This module turns from the possible roles of inner speech to a narrower question: what does the evidence actually show about people who report little of it? Earlier course topics distinguished a person’s usual experience from their ability to use words during a particular task. That distinction matters here.
In this lesson, you will examine Nedergaard and Lupyan’s 2024 study, Not Everyone Has an Inner Voice. Its central finding is a selective association: questionnaire scores related to inner speech predicted performance on some verbal tasks, but not on two other tasks the researchers tested. The study does not establish that inner speech caused the differences, or that participants in its low-scoring group never experienced inner speech.
How the study was set up
The researchers recruited people who had previously completed the Internal Representations Questionnaire (IRQ). They selected participants with relatively low or high scores on its verbal factor, which includes statements about experiences such as thinking through problems as a conversation with oneself. After exclusions, there were 46 people in the lower-scoring group and 47 in the higher-scoring group. The selection thresholds were asymmetric: the lower group came from roughly the bottom 16% of verbal scores, while the higher group was drawn from scores above a threshold at roughly the top 40%.
These are groups defined by questionnaire responses, not two independently verified kinds of mind. Although the paper’s figures display the groups separately, its statistical analyses generally use verbal score as a continuous predictor. Neither the recruitment thresholds nor the group sizes tell us how common a complete absence of inner speech is in the population.
Meet the Neuroscientist Who Discovered That Some People Have No Inner Speech
In Giant's Shoulder's interview, “Meet the Neuroscientist Who Discovered That Some People Have No Inner Speech,” study author Johanne Nedergaard explains why the researchers chose their tasks and describes the broad pattern of results.
Watch the verbal tasks for the researchers’ reasoning about word recall and picture-based rhyme judgments. Then watch the other tasks for the task-switching and perceptual findings. Treat this as an explanation of the hypotheses; use the paper below for the precise methods and limits of the findings.
The four tasks tested different predictions. In rhyme judgments, participants saw two pictures and decided whether their names rhymed. In verbal working memory, they saw five written words in sequence and reproduced them. In task switching, they alternated between adding and subtracting three, sometimes with a cue indicating which operation to perform. In visual discrimination, they judged whether animal silhouettes were physically identical; the researchers examined whether category membership affected response speed differently across verbal scores.
Not Everyone Has an Inner Voice: Behavioral ...
Read the study by Nedergaard and Lupyan for its recruitment criteria, four task designs, and analysis plan. These details determine what its results can support.
Start in “The Present Study,” where the authors introduce the four predictions. Continue through “Methods,” including “Participants,” “Method: Rhyme Judgments,” “Method: Verbal Working Memory,” “Method: Task Switching,” “Method: Same/different Judgments,” and “Data Analysis” (around p. 618). Read the design and methods. Pay particular attention to how participants were selected, what each task required, and the distinction between the groups shown in figures and the verbal scores used in analysis.
The design can reveal whether reported verbal-thinking tendencies and task performance vary together. It does not directly observe someone’s moment-to-moment experience while they solve a problem. Nor does it assign people to have more or less inner speech. Those limits will matter when interpreting both the positive findings and the tasks that showed no association.
Where verbal score predicted performance
Participants with higher verbal scores were more accurate on the picture-name rhyme judgments overall. Verbal score did not predict how quickly they responded. Crucially, the researchers did not find reliable evidence that the accuracy association differed by rhyme type. A visually striking difference in one part of a figure should therefore not be treated as proof of a special effect for that type of rhyme.
Higher verbal scores also predicted better recall of the five-word lists. The study included words that sounded alike, words that looked alike in writing, and words that were dissimilar in both respects. Across participants, sound-alike lists were especially difficult, but the researchers did not find the predicted reduction in this phonological-similarity effect among lower scorers. The supported result is a broader association with recall performance, not evidence that the groups used wholly different memory systems.
