Virtual reality may help see past masking in children’s neurodevelopmental assessments

Children with attention-deficit/hyperactivity disorder (ADHD) and autism spectrum disorder (ASD) frequently alter or suppress their natural behavioral presentation when they know they are being observed, a phenomenon clinicians and researchers call masking or camouflaging. In a Perspective article published in Frontiers in Health Services, Crystal Goh of Cognitive Leap Solutions and Albert Rizzo of the University of Southern California’s Institute for Creative Technologies argue that this dynamic may be quietly undermining the very assessments designed to identify neurodevelopmental conditions — and that immersive virtual reality could offer a way to see through it. Their proposal is carefully framed as a hypothesis: the authors are explicit that no study has yet directly demonstrated that children camouflage less in virtual reality than in conventional clinician-administered testing.

The scale of the problem they describe is substantial. A pooled prevalence estimate places ADHD at 7.2% of children worldwide (95% confidence interval 6.7 to 7.8), and co-occurrence between ADHD and autism has been documented in up to 43.8% of autistic children. Diagnosis is frequently missed or delayed, particularly in girls, children from minority ethnic backgrounds, and those in under-resourced settings, and waiting times across healthcare systems routinely extend to years. The tools in widest use — parent and teacher rating scales and standardized performance tests — are vulnerable to observer bias, rater expectation effects, and cultural variation in what counts as atypical behavior.

Masking is most extensively documented in autism, where it encompasses mirroring social behavior, suppressing self-stimulatory movement, and scripting interactions. Those strategies carry documented personal costs, including heightened anxiety, burnout, and systematically delayed identification, especially in girls. Camouflaging in autism has been associated with elevated hair cortisol concentration, although that association did not hold within twin pairs, suggesting familial confounding. Neuroimaging and neurocognitive work adds further texture: in autistic women, camouflaging has been linked to ventromedial prefrontal cortex responses during self-representation, and the first neurocognitive study of masking found faster N170 event-related potential latencies and attenuated late positive potentials to facial emotion in adolescents who passed as non-autistic.

Evidence that camouflaging extends to ADHD is newer and thinner. Adults with ADHD report more camouflaging than comparison participants but less than autistic adults, and it is autistic traits rather than ADHD traits that predict camouflaging in those studies. Qualitative work suggests adolescent girls with ADHD engage in sustained impression management and behavioral suppression at school. The pediatric evidence base for ADHD masking remains considerably more limited than for autism, and its underlying neurobiology is less well characterized. Across both conditions, however, the practical consequence is the same: a child may appear more regulated during a forty-minute clinical evaluation than during ordinary daily life, producing a structural decoupling between what the assessment captures and the child’s most functionally relevant behavior.

The assessment room itself, the authors argue, embodies precisely the conditions that drive masking — structured novelty, an attentive adult examiner, and implicit performance demand. Parental presence can compound the effect, since children may perform differently on neuropsychological tests when a parent is watching. Goh and Rizzo propose two complementary mechanisms by which virtual reality might reduce this contextual distortion. The first is environmental: once a child is immersed in a virtual environment through a head-mounted display, the physical examiner recedes from the child’s perceptual field, potentially reducing one source of social-performance demand. The second is neurobiological: head-mounted systems can continuously record head, body, and gaze movement alongside cognitive task performance, within the same session and without additional hardware, capturing signals that reflect atypically organized oculomotor and motor circuitry and that are less accessible to deliberate control than overt task responses.

The neurobiological rationale draws on established circuitry. ADHD is associated with alterations across the cortico-striato-thalamo-cerebellar circuit, the subcortical architecture that coordinates cognitive intention with motor action, and atypical striatal dopamine signaling has been linked to characteristic psychomotor patterns — elevated movement frequency, expanded spatial range, reduced stillness, and diminished orientation toward task-relevant stimuli — although dopamine transporter findings are mixed and partly influenced by prior stimulant exposure. Motor differences are not exclusive to ADHD: coordination, gait, and postural stability differences in autism are documented across meta-analytic and large-cohort studies, making continuous motor measurement a shared, cross-diagnostic signal that standardized tools currently leave unmeasured.

