Cognitive accessibility, neurodiversity and ageing

A conformance standard promises testability. That promise sets its limits: a success criterion is written so that an evaluator can decide, from the artefact, whether a page passes. Cognitive accessibility is where that method stops, and the bodies doing the work document the stop rather than their critics.

This eighth of ten units argues that cognitive accessibility is the boundary case of the conformance project rather than the part still to be finished. Attention and memory belong to a task performed by a person, so they cannot attach to an element as a contrast ratio can. The critique then asks whose norm the artefact is measured against, and whether the deficit to be repaired sits in the user or in the design.

The standard writes its own limit

Cognitive accessibility work at the W3C began as user research and has stayed close to it. Seeman and Cooper (2015) organised the barriers of web use by cognitive function: attention, executive function, knowledge, language, literacy, memory, perception and reasoning. A criterion can be written against a control or a colour ratio, while attention belongs to a person doing something.

W3C (2021) later published objectives for content that works for people with cognitive and learning disabilities, from understanding what things are to processes that do not depend on memory. It is supplemental guidance, and it states that following it is not required for conformance. The objective to remove memory demands shows the shape. It names the concrete barriers, among them a long password the user has to recall and a telephone menu that requires remembering a number. It pairs each with a design pattern and a user-testing question, so a team gets a protocol where a criterion is unavailable.

Following this guidance is not required for conformance to WCAG.

The distance between cognitive needs and the standard is a gap in enforceability rather than in knowledge: a conformance regime binds people through what it can test. Friedman and Bryen (2007) reviewed twenty guideline sets, nine addressing cognitive disabilities, and found the recommendations resting on expert agreement rather than on trials. The limit on that guidance is that nobody has to run it, since an obligation is whatever the law or a contract names.

Ageing: one need or several?

Cognitive accessibility would be narrow if it concerned only diagnostic categories. Ageing makes it demographic, and the evidence is thickest here. Czaja and Lee (2006) report that the population over 65 is growing worldwide and that the fastest growing subgroup is those aged 80 and over. They frame the design problem around the cognitive and perceptual demands a system places on its user, who may work on a shared device over a slow connection. Older adults are willing to use technology, they find, but many report usability problems due in part to those demands.

Charness and Boot (2009) give the opposing emphasis, placing attitudes and abilities among the most powerful predictors of use. They conclude that normative age-related changes in ability must be taken into account in design and training. Ability here is a measured variable with a real effect rather than a stereotype to be corrected. Hanson (2010) studied the components of information search in younger and older adults and treated the present difficulties as foreshadowing future trends. Mitzner and colleagues (2010) ran focus groups with 113 older adults and found positive attitudes outnumbered negative ones. That contradicts the image of older adults as unwilling.

Neurodiversity: the deficit may be in the frame

Where the ageing literature argues about how much of the barrier is capability, the neurodiversity literature disputes the frame. Milton (2012) proposed the double empathy problem. Difficulty between autistic and non-autistic people is mutual and relational. The failure usually attributed to a deficit in the autistic person's understanding of other minds is better described as a failure of reciprocity. If the mismatch is mutual, the autistic person is not the faulty component.

Dalton (2013) carried the claim into interaction design, arguing that neurodiversity, like feminist HCI, criticises the implicit singular user. The target is not a subgroup to be added but a universal assumption to be removed. Kapp (2020) supplies the movement's own account, gathering autistic activists and academics to debate who is entitled to define the condition. Kientz and colleagues (2014) show the assumption in a review that organises technology for autism around intervention, communication support, data collection and diagnosis. It treats the autistic person as the subject of intervention rather than its author.

The reframing has evidence, and it stops short of a design

Gillespie-Lynch and colleagues (2017) surveyed 636 adults with varied relationships to autism. They found autistic participants showed more scientifically based knowledge than the others, more often described autism as a neutral difference and more often opposed the medical model.

Zolyomi and colleagues (2019) move from who is in the room to what the room does to them. Interviewing autistic adults about video calling, they built a model in which sensory sensitivities, cognitive load and anxiety shape channel preference. They found participants coping by sensory moderation and masking as non-autistic behaviour.

Sanchez-Gordon and Lujan-Mora (2017) suggest why in a review of 40 studies of accessible massive open online courses. The authors map those studies onto an eight-dimension lifecycle that places verification of compliance apart from validation of user needs satisfaction.

