Cognitive Accessibility on the Web: The Limits of WCAG and Pathways Toward Person-Centered Digital Inclusion

Introduction: Why WCAG Compliance Does Not Guarantee Cognitive Accessibility

WCAG compliance establishes necessary technical requirements, but it does not prove that a person understands the information, completes a task, or maintains control during the interaction. A study with users who have dyslexia identified persistent difficulties in accessing sites declared compliant, using accessibility tests and A/B tests to identify problems that WCAG 2.2 did not resolve (Jianan, 2025). Professional evaluation also reveals a gap between formal compliance and user experience, along with the limitations of automated tools (Pereira & Duarte, 2025). Cognitive accessibility therefore requires shifting the focus from the mere presence of requirements toward understanding, action, and participation. This shift requires clarifying concepts, recognizing diverse user profiles, and linking design to functional outcomes.

From Technical Accessibility to Meaningful Participation: Concepts, Frameworks, and the Scope of Cognitive Accessibility

Cognitive accessibility can be understood as the ability to find, interpret, recall, and use digital information in accordance with a person’s capabilities and circumstances. This definition does not reduce the issue to font types, alternative formats, or easy-to-read text. A conceptual framework on information accessibility proposes asking for whom information is produced, how it is created, how it is found, how it is used, and when it is considered reliable (Meltzer et al., 2026). Accessibility thus includes technical, operational, and social dimensions linked to the effective use of information.

The ICF Framework as a Bridge Between Cognitive Impairments, Design, and Digital Participation

The International Classification of Functioning, Disability, and Health (ICF) allows us to relate health conditions to functioning and participation, without making the diagnosis a sufficient explanation of the digital experience. Johansson, Steel, and Gustavsson developed a core set of 46 ICF categories for cognitive web accessibility, linked to ISO 21801-1 and translated into practical design considerations (Johansson et al., 2025). The value of this approach lies in connecting lived experiences, interface decisions, and participatory outcomes. It also limits an individualizing interpretation: difficulty may arise from the relationship between the person, the task, the content, and the environment.

Understanding information accessibility beyond technically accessible formats

A document in audio, Braille, Easy Read, or a screen reader-compatible format may still be difficult to locate, interpret, or trust. Information accessibility depends on the alignment between purpose, structure, language, context of use, and possibilities for action (Meltzer et al., 2026). Multimodal duplication can improve access when combined with plain language and augmentative and alternative communication, according to the views of social service professionals (Miroshnichenko & Gennadievna, 2025). Format alone does not guarantee comprehension or autonomy.

Diversity of Profiles and Contexts: Dyslexia, Aphasia, Intellectual Disability, Autism, and Mental Health

Cognitive profiles do not form a homogeneous group. Dyslexia can affect reading, visual scanning, and the recognition of relationships between elements; aphasia impairs aspects of language following a brain injury, including comprehension, reading, and writing (Jianan, 2025; Lee et al., 2025). People with intellectual disabilities may require less textual complexity, concrete examples, and predictable sequences; a study on health information found preliminary evidence that simplified texts could be understood by adults with very low reading levels (Benson‐Goldberg et al., 2024). In autism, ADHD, and other neurodivergent conditions, preferences regarding movement, visual density, color, working memory, and executive function may differ among users (Komala, 2026; Lalitha, 2025). Mental health and age-related cognitive decline introduce situational variations that make it impossible to design for a single profile (Khomenko, 2026; Miroshnichenko & Gennadievna, 2025). This diversity leads to the following problem: a website may meet technical criteria yet still impose disproportionate cognitive demands.

When WCAG is met but the experience remains inaccessible

The gap between conformance and understandable use arises when verifiable criteria fail to capture the mental load of a complete task. Conformance can coexist with contextually ambiguous links, dense text, unexpected sequences, excessive steps, or errors that are difficult to recover from. Research with people with dyslexia confirms that a declaration of compliance alone does not predict a successful interaction (Jianan, 2025). Professional evaluation identifies the same tension between technical compliance and user needs, especially when automated tools do not adequately assess mobile content or situational experiences (Pereira & Duarte, 2025).

Difficulties with comprehension, reading, and navigation on compliant websites for people with dyslexia

For people with dyslexia, the problem is not limited to whether the text is technically perceivable. Page layout, information density, instructions, and the relationship between controls can increase the effort required to read and locate an action. The study of compliant websites used accessibility tests and A/B testing to identify barriers experienced by users, revealing a discrepancy between what WCAG stipulates and what people actually need when accessing content (Jianan, 2025). The evidence does not allow us to attribute all difficulties to a single visual trait or to establish a universal solution. It does, however, allow us to assert that evaluation must incorporate tasks, preferences, and outcomes of users with dyslexia, rather than inferring the user experience based solely on the code.

