Ray G. Adams

Cognitive and learning difficulties and how they affect access to IT systems

In October 2005, the IBM Human Ability and Accessibility Center and T.J. Watson Research Center hosted a symposium on “cognitive and learning difficulties and how they affect access to IT systems”. The central premise of the symposium was the recognition that cognitive and learning difficulties have a profound impact on a person’s ability to interact with information technology (IT) systems, but that little support is currently being offered by those systems. By bringing together internationally renowned experts from a variety of different, but complementary, research fields, the symposium aimed to provide a complete overview of the issues related to this topic. This paper summarises the discussions and findings of the symposium.

A systematic basis for developing cognitive assessment methods for assistive technology

We address the development of new assessment techniques for people of an employable age with acquired physical disabilities, such as head injuries and musculo-skeletal injuries. Specifically, we focus on cognitive function, which is fundamental to Assistive Technology and Rehabilitation engineering.

From Cultural to Individual Adaptive End-User Interfaces: Helping People with Special Needs

Culture heavily influences human---computer interaction (HCI) since the end-user is always within a cultural context. Cultural and informational factors correlate to jointly influence the look and feel of interactive systems, e.g. widget position or information density. Every single individual also develops a specific culture (eating style, walking style etc.), i.e. characteristics and behavior as well as attitudes and values. Consequently, individual adaptability can be essential to cover individual needs of the culturally but uniquely imprinted end-users with special needs e.g. reducing the workload by recognizing and knowing the individual expectances of the end-user. This improves usability and leads to shorter training and improves universal access.

Universal access to assistive technology through client-centred cognitive assessment

As a basis for user needs and system design assessments in assistive technology (AT), we have developed a new conceptual framework and battery of tests and research paradigms, on a continuing improvement basis. The framework consists of three levels, to provide; overall guidance, specific models for understanding data and task analysis. Following earlier work [1], we report three case studies, considering attentional problems and user-specific needs, to validate our core test elements and conceptual framework. Our systematic method generated specific benefits for our users and pointed out the need for cognitive software in assistive technology.

User sensitive research in e-learning: exploring the role of individual user characteristics

The increasing need for active and accessible learning in the inclusive knowledge society drives the demand for e-learning that engages users much more effectively than ever before. In this context, it is crucial to conduct research that embraces innovation in user sensitive design, or else influential individual user differences may be overlooked. The objective of this paper is to explore the creation of successful e-learning systems that are able to increase users’ learning performance and enhance their personal learning experiences. The paper reports two converging and complimentary approaches, namely case studies and experimentation. First, case studies are used to explore the extent to which effective e-learning systems comply with eight specific factors. Of the eight, accessibility, individual differences and student modeling turn out to be the weakest points in current practice. Second, an empirical study investigates the influences of user individual user differences on users’ learning outcomes in an e-learning environment. The experiment found that individual differences in motivation to learn and expectations about e-learning significantly impacted users’ learning achievements. Third, based on these studies, improvements in research methodology are identified towards greater consideration of user sensitive research issues, thus enabling us to outline improved experimental procedures. Further experiment results should provide us with better insights into the arguments needed to carefully assess benefits of developing and involving a user model in an e-learning application. Consequently, evaluation and justification could now encompass both system performance as well as user performance.

Universal access through client-centred cognitive assessment and personality profiling.

The demand for universal access to information in the evolving Information Society produces an inexorable move towards complex, powerful and interlinked technological solutions. In this context, user requirements must be captured by more powerful user models, based upon more advanced user centred methods. Traditional HCI techniques may not work well in the new context of future and emerging technologies. Earlier work [1] observed significant dissociations between observed task performance and self report, raising profound and serious problems for user modelling methods. This empirical paper evaluates three different types of method used in user modelling; task performance, self-report and the personality inventory. Four case studies with individuals with acquired disabilities are reported here. The relationships between these three aspects of the user’s profile (self report, task performance and the personality inventory) are more complex than expected and provide different, sometimes contradictory, perspectives of user needs. A potential explanatory framework is offered briefly to guide future user modelling work. More importantly, any code of practice for Universal Access must not rely on any one method alone but must combine methods to minimise conceptual and practical errors. User profiles for adaptive technology must also employ multiple methods, if such technology is to be reliable in practice.

