The demographic change leads to a considerable increase in user diversity. On the one hand, ‘digital natives’ grow up with ever changing interactive systems and are able to easily adjust to new interfaces. On the other hand, older users form a highly heterogeneous group including a broad user spectrum between ‘digital immigrants’ which are reluctant or unable to embrace new devices and highly motivated users with a high willingness learning to interact with modern technical systems. Besides of the motivation problem there is also a considerable diversity in sensory, cognitive, and motor skills in older users. Thus, a single interface cannot support the different requirements and abilities of all users. Adaptation has been recognized as a solution for supporting such requirements that are drifting apart. Adaptation may open up an interactive system to a larger user population and also allows a barrier-free interaction design.
Recognition and representation of user characteristics. A central topic of adaptive systems is the recognition and representation of user characteristics within the system. This representation is usually realized by a user model component which performs ’learning, inference, and decision making’ (Oppermann, 1994) and stores the outcome of these computations into the user model. The contents of the user model depend on the specific requirements of an adaptive interactive system. This model comprises information such as user preferences, an interaction history (based on information from the interaction model), goals, or predictions. Typical tasks of a user modeling component, for example, are making assumptions about user characteristics, performing classification of users, recording of user behavior, and drawing additional inferences (cf. Kobsa 2001, 2004). User models usually consist of different data types, like simple flags or response scales of questionnaires, up to complex data structures for sophisticated user modeling algorithms. These algorithms, such as neural networks, Bayesian networks or Markov chains (e.g. Witten et al., 2011), derive new information either from the data stored in the user model or from online recorded data. Which data is included in the model depends on the purpose of the application. It can include personal information such as users’ names and ages, their interests, their skills and knowledge, their goals and plans, their preferences and their dislikes or data about their behavior and their interactions with the system.
Integrating computer science and cognitive psychology. Our goal of user modeling is to extent common concepts from computer science by psychological knowledge. Thus, we are employing theoretical models from the field of cognitive and physiological psychology to infer about the characteristics and the actual state of younger and older users during the interaction with technical systems. For example in the field of driving, we adopted a kinematic model of Plamondon (e.g. Plamandon et al., 2003) to assess variations in central and neuromuscular parameters of younger and older drivers during the performance of lane change tasks (Rinkenauer & Hofmann, 2011). In our project ‘GripAssist’, we are applying models from the field of movement science to analyze sequential movements (Wing & Kristofferson, 1973, Wing et al., 1998, Bremer & Rinkenauer, 2013) as well as sensory fusion processes in grip coordination (Rinkenauer et al., 1999). The goal of using psychological concepts is to optimize surrogate sensory feedback for an advanced interaction with orthoses and prostheses. In our recent project IndiRes, we are employing models from movement coordination (Marteniuk et al., 1984, Schmidt et al., 1979, Rinkenauer, 2000) to track posture and movements of the hand, head, and torso to infer about changes in fatigue and motivation in the context of office work. The advantage of integrating theoretical concepts from the field psychology is that those concepts allow to infer and interpret user behavior more directly as, for example, data mining algorithms. In our future research we are going to investigate to what extent the combination of machine learning algorithms and psychological concepts can be used to control the individual adaptation of interfaces and technical environments more effectively.
Regional and supraregional networks and cooperations
As for the topic of information processing, age-related aspects of human machine interaction will be considered also for older adults.