I found out at a dinner table that my friend Maya doesn't have a voice in her head.
I'd been complaining that mine wouldn't shut up on the drive over. It replayed a work conversation twice, drafted a text I'll never send, and narrated the traffic like a low-budget nature documentary. Maya looked at me the way you'd look at someone who just casually admitted they can hear the neighbors' thoughts. "Wait," she said. "There are actual words?"
There are actual words. For me, nearly always. For her, apparently never. She thinks in pictures, she says, and in something she can only describe as pressure, and when she needs to speak she has to translate. She'd spent thirty years assuming everyone did that. I'd spent thirty years assuming everyone had the narrator.
We were both wrong, and it turns out this is one of the stranger facts about being a person: not everyone has an inner voice, and almost nobody finds out until adulthood, usually mid-conversation, usually with that exact expression on their face.
the study that gave it a name
In 2024, Johanne Nedergaard of the University of Copenhagen and Gary Lupyan of the University of Wisconsin dash;Madison published a paper in Psychological Science coining a word for the low end of this spectrum: anendophasia, the absence of inner speech. People with it aren't confused or evasive when you ask what their thinking is like. Nedergaard describes them as thinking in images and then having to translate those images into words when they need them. The words just aren't running in the background.
And it's not merely a difference in vibes. Nedergaard and Lupyan compared 46 adults who reported very little inner speech with 47 who reported a lot, and found the low-inner-speech group did worse on a verbal memory task and struggled more with rhyme judgments: the kind where you see a picture of a sock and a picture of a clock and have to decide, silently, whether their names rhyme. If you have a narrator, this is trivial. Your inner voice just says both words and checks. If you don't, there's nothing to check against.
Here's the part I can't stop thinking about, though. When the low-inner-speech participants were allowed to say the words out loud, the performance gap disappeared. Spoken speech did the exact work the silent voice does for everyone else.
Sit with that for a second. Talking wasn't the output of their thinking. It was the thinking.
we're terrible witnesses to our own heads
The deeper lesson here isn't about Maya, or about whichever camp you're in. It's that we are shockingly bad at knowing what our own minds are doing.
Psychologist Russell Hurlburt has spent decades on this. His method is beautifully simple: give people a beeper that goes off at random moments and train them to report exactly what was in their experience right then. Not what they think their thinking is usually like. What was actually there, at the beep. Using this method, he and Christopher Heavey found that inner speaking shows up in only about a quarter of sampled moments on average, with enormous variation between people. Some talk to themselves inwardly almost constantly. Some essentially never do. And plenty of people in both groups are surprised by their own results.
Hurlburt's blunt summary, in a Popular Science piece on the question: "people are ignorant about their own inner experience." The same article offers the best metaphor I've seen for why. Maybe the inner voice is like the fridge light. It seems on all the time because it switches on the instant you open the door to check.
That's the trap of introspection generally. The Stanford Encyclopedia of Philosophy calls inner speech one of the main ways thought becomes available to us at all. The little voice is a big part of how we know what we think. Which means the instrument you'd use to observe your mind is the mind. Of course the readings are suspect.
One honest caveat: the science here is young and contested. Andreas Lind at Lund University published a commentary in 2025 arguing that Nedergaard and Lupyan showed real differences in how often people use inner speech, but didn't prove anyone has literally zero. Maya might have a very quiet narrator rather than none. Nobody knows yet. Including, notably, Maya.
the part that applies to you either way
It would be easy to read all this as a ranking. Don't. The differences in the study were task-specific and modest, and "worse at silently judging whether sock rhymes with clock" is not exactly a life sentence. The interesting finding isn't that some people's heads are quieter. It's what fixed the gap.
Speaking out loud.
Because here's my suspicion, as someone with a maximally chatty narrator: the inner voice isn't as useful as it feels. Mine loops. It rehearses the same argument fourteen times without ever finishing it. It's the fridge light and the fridge hum. What it almost never does is complete a thought and let me look at it from the outside.
Said out loud, a thought has to become an actual sentence. With a beginning and, crucially, an end. You hear where it's vague. You hear yourself say "and I guess what I'm really upset about is" and finish the sentence in a way the silent loop never forced you to. External speech isn't a transcript of thinking that already happened somewhere private. Sometimes it's the first draft of the thinking itself. That's a large part of why we built Joice as a voice journal rather than another blank page: you talk, the thought becomes words you chose out loud, and what stays behind is a transcript you can search. The sentence, finally outside your head, without the loop attached.
Here's the experiment, then, and it works whether your head is a radio station or a silent film. Take the thing you've been turning over for the past week — the decision, the grudge, the conversation you keep rehearsing — and say it out loud, once, alone, in plain words. Start with "okay, so what's actually going on is" and don't stop until the sentence ends.
Notice what happens. The loop needed the silence to keep looping. Out loud, it's just a sentence. And a sentence is something you can finally do something about.