The coffee shop that sounds like focus but isn't
There is a familiar story we tell ourselves: that a little ambient noise helps us work. The clatter of a café, a playlist with vocals, a podcast murmuring in the background. It feels productive. And yet you read the same paragraph three times, or you finish an hour and can't quite say where it went.
The feeling and the fact are pulling in opposite directions. What the science of attention shows is oddly specific: your brain does not treat all sound the same. Steady, featureless noise is easy to tune out. But sound that changes—especially speech and lyrics—gets processed whether you consent to it or not, and it quietly competes for the very machinery you're using to think. Psychologists call this the irrelevant sound effect, and once you understand it, the café stops sounding like focus.
What your working memory is actually doing
To see why sound matters, start with how you hold a thought in place. When you keep a phone number in mind long enough to dial it, or track the thread of a sentence you're writing, you're using a part of working memory that the psychologist Alan Baddeley called the phonological loop. It's a small, temporary store for verbal information, paired with an inner voice that rehearses that information to keep it from fading.
That inner rehearsal is the quiet engine behind reading, writing, reasoning through a problem in words, and following a train of thought. It's fast, it's largely automatic, and it has a limited capacity. Crucially, it is built to handle sound-based, sequential information—the order of words, the shape of a sentence, the steps of an argument.
Here is the vulnerability. The phonological loop doesn't have a bouncer at the door. Sound that reaches your ears—particularly sound with speech-like structure—gets admitted into the same store you're trying to use for your own thoughts. Your inner voice and the outer voice end up sharing one small room.
The experiment that isolated the effect
In the early 1980s, Baddeley and his colleague Pierre Salamé ran a now-classic set of studies. People were asked to remember short sequences of numbers while irrelevant speech played in the background—speech they were explicitly told to ignore. Their recall got worse. That's not surprising on its own; distraction distracts.
The surprising part came from the controls. The speech impaired memory even when it was in a language the participants didn't understand, so meaning wasn't the culprit. And bursts of steady, unchanging noise did far less damage than speech did. Something about the acoustic pattern of speech, not its content, was reaching in and disrupting the sequence people were trying to hold.
Later work by Dylan Jones and colleagues sharpened this into what's called the changing-state hypothesis. Sound disrupts serial memory in proportion to how much it changes from moment to moment. A single tone repeated over and over—a steady "state"—is almost harmless. A stream of varying sounds, where each moment differs from the last, is disruptive. Speech is the ultimate changing-state signal: every syllable is acoustically different from the one before. Your brain automatically registers that stream of changes, and those changes collide with your own attempt to keep things in order.
The unsettling implication is that this isn't about willpower or concentration. The interference is obligatory. You can't decide to stop processing the acoustic structure of nearby speech any more than you can decide to stop hearing it.
Why music with lyrics is the quiet saboteur
This is where a lot of people are caught off guard, because it touches something we're fond of: the working playlist.
The psychologist Nick Perham has studied how background music affects exactly this kind of serial, verbal task. His findings are inconvenient for anyone who swears by their focus soundtrack. Music tended to impair performance on serial-recall tasks compared to quiet—and, tellingly, it did so whether or not people liked the music. Enjoyment didn't buy back the cost. What mattered was the acoustic profile. Music that's constantly shifting, and especially music with lyrics, behaves like changing-state speech: it feeds words and rapid acoustic variation into the same loop you're using to read and write.
There's even a wrinkle worth naming honestly. Some people report that a familiar, repetitive track fades into the background and helps them settle. That's consistent with the theory rather than against it—a song you know so well it becomes predictable is closer to a steady state than a changing one, and predictable sound demands less of your attention. But the moment lyrics arrive, or the music surprises you, you're back in the danger zone. Which is why instrumental, low-variation sound tends to be the safer bet for verbal work, and why the podcast you "aren't really listening to" is one of the worst companions for writing an email.
Not all work is equally vulnerable
Here's the useful nuance: the irrelevant sound effect hits verbal, sequential work hardest. Reading, writing, editing, coding, arithmetic, holding an argument together—anything that leans on that inner rehearsal is exposed.
Tasks that don't run through the phonological loop are more resilient. Spatial work, rote physical tasks, some kinds of visual design or sketching—these can coexist with background chatter far more comfortably, because they aren't competing for the same store. This is why the same open-plan office can feel fine when you're doing mechanical cleanup and unbearable when you're trying to draft something in words. It's not your mood swinging. It's whether the task you picked up shares a room with the sound around you.
Speech from other people carries an extra sting. Overheard conversation—especially a half of a conversation, like someone on a phone call—is unusually capturing, partly because its unpredictability keeps pulling your attention to fill in the gap. A predictable hum is easy to ignore. An unpredictable voice is not.
What to actually do about it
The fix isn't monastic silence, which most of us can't arrange anyway. It's matching the soundscape to the work.
When you're doing something verbal and demanding, protect the loop. That means removing speech first—no lyrics, no podcasts, no nearby conversation you can parse. If you want sound, reach for the steady and predictable over the varying and surprising: instrumental music you already know, or broadband noise with no structure to track. The goal isn't stimulation; it's giving your attention nothing new to register.
And notice when you're reaching for background noise to avoid something else—the discomfort of a hard blank page, the boredom before the work grabs you. Often the sound isn't helping you focus. It's helping you not start. The most honest move is sometimes to cut the audio entirely for the first ten minutes and let the task itself become the thing that holds your attention.
The deeper point is that focus isn't only a matter of intention. It's a matter of what your environment is quietly loading into the same small store you're trying to think with. You can guard that store, but only if you know it's there.
Where this meets Reclaim
Most of the sound that breaks your concentration doesn't arrive through your headphones—it arrives through your phone, in the form of notifications that are, acoustically and psychologically, the ultimate changing-state interruption. Reclaim is built for exactly that: to hold back the alerts and pulls that keep feeding your attention something new to process, so the room you think in stays quiet enough to actually use. It won't turn off the café, but it will stop the one device in your pocket from talking over your own inner voice.
If you want to give your working memory a fighting chance, you can start by closing the loudest door. Reclaim is at reclaim.lumenlabs.works—a small way to make the quiet you need easier to keep.