The paper also reports a suggestive strategy pattern. Participants were asked whether they had said words aloud while doing the rhyme and memory tasks. Among those who reported doing so, the performance gaps associated with verbal score narrowed or disappeared in the reported comparisons. The image below shows the rhyme-judgment pattern.

The pattern is consistent with overt speech helping some people solve these verbal tasks. It does not prove that higher scorers had been using the same strategy silently, or that speaking aloud caused the gap to close: participants chose for themselves whether to speak, and reported it afterward.
Not Everyone Has an Inner Voice: Behavioral ...
Return to the study for the results of all four tasks. Read the prose alongside the figures rather than inferring the conclusions from plotted group averages alone.
In “Results” (approximately pp. 618–621), read from “Rhyme Judgments and Inner Speech” through “Inner Speech and Categorical Effects on Visual Discrimination”: the task results. The intervening subsections are “Strategies for Rhyme Judgments,” “Verbal Working Memory and Inner Speech,” “Strategies for Verbal Working Memory,” and “Inner speech and task switching.” Track accuracy separately from response time, and note both the significant associations and the interactions the authors did not find.
Where verbal score did not predict a difference
On the task-switching test, alternating between operations—particularly without an external cue—was harder than repeatedly doing one operation. But verbal score did not predict accuracy or speed, and there was no reliable evidence that the groups were affected differently by the switching conditions. This is narrower than saying inner speech plays no role in planning or self-cueing: this particular task may permit other effective strategies.
The visual-discrimination task produced a category effect across participants. When two silhouettes differed, people responded more slowly if both belonged to the same category—for example, two different cats—than if one was a cat and the other a dog. The size of that effect did not vary with verbal score. That finding does not settle whether language ever influences visual categorization; it says that the predicted difference was not detected here.
The study thus resists either sweeping conclusion. Lower verbal scores were associated with poorer performance on two tasks involving words or word sounds, not poorer performance across all four tasks. Conversely, a null association on one switching task cannot establish that inner speech never assists behavioral control.
What “anendophasia” does—and does not—name here
The authors propose anendophasia as a term for a lack of inner speech. A useful name, however, is not itself a measurement. The authors explicitly note that many participants in their “low inner speech” group reported some inner speech. The study therefore primarily compares people with different reported degrees of verbal inner experience; it does not isolate a group proven never to experience it.
There are two further boundaries to the inference. First, responses to a general questionnaire may not accurately describe what happens at every moment, including during these tasks. Second, an association can have more than one explanation. A tendency to use verbal strategies might contribute to performance, but other differences related to questionnaire score could also matter. The authors discuss these limitations rather than claiming that the tasks provide a direct view into participants’ minds.
Not Everyone Has an Inner Voice: Behavioral ...
Finish with the authors’ interpretation and their own stated limits. This is where the paper distinguishes its proposed label from what its participant selection established.
In “General Discussion” (around p. 621), read the task interpretation. In “Anendophasia: A Lack of Inner Speech,” read the paragraph introducing the proposed term. Then read the strategy discussion under “What Have We Learned About Anendophasia?” and the limitations under “Limitations” (around p. 622). Notice where the authors’ interpretations go beyond what was directly measured.
A careful one-sentence reading of the evidence is: In this selected sample, higher IRQ verbal scores were associated with more accurate rhyme judgments and better word-list recall, but not with reliably different task-switching or visual-category effects. That wording preserves a meaningful finding without silently replacing “scored lower on a questionnaire” with “had no inner speech.”
Conclusion
Nedergaard and Lupyan provide evidence that self-reported differences related to inner speech have selective behavioral correlates, especially on the verbal tasks they tested. Reported speaking aloud is a plausible clue about strategy, but it was not a randomized intervention. Most importantly, the low-scoring group included people who reported some inner speech, so the study cannot establish either causation or the consequences of its complete absence.
The next lesson will place these results beside a methodological critique and evidence from complementary approaches. The aim is to assess public claims about anendophasia without turning a questionnaire score into a diagnosis or dismissing the behavioral findings it helped predict.