Summary:
A cognitive psychology study from the University of Tokyo reveals that individuals with and without autism spectrum disorder (ASD) process verbal expressions of probability, such as “likely” or “possible”, through distinctly different decision-making frameworks. Published in Psychonomic Bulletin & Review, the research evaluated 138 participants weighing medical treatment recommendations framed in either words or percentages.
Researchers identified three distinct cognitive profiles of decision-makers, discovering that individuals with ASD were heavily swayed by positive or negative verbal framing rather than the underlying numerical probabilities they inferred. The findings underscore the need for clinicians and educators to pair verbal risk estimates with explicit numerical data and interactive dialogue to ensure informed medical decisions.
Key Facts:
- Three Decision-Making Profiles: Cluster analysis across 138 adults revealed three distinct cognitive approaches to uncertainty: framing-driven (predominantly individuals with ASD), number-driven (predominantly neurotypical controls), and balanced integrators (a mix of both cohorts).
- Susceptibility to Verbal Framing: Participants with ASD assigned more uniform numerical values across different verbal phrases, yet their final medical choices were strongly driven by whether the wording was framed positively (“possible”) or negatively (“unlikely”).
- Clinical Communication Imperative: Subtle shifts in wording significantly alter high-stakes choices; researchers urge healthcare professionals to combine clear percentages with contextual dialogue rather than relying solely on ambiguous probability adjectives.
Source: University of Tokyo
Whether weighing the side effects of a new medication, interpreting climate forecasts, or navigating financial investments, human beings constantly make critical choices based on uncertain information. In clinical settings, physicians and educators routinely communicate risk using qualitative verbal descriptors, telling patients a treatment is “likely to succeed” or a side effect is “unlikely”, rather than citing exact statistical percentages.
However, verbal probability terms are inherently ambiguous. One person’s “possible” may translate to a 20% likelihood, while another interprets the exact same word as a 60% chance.
For individuals on the autism spectrum—a neurodevelopmental condition characterized by distinct differences in social communication, pragmatic language processing, and cognitive flexibility, interpreting these imprecise verbal cues presents unique challenges.
In a study published in Psychonomic Bulletin & Review, researchers from the University of Tokyo investigated how adults with and without autism spectrum disorder translate verbal and numerical probabilities into concrete medical decisions.
Led by Hidehito Honda, Hirokazu Kumazaki, and Professor Kazuhiro Ueda from the Graduate School of Arts and Sciences, the team uncovered three distinct cognitive profiles governing how people process uncertainty, demonstrating that subtle phrasing choices heavily steer decision-making in autistic individuals.
Testing Medical Risk in Words and Numbers
To evaluate how linguistic framing interacts with statistical reasoning, the researchers recruited 138 adult participants: 48 individuals diagnosed with ASD and 90 neurotypical control participants.
Participants were presented with a realistic clinical decision-making scenario. They were asked to imagine advising a close friend considering an expensive medical treatment for chronic migraines, rating how strongly they would recommend the therapy based on a physician’s prognosis.
Across trials, the doctor’s prognosis was systematically varied across two formats:
- Explicit Numerical Probabilities: Exact percentages detailing treatment efficacy (e.g., “a 30% chance the treatment will be successful”).
- Verbal Probability Expressions: Qualitative descriptors carrying either positive directional framing (e.g., “it is possible”) or negative directional framing (e.g., “it is unlikely to succeed”).
Following the recommendation task, participants were also asked to explicitly translate each verbal expression into a numerical percentage, allowing researchers to separate a person’s internal numerical estimate from the emotional or directional pull of the phrasing itself.
Three Distinct Profiles of Decision-Makers
“ASD is often associated with differences in communication and information processing, so we wondered if people with ASD might interpret expressions of uncertainty differently,” explained Professor Kazuhiro Ueda. “While we found that participants with ASD tended to assign more similar numerical meanings to different verbal phrases, we more importantly identified three distinct patterns of decision-making, revealing variation both within and across the groups.”
When the investigators analyzed how participants weighted wording versus implied numbers, three distinct behavioral clusters emerged:
- Group 1 (Framing-Driven Decision-Makers): Composed predominantly of individuals with ASD, this cohort made choices heavily dictated by the positive or negative polarity of the wording, placing far less weight on the actual numerical probability they had assigned to those words.
