Putting a number on the silent mind

For a long time the assumption — in psychology, in education, in everyday life — has been that thinking happens in language. Self-help books invite you to "talk to yourself"; therapists ask what your inner critic is saying; teachers tell children to "use their inner voice" while reading. The implicit picture is that thought is verbal, and that everyone shares this picture by default.

Recent self-report work has chipped at that assumption: the experience of inner speech in adults varies from near-constant to entirely absent. What was missing was a careful behavioural study that asked whether the variation actually matters for what the mind can do. Nedergaard and Lupyan’s 2024 paper in Psychological Science is that study — and it gives the absence of inner speech a name worth using: anendophasia.

What the paper did

Four experiments comparing adults who reported low levels of inner speech (n = 46) with adults who reported high levels (n = 47), recruited from the wider population rather than from clinical samples. Each experiment targeted a different cognitive task that prior work had connected to verbal cueing or inner-speech rehearsal: verbal working memory, rhyme judgement, task-switching, and categorical effects on perceptual judgements.

What the paper found

• Verbal working memory was lower in the low-inner-speech group. The kind of memory load that benefits from rehearsing items "in the head" performed worse for adults who do not, by their own report, do that rehearsal in language.
• Rhyme judgement was harder for the low-inner-speech group. Asked whether two written words rhyme — a task that typically activates phonological inner speech — adults without that inner soundtrack found the judgement slower and less accurate.
• Task-switching performance was unaffected. Switching between rule sets in a structured task did not differ between the two groups, despite prior theory linking task-switching to "endogenous verbal cueing" — the idea that we narrate the rule we are about to apply. Either that narration is doing less work than thought, or low-inner-speech adults are managing the switch some other way that delivers the same outcome.
• Categorical perceptual judgements were unaffected. Judging colours and other graded categories did not differ between the groups, again against earlier theoretical expectations.

The pattern is selective rather than global. Where a task explicitly leans on phonological inner speech (rhyme, rehearsal-based working memory), the absence of that speech costs something. Where the task could in principle be solved verbally but does not require it (switching, perceptual categorisation), the absence costs nothing measurable.

Why it matters

The paper is useful for three reasons. First, it gives the field a name — anendophasia — that distinguishes the lack of inner speech from related phenomena (no auditory imagery, no internal monologue, no thought at all). Second, it shows that the variation is real enough to register in standardised lab tasks, which means the self-report numbers were not an artefact of how questions about inner experience were being asked. Third, it shows the cost is bounded: low-inner-speech adults are not failing on cognition broadly, only on specifically phonological tasks. That is the right shape of finding to motivate further careful work without overclaiming.

For readers of this site, the paper is the first peer-reviewed reference point underneath the anendophasia glossary entry and the cluster paragraph in the cornerstone guide. If you have always experienced thought as wordless, this is the paper that says — quietly, with numbers — that the experience is real, has a name, and has measurable consequences in some places and none in others.