Contested question: where does the deficit sit, and what would settle it?

The disagreement is not about facts that are hard to obtain but about what follows from facts already in evidence. One position holds that the deficit is a capability. Charness and Boot (2009) place ability among the strongest predictors of use. Czaja and Lee (2006) locate much of the usability problem in the demands systems make, and Kientz and colleagues (2014) describe a field organised around intervention and assessment. That position predicts that designs reducing demand should improve performance and lower reported difficulty.

A second position holds that the deficit is relational. Milton (2012) makes the mismatch mutual, Dalton (2013) treats the singular user as an assumption rather than a description, and Kapp (2020) grounds the claim in the movement's testimony. Gillespie-Lynch and colleagues (2017) show autistic adults holding the relevant expertise, and Zolyomi and colleagues (2019) supply the mechanism, since masking is a cost paid to a design not built for the user. A third position concerns the evidence rather than the user. Seeman and Cooper (2015) and the note that followed (W3C, 2021) show cognitive guidance resting on user research and expert agreement. Sanchez-Gordon and Lujan-Mora (2017) show the field holding compliance verification apart from user-need validation.

The two sides rest on evidence of different kinds. The capability case rests on a review of predictors and on a demographic argument, and both measure the user. The relational case rests on a theoretical reframing, on a survey of 636 adults about who holds expertise, and on one interview study of autistic adults. The third rests on a synthesis and on a review of 40 studies, which map how the field is organised rather than test an intervention.

The strongest objection is that the reframing is untested as a design. Gillespie-Lynch and colleagues measure who should be in the room, and Zolyomi and colleagues describe the costs a design imposes, but no source here shows a neurodiversity-informed design outperforming a capability-informed one. A reframing that cannot predict a better outcome is a claim about authority rather than about design.

What would settle it is a comparison this literature lacks. It would take one cognitive demand a system imposes, such as a process that depends on memory, and build two remedies. One remedy would reduce the demand, and the other would help the user meet it. It would measure both against a control, while recording the interface's demand separately from the user's capability. Zolyomi and colleagues (2019) give the mechanism to look for.

What to do while the dispute is open

While the dispute is open, name whose norm the interface encodes. Record the demand the interface makes separately from the user's measured capability, so that a change to either one is visible, on the shared device and the slow connection the user actually has.

Where the argument is heading

The direction of travel is fixed by work that exists rather than by fashion. The W3C's cognitive guidance will probably stay supplemental, since its own groups do not expect it to become a recommendation. Enforcement therefore runs through procurement and law. The route to evidence runs through participation. Gillespie-Lynch and colleagues (2017) make the case for autistic people as partners in research, and Kapp (2020) documents the movement that demanded it. Zolyomi and colleagues (2019) show the design opportunities that open once the costs users pay are treated as data. Dalton's (2013) critique of the singular user is the most portable of these claims, since it applies to any interface built for an average no user occupies. The next unit asks who holds authority in inclusive and participatory research.

The sources this unit rests on

Three entries below are normalised to the house spelling where the published form uses a different variant.