- Group 2 (Number-Driven Decision-Makers): Composed mostly of neurotypical participants without ASD, this group relied primarily on the underlying numerical likelihood they inferred from the doctor’s words, remaining relatively insulated from positive or negative linguistic framing.
- Group 3 (Balanced Integrators): A mixed cohort containing both autistic and non-autistic individuals whose decisions were equally shaped by both verbal framing and inferred numerical probability.
Improving Informed Consent in Healthcare and Education
The researchers note that because the study sample encompassed a wide range of ages and intellectual abilities, further longitudinal and cognitive testing is required to determine the exact psychological mechanisms—such as differences in pragmatic inference or cognitive load—that place an individual into a specific decision-making cluster.
Nevertheless, the real-world implications for clinical communication are immediate. If a physician describes a 30% success rate as “possible” versus “unlikely,” a framing-sensitive patient may reach opposite treatment conclusions based entirely on word choice rather than clinical reality.
“Even so, our findings suggest that the way we communicate uncertainty deserves more attention, especially when people make important decisions about health, education or everyday life. We could see that even subtle differences in wording can influence people’s decisions,” said Ueda.
To ensure genuine informed consent and equitable autonomy, the authors advocate for a shift toward explicit, transparent communication protocols.
“We would like to see greater awareness that communicating uncertainty is not simply a matter of choosing understandable words,” Ueda concluded. “It is important to clearly explain the context and background of the information being provided and to understand which aspects of that information influence each person’s judgments and decisions. We hope our findings will encourage professionals, particularly in healthcare and education, to explain uncertain information more explicitly and check whether their intended meaning has been understood.”
Questions and Answers
How do people with autism interpret probability words like “likely” or “possible”?
The University of Tokyo study found that participants with autism spectrum disorder (ASD) tended to assign more similar numerical percentages across different verbal phrases compared to neurotypical peers. More importantly, when making medical decisions, individuals with ASD were far more influenced by whether a phrase was framed positively or negatively rather than the underlying numerical probability.
What are the three types of decision-makers identified in the study?
Researchers identified three cognitive profiles: Group 1 (mostly individuals with ASD), who rely heavily on positive or negative wording; Group 2 (mostly neurotypical adults), who base choices primarily on inferred numerical probabilities; and Group 3 (a mix of both groups), who weigh wording and numbers equally.
How should doctors and educators communicate risk based on these findings?
Professionals should avoid relying solely on vague verbal terms like “possible” or “unlikely.” Instead, they should provide clear numerical percentages alongside explicit context, and engage in two-way dialogue to verify that the patient or student has understood the intended meaning.
Funding:
This research was supported by Grants-in-Aid for Scientific Research (Grant Numbers JP22H03911 and JP25H01131).
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by our staff.
About this autism Research:
- Media Contact: Nicola Burghall
- Source: University of Tokyo
- Image Credit: Image credited to Neuroscience News
- Original Research is Open Access: Psychonomic Bulletin & Review (Oct 1, 2026). “Understanding verbal and numerical probability information and decision-making among individuals with and without autism spectrum disorder.” Authors: Hidehito Honda, Hirokazu Kumazaki, and Kazuhiro Ueda.
- DOI: 10.3758/s13423-026-03032-0
Abstract
Understanding verbal and numerical probability information and decision-making among individuals with and without autism spectrum disorder
This study examined how individuals with and without autism spectrum disorder (ASD) interpret probabilistic information expressed verbally or numerically, and how such interpretations influence decision-making.
Two experiments were conducted: Experiment 1 included 48 participants with ASD and 41 participants without ASD in the final analyses, and Experiment 2 included an independent sample of 49 participants without ASD. Participants translated verbal probability expressions (e.g., likely, doubtful) into numerical estimates and made hypothetical treatment recommendations based on either verbal or numerical probabilities.
Compared with non-ASD individuals, participants with ASD showed less differentiated numerical translations across verbal probability expressions, indicating greater similarity in the numerical meanings assigned to different expressions. A hierarchical Bayesian analysis followed by cluster analysis identified three decision-making profiles.
Notably, individuals with ASD were more likely to belong to a cluster characterized by stronger sensitivity to the directionality (positive vs. negative framing) of verbal probability expressions and weaker reliance on numerical translation. In contrast, non-ASD individuals were more frequently represented in a cluster that relied primarily on numerical information.
These findings suggest that individuals with and without ASD differ in how verbal probability expressions are interpreted and incorporated into decision making, highlighting the importance of considering individual differences in the communication of uncertainty.