Cognitive Load in Self-Service Systems, Medical Applications, and Data-Rich Repositories

Self-service systems centralize decision-making, contextual memory, and error handling. An evaluation of 20 self-service technologies in Japan linked interaction modalities, perception, action, and interface components to cognitive processes, and showed that different design practices affect interaction (Mäkipää & Iijima, 2025). In medical applications, a review combined with a heuristic evaluation of MyChart, Medisafe, and Ada identified six barriers: excessive navigation depth, information overload, unpredictable behavior, insufficient error recovery, lack of multimodal communication, and limited personalization (Khomenko, 2026). In data repositories, the volume of metadata, the shift between general and specific searches, and the loss of context create burdens that the FAIR principles do not address on their own (Aga, 2026). These observations converge on a single idea: cognitive accessibility depends on how decisions and memory are distributed during the task, not just on the availability of the content.

Complex Language and the Consequences of Inaccessibility in Education, Health Care, and Public Services

Abstract language and inconsistent sequencing can prevent a person from understanding health information and participating in decision-making. In national palliative care resources, eight people with cognitive disabilities pointed out dense vocabulary, abstract phrasing, and irregular order, but valued segmentation, plain language, clear visual design, and questions directed at professionals (Er-raji et al., 2026). In public health, simplifying COVID-19 documents using complexity-reduction guidelines yielded initial evidence of improved readability and comprehension among adults with intellectual and developmental disabilities (Benson‐Goldberg et al., 2024). Online education faces a similar diversity of challenges: a systematic review describes barriers to teachers’ implementation of universal design and calls for evidence-based strategies (Yang et al., 2024). In public services, when information cannot be understood, decision-making, self-care, and effective access are restricted—even if the webpage meets formal requirements.

Evaluating and Ensuring Cognitive Accessibility: Implementation Barriers and New Tools

Assessing cognitive accessibility requires combining technical inspection, task observation, and the involvement of people with direct experience. Professional practice reveals gaps in training, limitations of automation, and difficulties in translating user needs into organizational processes (Pereira & Duarte, 2025). Content production adds another layer: writers, designers, developers, and institutional leaders make decisions that can either enhance or undermine value throughout the service. Therefore, evaluation should not be viewed as a final phase, but rather as a cycle of design, testing, review, and monitoring.

Limitations of Manual Evaluation, Monitoring, and Automation Using Language Models

Manual evaluation allows for the examination of comprehension and context, but it is time-consuming and depends on the competence of the person conducting it. An international study involving 27 professionals identified training deficiencies, limitations of automated tools, and a disconnect between compliance and user experience (Pereira & Duarte, 2025). Language models can expand the coverage of certain manual criteria: a study that examined non-textual content, link purpose, and part-of-speech achieved an overall detection rate of 87.18% in its test cases, whereas conventional automated evaluators either missed issues or issued only warnings (López & Pereira, 2024). This result does not equate to validating cognitive understanding, because the study evaluated three WCAG criteria and not participation in real-world tasks. Automation should support prioritization and human review; it should not replace user testing or turn a textual prediction into evidence of accessibility.

The Role of Content Creators, Professionals, and Users in Design and Evaluation

Content creators determine the clarity of titles, instructions, examples, sequences, and error messages. Interviews with 15 content creators from Finnish educational institutions, companies, and labor unions identified a lack of time, insufficient knowledge, and weak organizational support as barriers; facilitating conditions included clear guidelines, competent staff, and assistive technology (Kärpänen & Pöyry, 2025). Interviews with entrepreneurs also described obstacles to designing understandable services, despite the expected benefits for users and organizations (Kärpänen, 2024). Professionals need training and shared methods, while users must be involved in problem definition, not just in final validation. The participation of people with aphasia, their family members, and healthcare professionals has been incorporated into a co-design protocol to create a browser extension, training tools, and specific guidelines (Lee et al., 2025).

Co-design, empathetic communication, and personalization through tools such as UserLens

Co-design makes it possible to transform experiences of exclusion into observable decisions regarding content and interface. The participatory review of palliative care resources yielded concrete recommendations and has already guided institutional reviews, although its sample was small and its scope contextual (Er-raji et al., 2026). Empathetic communication about barriers can help those who produce and consume content understand their effects from the perspective of people with disabilities, rather than viewing WCAG as isolated technical documentation (Pascual & Granollers, 2025). UserLens proposes that users adjust layout, colors, interaction, and presentation in real time using profiles for autism, dyslexia, low vision, or ADHD (Komala, 2026). This solution offers control, but its usefulness requires independent evaluation regarding usability, stability, privacy, and effects on specific tasks. Customization should complement well-structured content and not shift the entire responsibility for accessibility onto the user.

