Most of us have a voice that never stops. It reads these words as you go, it rehearses the sentence you are about to say, it argues with you at two in the morning. Ask people whether they have it and almost everyone says yes, without hesitation, the way they would confirm that they have ten fingers.
Then in 2024 a pair of psychologists published a paper about the people who say no.
Johanne S. K. Nedergaard, at the University of Copenhagen, and Gary Lupyan, at the University of Wisconsin-Madison, put a name to the condition: anendophasia, from the Greek for "no inner voice." Their paper in Psychological Science compared adults who reported low levels of inner speech, forty-six of them, against forty-seven who reported high levels. The word was coined deliberately as a mirror of aphantasia, the now well-known inability to form mental images.
What makes the study worth reading is not that some people lack a narrator. It is what did not happen when they were tested.
The teams ran four kinds of task. One measured verbal working memory. One was a rhyme judgment: participants saw pairs of pictures and had to decide whether the names rhymed, with twenty orthographic rhymes such as sock and clock, twenty non-orthographic rhymes such as drawer and door, and twenty non-rhyming controls, all under a five-second deadline. One was a task-switching exercise in which people added or subtracted three across five blocks, some blocked, some alternating and cued by colour, some cued by symbol, and some uncued, where the participant had to hold the previous operation in mind. The last looked at categorical perception.
The expectation was straightforward. A long line of earlier work, going back to studies by Baddeley and by Emerson and Miyake in the early 2000s, had shown that when you force people to repeat an irrelevant word aloud, they get worse at tasks that seem to need an inner voice. Suppress the voice and the machinery stumbles. So if some people simply do not have the voice, their task-switching should look different.
It did not. Task-switching performance was unrelated to how much inner speech a person reported. So was categorical perception. The prediction failed.
Where the two groups did diverge was narrower and stranger. The low-inner-speech group performed worse on verbal working memory. And they were less accurate at judging rhyme, though no slower. Everyone was best on the pairs that did not rhyme at all, at 95.7 percent correct. Orthographic rhymes, the kind a person can almost see, came in at 87.5 percent. Non-orthographic rhymes, which you cannot solve by eye and must sound out, dropped to 79.5 percent. The group with less inner speech lost more ground on exactly the category that requires you to hear the word.
There was a hint of how they coped. About half of each group said they spoke the words out loud, and doing so reduced the accuracy advantage that came with higher verbal scores. The people who lacked a silent voice appear to have been doing aloud what everyone else did privately.
How common is this? You will see the figure five to ten percent of the population repeated online. Treat it with suspicion. When Nedergaard and Lupyan analysed their own questionnaire sample of 1,037 people, they found no one at all who reported a complete absence of inner speech. Two people sat at the very bottom of the scale. Russell Hurlburt, who has spent decades sampling what people actually experience in the moment, finds inner speaking occupying roughly twenty to twenty-five percent of waking life, ranging from near zero to near total.
The dispute has since become public. In May 2025, Andreas Lind of Lund University published a commentary in the same journal arguing that self-report cannot establish absence and that there is no convincing evidence anyone lacks inner speech entirely. Nedergaard and Lupyan replied that first-person reports are data too, and that the same argument was once made about aphantasia. In January 2026, Hurlburt entered the fray with a distinction that may matter most: there is a difference between being completely anendophasic and being mostly so, and the mostly-anendophasic may well exist.
Aphantasia offers the useful precedent. The term was coined by Adam Zeman and colleagues in Cortex in 2015, after a patient known as MX lost his mental imagery following a heart procedure and then twenty-one people with lifelong imagery loss wrote to Zeman in response to a magazine article. It is named from the Greek phantasia, and its opposite is hyperphantasia.
There is one more thing worth noticing. The inner voice is the only voice with no recording. It cannot be played back, cannot be verified, cannot be handed to anyone else, and until recently it was assumed to be universal. What Nedergaard and Lupyan did was ask a question the rest of us had never thought to ask, and find that the answer was not what anyone expected.
Notice what that means. Everything you have ever thought in that voice has left no trace. It could not be recorded, could not be handed over, could not be checked by anyone, including you. When a person dies, that voice dies entirely, and nothing of it survives.
We take this as normal because we have nothing to compare it to. The anendophasic make the assumption visible: a whole way of being a person can move through the world and leave nothing behind at all.
The difference between the two groups is not that one thinks and the other does not. It is that one of them is, by its nature, unrecordable.
Which is worth sitting with. If what carries a person across a very long stretch of time is only what they left somewhere it could be found, then the inner voice — the most private and most continuous part of being you — is also the one part guaranteed not to make the crossing.
https://3026alive.com