hand, HIID are often static and changes on the users’ abilities are seen in a larger time- scale.
better system adaptations (Kondraske, 1995)(Persad et al., 2007). These processes typi- cally require a large battery of tests with non-technology-related tasks, which assess users’
low-level capabilities. For instance, Persad and colleagues (Persad et al., 2007) propose a Capability-Demand Theory to evaluate product accessibility, where sensory, motor, and cognitive dimensions of user capability are assessed to estimate the number of people po- tentially excluded from using a given product (demands). For instance, a remote controller with small labels (visual demand) may exclude a high percentage of the population from using it (e.g. older adults and visually impaired). By quantifying the visual abilities of the target audience, we can predict how many people will not be able to use the product. This highlights the variability of users as well as the need to build new solutions to accommo- date a wide range of capabilities. Still, this approach is often static; that is, it assumes that the product is immutable. Thus, accommodations to technology require the development of completely new solutions or add-on’s that lowers the device’s demands.
In 1995, Newel proposed the concept of Extra-Ordinary Human-Computer Interaction (Newell, 1995), recognizing that all users have abilities, and some users have extra-ordinary abilities. For the first time, this drew the parallel between “ordinary” people operating in an “extraordinary” environment (e.g. adverse noise and lightning conditions, high work load) and “extra-ordinary” (disabled) users operating in an ordinary environment.
The author acknowledges, but does not quantify, that context can temporarily reduce people’s abilities in ways similar to health-related impairments. Similarly to Sears and colleagues (Sears et al., 2003) definition of context and situationally-induced impairments and disabilities, Extra-Ordinary Human-Computer Interaction was the first attempt to relate human abilities to context.
Around the same time, the concept of User Interfaces for All (UI4All) was proposed (Stephanidis, 1995) as a vehicle to efficiently and effectively address the numerous and diverse problems related to the accessibility of interactive applications in different contexts of use. Particularly, UI4All promotes the use ofunified user interfaces (Stephanidis, 2001) to support user independent interface development. In a unified user interface, only the core functionality is developed, while abstract user interface representations map to one concrete interface template, either at configuration- or at run-time. Special purpose user interface software components automatically manage the specific issues and adapt the interface to a particular user or user group (Harper, 2007). Nonetheless, this approach lacks the means to measure and develop interfaces for these users, placing the burden on the developer.
Universal Design (Vanderheiden, 1998) and Design for All (Stephanidis and Salvendy, 1998) are very similar concepts that emerged in Europe and introduced the visionary goal of an Information Society for all. These approaches focus on applying a set of principles, methods, and tools to develop technological products and services that are accessible and usable by all citizens, therefore avoiding the need for adaptations or specialized design.
These approaches emerged as a response to the limitations of add-on approaches like as-
sistive technologies. Universal design advocates a “one size fits all” solutions, which in our view, may suit door handles or building entrances, but comes short on many technological domains.
After Universal Design and Design for All, emerged the concept of Universal Usability.
Similarly, this approach provides guidelines for designing interfaces that are usable by the widest range of people that is possible (Harrin, 2008)(Shneiderman, 2000)(Vanderheiden, 2000). It attempts to answer questions, such as “what can everyone do?” . As universal design and design for all approaches, universal usability also focuses on a “one size fits all” ideal. However, it does not specifically aim at people with disabilities, and is more concerned with the gap in access due to other aspects, such as gender, literacy, economic status, culture, etc.
User-Sensitive Inclusive Design came primarily from the UK and was proposed by Newell and Gregor in 2000 (Newell and Gregor, 2000). It acknowledges that “design for all” is a very difficult, if not even impossible task. The authors state that:
“Providing access to people with certain types of disability can make the prod- uct significantly more difficult to use by people without disabilities, and often impossible to use by people with a different type of disability”.
Inclusive design emerges from the premise that disabled users have a much greater variety of characteristics than able-bodied people and it is usually difficult to find “representative”
user groups. The use of the term “inclusive” rather than “universal” reflects the view that
“inclusivity” is a more achievable, and in many situations, appropriate goal than “univer- sal design” or “design for all”. Also, the authors suggest that some significant differences must be introduced into the User Centered Design Paradigm. Therefore, “sensitive” re- places “centered” to underline the extra levels of difficulty involved when the range of functionality and characteristics of the user groups can be so great that it is impossible to produce a representative sample of the user group.
More recently, Wobbrock et al. (Wobbrock et al., 2011) proposed a new approach called Ability-Based Design, which consists of focusing on ability throughout the design process in an effort to create systems that leverage the full range of human potential. As user- centered design shifted the focus of interactive system design from systems to users, ability- based design attempts to shift the focus of accessible design from disability to ability. This concept provides a unified view of able-bodied and disabled users, as well as health-related impairments and situationally-induced impairments. The authors focus on how systems can be made to fit the abilities of whoever uses them, which is closely related with the notion of adaptation.