Critical Synthesis: From Compliance Logic to Autonomy and Adaptive Design

The literature reviewed shifts the focus of evaluation from the technical properties of a webpage to the relationship between the person, information, task, and context. WCAG retains value as a common foundation, but its criteria do not fully describe comprehension, cognitive load, or resilience. Cognitive accessibility must integrate design, content, institutional support, and user participation. This synthesis highlights a tension between shared rules and individual adaptations.

Standardization versus personalization of cognitively accessible experiences

Standardization provides consistency and allows for the establishment of minimum requirements, while personalization acknowledges differences between individuals and moments of use. A user autonomy framework proposes adjustments to contrast, typography, motion, and scale through a configurable mode, without imposing a single experience (Lalitha, 2025). UserLens extends this logic to profiles and page reorganization (Komala, 2026). However, personalization options can increase decision-making and cognitive load if they are numerous, opaque, or difficult to configure. The solution is not to choose between standards and preferences, but to establish clear minimums and offer understandable adaptations.

Accessibility as Support for Autonomy, Not as a List of Requirements

Digital autonomy involves understanding options, anticipating consequences, correcting errors, and making decisions without constantly relying on someone else. Palliative care resources were valued when they included specific questions for discussing with professionals, because they supported participation in decision-making (Er-raji et al., 2026). In medical applications, error recovery, stable navigation, and personalization are related to tasks whose failure can affect adherence and safety (Khomenko, 2026). Therefore, a list of requirements must translate into context-specific outcomes: what the person can do, with how much effort, and under what conditions.

Cross-cutting benefits of cognitive accessibility for care, learning, and participation

Cognitive improvements can benefit people with and without disabilities. In a psychophysiological study involving 20 participants without disabilities, improvements in cognitive accessibility were associated with greater cognitive engagement during reading, while the absence of such improvements was linked to a decline in that engagement over time (Ekin et al., 2025). In online education, universal design aims to address student diversity, although its implementation faces pedagogical and organizational barriers (Yang et al., 2024). These findings support a broad rationale but do not justify disregarding the specific needs of those who require individualized supports.

Gaps in the Evidence and Priorities for Overcoming Current WCAG Limitations

The available evidence identifies problems with varying clarity and still offers few comparable metrics. There are qualitative studies, protocols, reviews, and heuristic evaluations, but longitudinal comparisons linking design changes to sustained comprehension and engagement are scarce. Future research should avoid treating cognitive accessibility as a stable property of the interface.

Lack of Operational Criteria and Comparable Metrics for Comprehension and Cognitive Load

WCAG allows for the verification of specific conditions, but the reviewed literature does not converge on a single metric for comprehension, effort, control, or recovery. Tests involving individuals with dyslexia, self-service studies, and evaluations of medical applications use different methods, making it difficult to compare results across contexts (Jianan, 2025; Khomenko, 2026; Mäkipää & Iijima, 2025). Language models can expand the detection of certain problems, although they do not yet measure situated comprehension on their own (López & Pereira, 2024). Operational criteria are needed that combine task outcomes, demonstrated comprehension, errors, time, confidence, and the ability to request or withdraw support.

Need for Longitudinal, Participatory, and Context-Specific Evidence

Participatory studies show that users with cognitive disabilities identify problems that a formal inspection may overlook (Er-raji et al., 2026; Lee et al., 2025). However, several studies are exploratory, have small sample sizes, or feature designs still under development; therefore, they do not establish lasting effects or generalizability across domains. Priorities include longitudinal follow-ups, comparisons between user profiles, testing in education, health care, public services, and repositories, and evaluation of adaptations during real-world tasks. Research should also investigate whether personalization reduces user burden or shifts it toward configuration and maintenance.

Conclusion: Toward Web Guidelines That Measure Comprehension, Control, and Participation

Compliance with WCAG should remain a technical threshold, not the end goal of cognitive accessibility. Evidence shows that comprehension, language, sequencing, contextual memory, and error recovery can fail even on compliant sites (Jianan, 2025; Pereira & Duarte, 2025). More appropriate guidelines should measure whether people find reliable information, understand their options, complete tasks, and maintain control with adjustable supports. Such a framework requires common criteria, assisted automated evaluation, and direct participation by diverse users. The transition from compliance to autonomy will be valid only if it demonstrates improvements in actual participation, not merely in interface metrics.

