Learning to read may shape how adults reason about what others believe

A study of literate and illiterate women in Turkey found that reading fluency was strongly linked to performance on a test of reasoning about what one person thinks another believes. The findings, published in PLoS One, suggest that literacy and the schooling that typically comes with it may shape this kind of advanced social reasoning in adulthood.

A central concept in social psychology is “Theory of Mind,” which refers to the ability to recognize that other people have their own thoughts, desires, and intentions. One way to measure this is through a false-belief task. This test assesses whether an individual understands that someone else can hold a belief that contradicts reality.

A more advanced version of this test is the second-order false-belief task. Rather than just asking what one person believes, this task requires reasoning about what one person thinks another person believes. This level of social reasoning typically matures between ages 8 and 10, around the time many children are becoming fluent readers.

Past research suggests that language acquisition and impulse control both help people reason about others’ perspectives. For example, a 2009 study of deaf adults using a newly emerging sign language in Nicaragua found that those whose signing improved over two years also improved on false-belief tests, which the authors took as evidence that language drives this ability. Similarly, a 2001 study of preschoolers found that children better at holding back automatic responses tended to do better at tracking others’ beliefs, though evidence in adults has been mixed.

Other research has shown that reading stories relates to social understanding. For instance, a 2018 meta-analysis found that reading fiction produced a small improvement in social understanding compared with reading nonfiction or nothing.

Free daily newsletter

Tan Arda Gedik of Bilkent University and Friedrich Alexander Universität Erlangen-Nürnberg and his colleagues set out to test whether differences in false-belief reasoning between literate and illiterate adults could be explained by reading fluency or related factors like intelligence and impulse control.

“Most theory of mind (ToM) studies have focused on children or highly educated neurotypical adults, leaving a gap regarding the roughly 700 million illiterate neurotypical adults worldwide, at least according to UNESCO,” Gedik, a cognitive scientist and instructor in the Department of Psychology at Bilkent University and the Department of Foreign Language Teaching at Middle East Technical University, told PsyPost. “Since literacy acquisition significantly alters brain structures that also support ToM, we wanted to investigate this overlooked population and see what differences (if any at all) we might observe.”

For their experiment, the researchers recruited 76 female native Turkish speakers. The sample included 41 illiterate women with an average age of about 49, and 35 literate women with an average age of about 46. The illiterate participants had recently started attending an adult education center and had received only a few months of formal schooling on average. The literate group, by contrast, had an average of over 17 years of formal education.

“I would like the readers to take a moment to appreciate the great lengths these women went to emancipate themselves from some awful sociocultural practices they were subjected to by attending literacy classes (sometimes in secret from their family members, sometimes with bruises on their arms), learning one letter at a time,” Gedik said. “We hope that each and every one of our participants will make their wishes come true (i.e., finding a job, getting a driver’s license or just being able to help their children with their homework).”

Participants completed several tests designed to measure their cognitive and academic abilities. To assess reading fluency, the researchers timed the participants for one minute as they read lists of real Turkish words and made-up words. They measured inhibitory control, which is the mental ability to override an automatic response, using an emotional Stroop task. In this task, participants looked at happy and sad face emojis on a screen and had to state the opposite emotion out loud.

The researchers also measured nonverbal intelligence using Raven’s Colored Progressive Matrices. This task requires participants to look at a series of visual patterns with a missing piece and select the correct option to complete the design. By using visual puzzles, the task assesses reasoning abilities without relying on spoken or written language.

The illiterate group averaged about 12 out of 36 on this test, versus about 30 for the literate group. The authors note that the illiterate group’s scores match those of other adults with very little schooling and reflect limited education rather than cognitive impairment.

Finally, the authors tested false-belief understanding using two short stories, presented orally with the help of artificial intelligence-enhanced images. In one story, a boy hides a chocolate bar in a drawer, and his sister moves it to a toy box while he is away. However, the boy secretly watches her move it through a window.

