Beyond Diagnosis: Rethinking Cognitive Differences in Children
Picture from Yan Kurkau, Canva
All children do not follow identical paths of development. Within the same classroom, some children grasp new concepts quickly but struggle to sustain focus, while others appear attentive yet find it difficult to organize their thoughts or manage their impulses. These variations are often subtle at first, but over time they shape how a child learns, behaves, and experiences school and life. Teachers and parents often respond to these patterns through decisions about expectations, classroom strategies, and whether additional help is needed. In practice, wandering attention may be understood as distraction in one context, as fatigue or lack of motivation in another. Those decisions depend on how well everyday observations fit the interpretive lens used to make sense of behaviour. The closer the lens aligns with how a child actually thinks, the more likely the response will help rather than frustrate the learner.
Some of these variations may signal more persistent difficulties, which educational and clinical systems attempt to identify through diagnostic categories such as Autism Spectrum Disorder (ASD) or Attention-Deficit/Hyperactivity Disorder (ADHD). Such diagnoses can help secure supports that enable children to navigate school and daily life. They offer a shared language that allows educators, clinicians, and families to coordinate support. However, children who share the same diagnosis may have remarkably different levels of functioning and different support needs. Two children carrying the same label may struggle for entirely different cognitive reasons and respond to different forms of assistance. One may benefit from structured reminders, while another may benefit more from reduced sensory distraction, even though both fall under the same category.
At the same time, children whose difficulties do not align neatly with formal criteria may receive no diagnosis at all, and their challenges may go unrecognized. Across both diagnosed and non-diagnosed children, the specific cognitive needs underlying each child’s challenges and strengths risk being overlooked. The categories succeed at identifying groups of children, yet they do not always capture how each child learns. The practical consequence is a mismatch: support may follow the label rather than the learning difficulty itself.
Looking Beyond Labels: A Data-Driven Approach
To understand cognitive heterogeneity among children and adolescents with or without neurodevelopmental disorders, a team at Western University led by Sarah Al-Saoud and Dr. Emma Duerden used machine learning approaches to identify cognitive profiles using data collected between 2019 and 2020. Instead of beginning with diagnostic labels, the researchers first examined patterns of thinking and only afterward asked how those patterns related to diagnosis. This reversal matters because it tests whether diagnostic categories reflect underlying learning structure or only approximate it. The team aimed to determine whether data-driven cognitive profiles corresponded with the traditional diagnoses individuals had received.
To examine this, the researchers used data from children and adolescents who had received different diagnoses, including ADHD (510 individuals), ASD (42 individuals), both ADHD and ASD (42 individuals), and no diagnosis (935 individuals). Participants completed a 12-task battery assessing cognitive functions such as short-term memory, verbal ability, and reasoning. Each activity captured cognitive processes used in school, such as remembering instructions, resisting distraction, or mentally rotating shapes. Rather than measuring academic knowledge directly, the tasks assessed underlying cognitive abilities that support learning across contexts. The tasks were short, “game-like” versions of standard cognitive psychology measures administered through an online research platform.
Grouping Children by Patterns of Cognition
After data collection and cleaning, machine learning techniques were used to group children based on similarities in their performance across tasks, creating cognitive profiles independent of diagnostic status. The algorithm compared responses across tasks and grouped children who approached problems in similar ways. The analysis produced six distinct profiles. Children were organized by patterns across tasks rather than overall scores. The groups reflected configurations of strengths and challenges rather than simple differences in intelligence or diagnosis. Placement depended on the pattern of abilities, not performance level alone.
Six Profiles, Not One Diagnosis
Across the six profiles, some individuals showed slower responding and reduced task persistence, whereas others displayed strong selective attention, deductive reasoning, or broadly high accuracy across tasks. Some profiles were characterized by relative delays in memory, while others showed advantages in spatial manipulation and working memory. Another profile showed a more impulsive response style, completing many attempts but with lower accuracy. These patterns resemble familiar classroom observations: hesitation, rapid guessing, careful reasoning, or difficulty holding instructions in mind.
Together, these profiles indicate that cognitive differences are not simply a matter of being “good” or “poor” performers. They instead describe how abilities are organized across tasks and domains. A child who works slowly and carefully reflects a different cognitive pattern than one who works quickly and inaccurately. The distinction concerns strategy and processing style rather than effort or willingness to learn.
One of the study’s primary questions was whether these cognitive profiles aligned with traditional diagnostic categories. If diagnostic categories captured cognitive organization, children sharing a diagnosis would tend to appear in the same profile. The results showed that diagnostic labels did not predict group membership and therefore did not reliably organize cognitive functioning. Children with the same diagnosis appeared across multiple profiles, and each profile included children with different diagnostic backgrounds. The categories account only partially for learning differences.
When Diagnosis is Not Enough
Additionally, individuals without a formal diagnosis appeared across the same cognitive profiles as those with diagnoses. These individuals were represented in all identified groups, including those characterized by cognitive difficulties and those showing relative strengths. Thus, variability in cognitive functioning was not confined to clinically diagnosed populations but also extended to non-diagnosed individuals. These results indicate that individuals without a formal diagnosis also have varied cognitive needs. This suggests that meaningful learning differences can be present even in the absence of a formal diagnosis. Support decisions may therefore be informed by patterns of learning rather than waiting until difficulties meet formal diagnostic thresholds.
From Categories to Cognitive Patterns
Evidence for this perspective also comes from earlier research examining executive functioning across typical development, ADHD, and ASD. A previous data-driven study identified three recurring profiles involving flexibility and emotional regulation, inhibition, and working memory and planning. Children from different diagnostic groups appeared within each profile, and typically-developing children showed the same underlying structure. Taken together, results from different approaches suggest that cognitive organization follows developmental dimensions that extend across diagnostic boundaries.
In everyday learning contexts, cognitive profiles reshape how support is designed. A learner with working-memory difficulty may need written steps, previews, or reminders, while a learner who responds impulsively may benefit from pacing and structured response opportunities. Some profiles also include relative strengths, such as strong reasoning or visual processing, allowing instruction to build on what already works while supporting weaker processes. The focus shifts from matching support to a label toward matching support to how learning operates.
These findings should be interpreted cautiously. Different sets of cognitive measures can produce different groupings, and no single method guarantees a uniquely correct number of profiles. The important point is not the precise boundaries of groups but the consistent presence of meaningful differences in learning patterns. The profiles arise from performance within a defined task battery and therefore complement, rather than replace, broader assessment. Diagnostic categories continue to guide communication and access to services, while profile information helps specify what to target within those structures. Longitudinal research will clarify how stable these patterns remain and how they relate to later educational outcomes. Overall, the findings suggest a shift in emphasis. Diagnostic categories describe populations; cognitive organization describes how children learn.
Original Article: Al-Saoud, S., Nichols, E. S., Brossard-Racine, M., Wild, C. J., Norton, L., & Duerden, E. G. (2025). A transdiagnostic examination of cognitive heterogeneity in children and adolescents with neurodevelopmental disorders. Child Neuropsychology, 31(2), 293-311. doi: 10.1080/09297049.2024.2364957