User modelling and social intelligence

There is a growing body of evidence that key components of human cognition can be used to identify important aspects of accessibility design for universal access in the information society, through user modelling. However, there is an equal growth in an appreciation of the contexts within which any interactive system must function, including the vocational and social contexts. If so, there is an important need is to extend cognitive user models to respond to and make predictions about the vocational and social contexts that make up the information society. Whilst many aspects of social intelligence can, it seems, be subsumed under current cognitive architectures of the user, there is the practical danger that the contribution of social intelligence may be underestimated when considered as a subset of the knowledge domains or skills sets of human cognition. To counter this practical development problem, the concept of the social intelligence interface is introduced as a developmental construct to inform the inclusive design process.

User modeling for intelligent interfaces in e-learning

The emergence of the accessible knowledge society for all underlines the need for all to acquire the necessary knowledge and skills for inclusion. One way to do so is through e-learning, which itself should be accessible to all intended users. The mere provision of knowledge systems is not enough, since the need is for active and accessible learning that engages the participants effectively. This study explores the hypotheses that the solution is to be found in systems that: (i) support rather than replace the tutor, (ii) are accessible to the intended populations of users, (iii) can be adapted to the needs and individual characteristics of diverse users, (iv) are smart enough to adapt in real time to changing user needs, (v) reflect best practice in learning psychology, (vi) provide a high quality student experience, (vii) provide a high quality tutor experience and (viii) use valid student modeling. These hypotheses are evaluated through the five case studies. Accessibility and student modeling turnout to be the weakest points. All the other features are well represented in the case studies. None placed any kind of substantial emphasis upon accessibility. Only one of the case studies (case study 2; Cognitive Tutor Authoring Tools) makes a substantial effort in user modeling, being based upon the generic, cognitive model ACT. Even then, the focus tends to be on the typical or average user and does not address the problems of user diversity. Clearly, accessibility and user modeling need much more work in e-learning.

Creating smart and accessible ubiquitous knowledge environments

Digital libraries offer substantial volumes of declarative knowledge to the information society. This paper explores the extent to which current and future digital libraries, also known as ubiquitous knowledge environments, can be made sufficiently usable, accessible and smart to support an inclusive information society and the aspiration of universal access. Using a range of converging methods to evaluate a random sample of such digital library websites, it is concluded that, whilst they act as substantial and functional repositories for knowledge, there is potential to improve, particularly in accessibility and smartness. The current methods are validated through the substantial statistical significance levels and by the meaningful patterns found in the resulting data. A new measure of system smartness is introduced and found to provide a useful metric for present purposes, though it is clear that further work will be needed.

Evaluating the Next Generation of Multimedia Software

Developing new multimedia, software applications is a particularly fascinating but challenging task, depending, as it does, on the thorough evaluation of new concepts and prototypes. As complexity and choice increases exponentially, future development will be increasingly challenging for emerging, intelligent, multimedia and virtual reality applications, including Web 3D. In the past, we have been able to rely on generic evaluative heuristics. But are they generalisable to the newer generations of software? In this study, participants saw demonstrations of two interactive, virtual reality, multimedia applications, namely Second Life and 3d mailbox. They used them to generate questionnaires to capture those key aspects of such applications that were important to them. They did not find it difficult to generate questions grounded in their own experience. The resulting items turned out to be validated by significant levels of internal consistency across dependent variables. Surprisingly, however, the new heuristics bore little resemblance to traditional or current, cognitive items. The overwhelming influence was that of the immersive impact of such applications rather than standard design issues or cognitive user factors. Clearly, new software innovations require equally new innovations in evaluation techniques in general and context specific heuristics in particular, but we do not yet have a conceptual foundation upon which to base them.