References

• Aga, D. (2026). Beyond FAIR: Designing cognitive accessibility in data-rich repositories. Zenodo (CERN European Organization for Nuclear Research). https://doi.org/10.5281/zenodo.20788950
• Benson‐Goldberg, S., Geist, L. A., & Erickson, K. A. (2024). Simplified COVID‐19 guidance for adults with intellectual and developmental disabilities. Journal of Applied Research in Intellectual Disabilities, 37(3), e13222. https://doi.org/10.1111/jar.13222
• Ekin, M., Krejtz, K., Duarte, C., Pereira, L. S., Marcus-Quinn, A., & Krejtz, I. (2025). Impact of web accessibility on cognitive engagement in individuals without disabilities: Evidence from a psychophysiological study. PLoS ONE, 20(7), e0328552. https://doi.org/10.1371/journal.pone.0328552
• Er-raji, N., Tahir, M., & Martin, L. (2026). A cognitive accessibility review of national palliative care resources for people with cognitive disabilities. BMC Palliative Care, 25(1). https://doi.org/10.1186/s12904-026-02067-3
• Jianan, L. (2025). WCAG compliance is not enough for websites: Accessibility issues and solutions faced by users with dyslexia when accessing WCAG-compliant websites. Jönköping University Library (Jönköping University). https://hj.diva-portal.org/smash/get/diva2:2015804/FULLTEXT01
• Johansson, S., Steel, E., & Gustavsson, C. (2025). An ICF core set for designing cognitive web accessibility: a bridge between design and the implications of cognitive impairments. Universal Access in the Information Society, 24(4), 3599–3614. https://doi.org/10.1007/s10209-025-01262-y
• Kärpänen, T. (2024). Barriers to creating value with cognitive accessibility features in digital services. Universal Access in the Information Society, 24(2), 1379–1392. https://doi.org/10.1007/s10209-024-01151-w
• Kärpänen, T., & Pöyry, E. (2025). Evaluating content creators’ perceptions of cognitively accessible digital services. Behaviour and Information Technology, 45(5), 855–867. https://doi.org/10.1080/0144929x.2025.2535742
• Khomenko, V. (2026). Accessibility barriers in mobile medical applications for users with cognitive impairments: analysis, impact assessment, and design recommendations. Zenodo (CERN European Organization for Nuclear Research). https://doi.org/10.5281/zenodo.20629608
• Komala, Mrs. B. (2026). UserLens—a web accessibility tool for autism, dyslexia, and cognitive disorders. International Journal for Research in Applied Science and Engineering Technology, 14(5), 318–323. https://doi.org/10.22214/ijraset.2026.81604
• Lalitha, A. (2025). Beyond compliance: A user-autonomy framework for inclusive and customizable web accessibility [Preprint]. In arXiv (Cornell University). https://doi.org/10.48550/arxiv.2506.10324
• Lee, J., Worthy, P., Deslandes, R., Burton, B., Copland, D. A., Jamieson, P., Barron, K., Togher, L., Shrubsole, K., Shiggins, C., Campbell, J., Hill, A. J., Wiles, J., Haslam, S. A., & Wallace, S. J. (2025). Bridging the digital divide for people with aphasia: a study protocol for co-designing web accessibility tools and guidelines. BMJ Open, 15(8), e099273. https://doi.org/10.1136/bmjopen-2025-099273
• López, J. M., & Pereira, J. (2024). Turning manual web accessibility success criteria into automatic: an LLM-based approach. Universal Access in the Information Society, 24(1), 837–852. https://doi.org/10.1007/s10209-024-01108-z
• Mäkipää, J.-P., & Iijima, J. (2025). Accessibility evaluation of interaction modalities and cognitive processes of self-service technology user interfaces in Japan. Information and Software Technology, 190, 107970. https://doi.org/10.1016/j.infsof.2025.107970
• Meltzer, A., Wehbe, A., & Barnes, E. (2026). What does it mean for information to be accessible to people with disabilities? An information accessibility conceptual framework. Disability & Society, 41(5), 1209–1231. https://doi.org/10.1080/09687599.2025.2608581
• Pascual, A., & Granollers, A. (2025). Improving communication about web accessibility barriers through an empathetic perspective to enhance the accessible user experience (AUX). Universal Access in the Information Society, 24(4), 3143–3163. https://doi.org/10.1007/s10209-025-01188-5
• Pereira, L. S., & Duarte, C. (2025). Evaluating and monitoring digital accessibility: practitioners’ perspectives on challenges and opportunities. Universal Access in the Information Society, 24(3), 2553–2571. https://doi.org/10.1007/s10209-025-01210-w
• Yang, M., Duha, M. S. U., Kirsch, B. A., Glaser, N. J., Crompton, H., & Luo, T. (2024). Universal design in online education: A systematic review. Distance Education, 45(1), 23–59. https://doi.org/10.1080/01587919.2024.2303494
• Miroshnychenko, O. A., & Gennadievna, V. I. (2025). Accessibility of the information environment for people with mental disabilities. Autism and Developmental Disorders, 23(1), 40–47. https://doi.org/10.17759/autdd.2025230105