A converging body of evidence supports the feasibility of capturing these signals. Merzon and colleagues showed that gaze behavior during a virtual reality prospective memory task distinguished children with ADHD from neurotypical controls, with a classifier trained on eye-movement features reaching an area under the curve of 0.92. Stokes and colleagues demonstrated that the mere presence of distractors disrupted on-task gaze in children with ADHD even when the distractor itself was never fixated — revealing attentional disruption in naturalistic contexts that task accuracy alone cannot detect. Combining eye tracking with conventional continuous performance test measures outperforms either modality alone, with integrated approaches achieving an area under the curve of 0.889 compared with 0.769 for the continuous performance test alone. On the motor side, Jeong and colleagues found that total distance traveled was the strongest single predictor of hyperactivity severity in children with ADHD assessed in immersive virtual reality, while average movement speed showed the broadest associations across symptom dimensions, and Hocking and colleagues showed that low-cost motion capture integrated with virtual reality could quantify whole-body movement efficiency, synchrony, and symmetry in autistic youth, though only in a small feasibility sample.

The authors caution strongly against interpreting these measures as diagnosis-specific biomarkers. Similar behavioral features may arise from attentional regulation, motor coordination, anxiety, sensory characteristics, medication effects, or simple familiarity with the technology, and the diagnostic specificity of motor and gaze signals across neurodevelopmental and psychiatric conditions remains an open empirical question. They point instead to their own implementation, the Virtual Classroom Assessment Tracker, administered via a head-mounted display in a simulated school classroom. The tool rests on a normative dataset of 837 neurotypical children aged 6 to 13 across seven United States clinical sites, and simultaneously captures twelve cognitive and motor metrics during a 13-minute AX-continuous performance task. In the neurotypical normative sample, boys showed greater motor activity and more impulsive responding while girls showed earlier inhibitory regulation; in preliminary clinical data from children with confirmed ADHD profiles, that sex differentiation appeared to converge across most motor metrics, with boys and girls showing broadly equivalent psychomotor profiles. The authors suggest, cautiously, that continuous motor measurement may be less influenced by the sex-typed behavioral expectations that shape observer-based assessment — but whether such measures improve identification of highly compensating or masking children remains untested.

The road from laboratory to clinic, the authors stress, is where most promising tools falter. They catalog the implementation and ethical challenges: gaze and movement data create high-dimensional behavioral records demanding data minimization, encryption, and retention limits; pediatric data may be reused beyond its original purpose without explicit consent and assent; algorithmic opacity could produce scores that perform differently across demographic groups, requiring explainable outputs, subgroup validation, and clinician review; sensor and firmware updates may silently alter measurement properties; costs and licensing may impede scale; and some children — autistic children in particular — may find head-mounted displays intolerable or become dysregulated, meaning the same sensory sensitivities that drive masking may also shape engagement with the technology itself. Cultural adaptation matters equally, since classroom designs, instructions, and norms may not generalize, and family attitudes toward the technology vary across cultural contexts. Clinicians, meanwhile, must shift from test-administrator to environment-manager and data-interpreter, competencies that pre-licensure training does not yet address, and the rapid pace of psychomotor development between ages 6 and 13 means normative adequacy is an ongoing calibration challenge rather than a one-time validation milestone.

The authors position virtual reality assessment, at its current stage of evidence, as an adjunctive source of standardized behavioral information — complementing developmental history, clinical interview, parent and teacher report, and validated rating scales, with the clinician remaining responsible for integration. They are explicit about what the literature does and does not show: feasibility and measurement potential, not incremental diagnostic value and not demonstrated reduction in masking. Testing the central hypothesis will require within-participant studies comparing conventional and immersive conditions, ideally with randomized order and experimental manipulation of observer presence, direct measurement of camouflaging, assessment of gaming and virtual reality familiarity, and blinded clinical classification to determine whether the new data improve accuracy beyond established methods. Prospective multisite work must establish test–retest reliability, developmental norms, demographic and cultural generalizability, accessibility, acceptability, usability, cost-effectiveness, and independent replication — particularly where commercially involved investigators evaluate their own systems. Whether immersive measurement can reach what masking conceals, the authors conclude, is both testable and worth testing.

Subject of Research: The potential role of immersive virtual reality in reducing masking during neurodevelopmental assessment of children with ADHD and autism

Article Title: Unmasking neurodevelopmental assessment: the potential role for virtual reality as a multimodal clinical adjunct

Article References: Goh, C. C. W., & Rizzo, A. S. (2026). Unmasking neurodevelopmental assessment: the potential role for virtual reality as a multimodal clinical adjunct. Frontiers in Health Services, 6, Article 1927026. https://doi.org/10.3389/frhs.2026.1927026

Image Credits: AI Generated

DOI: 10.3389/frhs.2026.1927026

Keywords: virtual reality, ADHD, autism, masking, camouflaging, neurodevelopmental assessment, eye tracking, motor behavior, psychomotor measurement, ecological validity, clinical implementation, pediatric diagnostics