The participants were asked where the sister thinks the boy will look for the chocolate. Answering correctly requires understanding the sister’s mistaken belief about the boy’s knowledge.

Under the stricter of two scoring methods, literate participants answered the belief questions correctly about 80 percent of the time, compared with about 20 percent for the illiterate participants. The illiterate group still answered a basic comprehension question about the stories correctly 87 percent of the time, suggesting they followed the narratives.

“We expected to see large individual differences (within and between groups) in false-belief understanding but the gap in understanding between the two groups was the most striking,” Gedik noted.

When looking at the cognitive predictors, reading fluency emerged as the strongest factor in explaining these differences. Reading scores alone accounted for 49 percent of the variance, or statistical variation, in the participants’ false-belief performance. Age accounted for an additional 6 percent of the variation, with older adults tending to score slightly lower.

Reading fluency, however, was almost perfectly tied to which group a participant was in and to years of schooling, and the results looked nearly identical when the researchers used education instead. As a result, the authors treat reading fluency as a stand-in for literacy and the education that typically accompanies it, rather than reading alone.

The illiterate adults also scored lower on the measure of inhibitory control, averaging about 6 out of 10 compared with about 9 for the literate adults. However, once the researchers factored in reading fluency, individual differences in intelligence and impulse control did not add any predictive power. The authors suggest that literacy, and the language and thinking skills that come with it, may account for much of the gap. They caution, though, that with 76 participants these null results should not be taken to mean intelligence and impulse control are unimportant for this kind of reasoning.

“The take home message is that literacy is not as simple as it seems and has consequences that go well beyond being able to decode written words,” Gedik explained.

The findings are in line with research covered by PsyPost in 2024, which found that lifetime exposure to fiction is linked to better theory of mind and perspective-taking. Both studies connect reading to improved social understanding. This new study measured reading at a much more fundamental level, though, focusing on basic literacy and word-reading fluency in adults rather than engagement with fictional narratives.

There are a few things to keep in mind about these findings. The study is correlational, meaning it cannot confirm that learning to read directly causes improvements in social reasoning. The sample was small, and the literate group was unusually highly educated, averaging 17 years of schooling with many graduate degrees, which widens the gap between the groups.

“The biggest misconception would be to interpret the study as saying that illiterate people don’t understand other people’s minds or that they fail to take perspectives of others,” Gedik clarified. “Our task measures one relatively specific form of perspective-taking. So we should not equate performance on this task with someone’s general capacity for empathy and social understanding, broadly speaking.”

Earlier accounts show that illiterate women understand basic deception, though standard false-belief tasks assess more advanced reasoning that the authors suggest may be strengthened by the kind of practice reading provides.

“As discussed in the paper, similar demographics have been documented to make references to perspective-taking or other forms of theory of mind,” Gedik added. “The question is: to what extent is this automatized/mastered?”

The sample was entirely female and specific to Turkish speakers, so the results might differ for men or for people from other cultural and linguistic backgrounds. Because illiterate women in this demographic often experience highly restricted social environments, their lower performance might also reflect a lack of diverse social interaction rather than reading itself.

“Next on the list is testing ideally the same participants (or participants similar to this demographic) on a nonverbal ToM (theory of mind) task to rule out language related limitations,” Gedik noted. “The other (perhaps obvious) next step would be longitudinal research following people as they acquire literacy and documenting changes in their ToM performance.”

“While we feel that these findings are significant and important, we are at the beginning of a long journey,” he said of the research’s practical impact. “We will need many more studies (ideally longitudinal and cross-cultural) to more confidently propose the effects of these findings.”

“Readers who are broadly interested in this line of work and population can find further literature and experimental studies in Gedik (2026): Literacy at Work: Ultimate Native Language Attainment, published by De Gruyter,” he concluded.

The study, “Literacy predicts performance on a second-order false-belief task,” was authored by Tan Arda Gedik, Vania De la Garza, and Ewa Dabrowska.