2.2.1. Challenges on Accessible Computing
Multiple approaches, same goal, different avenues. There are numerous approaches to accessible computing and they all share the common goal of providing access to tech- nological solutions, thus improving quality of life and guaranteeing equality of rights in an increasingly digital world. How to achieve this goal, however, may vary between ap- proaches. For instance, universal-related (Vanderheiden, 1998)(Stephanidis and Salvendy, 1998)(Harrin, 2008)(Shneiderman, 2000) approaches focus on lowering solutions demands so they can be used by everyone (Guerreiro et al., 2009), while others center their attention on individual needs (Newell and Gregor, 2000). Similarly, while some focus on measuring and modeling low-level human abilities (Persad et al., 2007) recurring to clinical assess- ment, others use human performance in computer-oriented tasks to accommodate users’
needs (Wobbrock et al., 2011). The real challenge consists in knowing which approach is more appropriate to solve the problem at hand.
Universal interfaces, a utopian goal? Building universal interfaces that can be used by everyone is the “saint graal” of accessible computing. A world where computerized devices can be equally used by all people, independently of their motor, sensory and cognitive abilities is certainly a utopian world. Although “universal” approaches seem to believe in this world, much due to the inherited design principles that grew out of architecture (Mace et al., 1990)(Steinfeld, 1994)(Story, 1998), this is nearly impossible to achieve when considering computer-oriented interfaces. The founders of universal design were mainly concerned with physical spaces and while this may be a fruitful approach for architecture solutions, such as door handles and building ramps, designing “one size fits all” computer interfaces is unfeasible. The number of tasks and uses of current mobile devices, as well as the spectrum of users’ profiles is too wide to be accommodated in a single interface. Even if we think of brain-computer interfaces, and we could explicitly assess users’ intentions, cognitive abilities would still play a crucial role. Nonetheless, we believe that current interfaces can be greatly improved to cover wider audiences, and new solutions can be designed for those who are currently excluded from access to information.
A unified view of users’ abilities. In order to provide interfaces that can be used by a wide range of users, designers and researchers should focus on users’ abilities, rather than their dis-abilities (Wobbrock et al., 2011). Accessible computer interfaces need to be designed for “what a person can do”, instead of “what disabilities does a person have?”.
Abilities vary across a limited range and are influenced by the person’s characteristics as well as the surrounding context (Newell, 1995). In our research work, we adopted this unified view of users’ abilities, dealing with the variability of health-related conditions (Chapter 5) and contextual factors (Chapter 4). Moreover, this approach enabled us to place both user groups on the same “ability scale” (Chapter 6) and design common solutions (Chapter 7).
A common need: assessing user’s abilities. A common need of any accessible com-
puting approach is the assessment of users’ abilities. There are several ways of doing this:
assessing demographic characteristics, such as age or height; performing clinical functional assessment (Cook and Hussey, 2001)(Oliveira et al., 2011a); or characterizing disabilities through subjective scales (WHO, 2009). While these approaches may be useful to have an overall view of a person’s profile, it is usually very difficult to relate and generalize them to computer-skills. An alternative, which was followed in the remainder of this dissertation, was to detect user’s abilities as they interacted with technology (Gajos and Weld, 2004) and, therefore, inform user interface at design-time. Another topic of interest is in when to perform this assessment. A “battery of tests” can be administered once, assuming that users’ abilities will not change over time (or very little), or periodically for users whose abilities are constantly changing. A more difficult solution would be to test it “in the wild”, without the need of assistance (Hurst et al., 2008) or previous knowledge about the task (Trewin et al., 1997). In this case a new set of challenges emerge, since we need to infer about users’ intentions in order to measure their performance.
Improve solutions by understanding the context. In an increasingly mobile world, abilities are influenced by the context in which they are exercised (Gregor et al., 2002).
Sensing context and leveraging collected data to improve users’ performance with current technologies is still in its early stages. First, we need to understand how factors, such as light (Barnard et al., 2007), noise, or mobility (Schildbach and Rukzio, 2010)(Nicolau and Jorge, 2012c) influence users’ abilities in order to remove the experienced difficulties.
Some works in activity recognition hold great promise to achieve this goal (Choudhury et al., 2008)(Hinckley et al., 2000). Recent works have attempted to use contextual data to provide better suited interfaces (Kane et al., 2008b)(Goel et al., 2012). In our work, we went a little further and used context information (motion data) as a unifying measure between users with HIID and SIID (Chapter 7).
Modeling, modeling, modeling. After performance has been accurately measured and/or context has been sensed, there still remains the challenge of describing (model) users’ abilities. This step is particularly important if adaptation is to take place. De- scribing abilities in terms of users’ performance is an open research challenge, especially considering the variability of users with impairments, even with the same health condi- tion. When considering both HIID and SIID, we expect this variability to increase. In Chapter 7, we propose several features and models that attempt to explain users’ perfor- mance.
Apply all the knowledge. Once users’ abilities have been measured and modeled through performance and/or context, we need to incorporate all that knowledge into the user interface. One option is to give full control to end-users enabling them to customize the interface to their needs and preferences (Koester et al., 2007a). However, users need to be aware of this feature. Still, there is no guarantee that they will, or more importantly, that they can do that, ending up with an interface that does not accommodate their abilities. An alternative is to automatically adapt the user interface, requiring little or
no effort on the part of the user. Nevertheless, this approach also has a set of challenges of its own (H¨o¨ok, 2000): lack of control, predictability, transparency, obtrusiveness, and privacy. Further research, should explore new ways of dealing with these issues, and if possible, combining the advantages of adaptive and adaptable interfaces.