• Czaja, S. J. and Lee, C. C. (2006). The impact of ageing on access to technology. Universal Access in the Information Society. doi:10.1007/s10209-006-0060-x Reported that the fastest growing group of people over 65 is those aged 80 and over, and framed the design problem around the cognitive and perceptual demands systems place on users, who are willing but report usability problems.
• Friedman, M. G. and Bryen, D. N. (2007). Web accessibility design recommendations for people with cognitive disabilities. Technology and Disability. doi:10.3233/tad-2007-19406 Reviewed twenty guideline sets, nine addressing cognitive disabilities, and distilled the recommendations on which they agreed, chiefly simplicity and consistency. The guidance rests on expert consensus rather than on trials.
• Charness, N. and Boot, W. R. (2009). Ageing and information technology use. Current Directions in Psychological Science. doi:10.1111/j.1467-8721.2009.01647.x Placed attitudes and abilities among the strongest predictors of technology use and concluded that normative changes in ability must be taken into account in design and training. It is the sharpest counter to the neurodiversity position.
• Mitzner, T. L., Boron, J. B., Fausset, C. B., et al. (2010). Older adults talk technology: Technology usage and attitudes. Computers in Human Behaviour. doi:10.1016/j.chb.2010.06.020 Ran focus groups with 113 older adults and found positive attitudes outnumbered negative ones, contradicting the stereotype of older adults as afraid or unwilling to use technology. It returns the explanation for low use to the system.
• Hanson, V. L. (2010). Influencing technology adoption by older adults. Interacting with Computers. doi:10.1016/j.intcom.2010.09.001 Examines the components of information search by younger and older adults and reads current problems as foreshadowing future trends, with the long-term implications of designing for older users in view. It turns the ageing literature into a set of design requirements.
• Milton, D. E. M. (2012). On the ontological status of autism: the 'double empathy problem'. Disability & Society. doi:10.1080/09687599.2012.710008 Reframed the deficit attributed to autistic people's understanding of other minds as a question of reciprocity, arguing that the difficulty between autistic and non-autistic people is mutual. It is the basis for the claim that the mismatch, not the person, is the design target.
• Dalton, N. S. (2013). Neurodiversity and HCI. CHI '13 Extended Abstracts on Human Factors in Computing Systems. doi:10.1145/2468356.2468752 Argued that neurodiversity, like feminist HCI, criticises the implicit singular user, so assumptions about cognitive processing need challenging. The plan records a workshop venue; the paper is in the CHI 2013 extended abstracts.
• Seeman, L. and Cooper, M. (2015). Cognitive Accessibility User Research. W3C First Public Working Draft. https://www.w3.org/TR/coga-user-research/ Collected user research on the barriers of web use, organised by cognitive function and by user group, as the basis for later guidance. The plan dates it 2013; the W3C page gives 15 January 2015, and the year is corrected here.
• Kientz, J. A., Goodwin, M. S., Hayes, G. R. and Abowd, G. D. (2014). Interactive Technologies for Autism. Synthesis Lectures on Assistive Technologies. doi:10.1007/978-3-031-01595-3 Reviewed interactive technology for autism and offered a classification scheme, organising the material around interventions, communication support, data collection and assessment and diagnosis. Its scope makes the autistic person the subject of intervention rather than the person who sets the objective.
• Gillespie-Lynch, K., Kapp, S. K., Brooks, P. J., Pickens, J. and Schwartzman, B. (2017). Whose Expertise Is It? Evidence for Autistic Adults as Critical Autism Experts. Frontiers in Psychology. doi:10.3389/fpsyg.2017.00438 Surveyed 636 adults and found autistic participants showed more scientifically based knowledge than the others, more often described autism as a neutral difference and more often opposed the medical model. It supports including autistic adults as partners in research.
• Sanchez-Gordon, S. and Lujan-Mora, S. (2017). Research challenges in accessible MOOCs: a systematic literature review 2008-2016. Universal Access in the Information Society. doi:10.1007/s10209-017-0531-2 Reviewed 40 studies of accessible massive open online courses and mapped them to an eight-dimension lifecycle holding verification of compliance apart from validation of user needs satisfaction. The field's own map separates compliance from the person's needs being met.
• Zolyomi, A., Begel, A., Waldern, J. F., Tang, J., Barnett, M. and Cutrell, E. (2019). Managing Stress: The Needs of Autistic Adults in Video Calling. Proceedings of the ACM on Human-Computer Interaction, 3(CSCW). doi:10.1145/3359236 Interviewed autistic adults about video calling and built a model in which sensory sensitivities, cognitive load and anxiety shape channel preference, finding coping through sensory moderation and masking. The plan calls this workplace research in the CHI proceedings; the paper concerns video calling and appears in the CSCW issue, and the note is corrected here.
• Kapp, S. K. (ed.) (2020). Autistic Community and the Neurodiversity Movement: Stories from the Frontline. Singapore: Palgrave Macmillan. doi:10.1007/978-981-13-8437-0 Gathered autistic activists and academics to describe the events that shaped the neurodiversity movement and to debate epistemic authority in autism research. It gives the movement's own account of the claims the unit tests.
• W3C (2021). Making Content Usable for People with Cognitive and Learning Disabilities. W3C Working Group Note. https://www.w3.org/TR/coga-usable/ Set out six objectives for content that works for people with cognitive and learning disabilities, from understanding what things are to processes that do not depend on memory, each linked to user testing. It states that following it is not required for conformance.