Eva Hudlicka & John Billingsley
Psychometrix Associates, Inc.
Lincoln, MA, USA
As decision-support systems (DSS) and user interface (UI) technologies mature and proliferate into critical applications, and increasingly heterogeneous user populations, it becomes particularly important that they accommodate individual user characteristics. Recent research provides increasing evidence that individual differences in general, and affective states in particular, have a major impact on performance ( Williams et al. 1997; Eysenck 1997; Mineka & Sutton 1992; Isen 1993; LeDoux 1992; Fallesen 1993). Specific influences include the following: 1) anxiety influences attention, by narrowing of attentional focus and predisposing towards detection of threatening stimuli ( Mineka and Sutton 1992; Eysenck 1997); 2) mood influences memory, by biasing recall of information that matches the current affective state (mood congruent recall) ( Bower 1981); and 3) obsessiveness influences decision-making by increasing "checking behaviors" and delaying decision making ( Williams et al 1997). Current lack of accommodation of individual variations in performance in most human-machine systems can lead to non-optimal behavior at best, and critical errors with disastrous consequences at worst (e.g., USS Vincennes).
To address this problem, we developed an affective, adaptive methodology and implemented it within a software prototype: the Affect and Belief Adaptive Interface System (ABAIS). The methodology consists of: 1) sensing/inferring the individual's affective state and performance-relevant beliefs (e.g., high level of anxiety; aircraft under attack); 2) identifying their potential impact on performance (e.g., focus on threatening stimuli, biasing perception towards identification of ambiguous stimuli as threats); 3) selecting a compensatory strategy (e.g., redirecting focus to other salient cues, presentation of additional information to reduce ambiguity); and 4) implementing this strategy in terms of