disliked.⁶ Evidence: Moderate Preference did not rescue performance. Liking a track changes how you feel about the interference, not whether it interferes.

Which is worth sitting with, because "but I like this playlist" is the single most common counter-argument, and it is the one the experiment was designed to test.

What happens when you pool everything

Juliane Kämpfe, Peter Sedlmeier and Frank Renkewitz meta-analysed the background-music literature across adult listeners and found essentially no overall effect on performance.⁷ Evidence: Strong That headline is easy to misread. A near-zero average across many studies is not the same as "music does nothing." It is what you get when music reliably helps some tasks, reliably hurts others, and the literature samples both.

The moderators are where the practical content lives, and two of them are reasonably consistent:

• Task type. Verbal tasks — reading, writing, anything that runs through inner speech — are the ones lyrics damage. Repetitive, low-demand, monotonous work is where background music most often helps, plausibly by holding arousal up.
• The person. Gianna Cassidy and Raymond MacDonald found that introverts were more impaired than extraverts by high-arousal music with lyrics.⁸ Evidence: Moderate Manuel Gonzalez and John Aiello later reported a similar interaction pattern, with the effect of music depending jointly on task complexity and the listener's preference for external stimulation.⁹ Evidence: Moderate

Both of those are personality-by-task interactions from modest samples, so grade them accordingly. But the direction is stable enough to be worth a personal experiment, and the shape of the finding explains why this argument never resolves online: two people can both be reporting their own results accurately and still disagree completely.

Where the evidence stands

The calibrated bottom line

Music is not a focus tool. It is a mood and arousal tool that sometimes has focus consequences, and the sign of those consequences depends on what you are doing.

A shape that follows from the evidence rather than from anyone's playlist: if the work runs through language — writing, reading closely, reasoning in sentences — the lyrics are competing for the machinery you are trying to use, and dropping them is the single change with the most support behind it. If the work is repetitive and the risk is drifting off rather than being pulled off, music is a reasonable bet. And if you are using music mainly to mask an environment you cannot control, you are solving a real problem, but you are solving it with a second sound source rather than with quiet.

The uncomfortable implication of the Perham and Vizard result is that your sense of how well a playlist is working is not a measurement of how well it is working. The only way to find out is the boring one: same kind of task, two weeks, one week with and one without, and judge it on what you finished rather than on how the sessions felt.

What none of this touches is the structure of the session itself. Audio does not decide how many things you are holding at once, or how often you break your own concentration to check something — and that is the larger of the two variables by some distance.