Article révisé par les pairs
Résumé : The ability to perform two tasks simultaneously has become increasingly important as attention-demanding technologies have become more common in daily life. This type of attentional resources allocation is commonly called "divided attention". Because of the importance of divided attention in natural world settings, substantial efforts have been made recently so as to promote an integrated, realistic assessment of functional abilities in dual-task paradigms. In this context, virtual reality methods appear to be a good solution. However to date, there has been little discussion on validity of such methods. Here, we offer a comparative review of conventional tools used to assess divided attention and of the first virtual reality studies (mostly from the field of road and pedestrian safety). The ecological character of virtual environments leads to a better understanding of the influence of dualtask settings and also makes it possible to clarify issues such as the utility of hands-free phones. After discussing the theoretical and clinical contributions of these studies, we discuss the limits of virtual reality assessment, focusing in particular: (i) on the challenges associated with lack of familiarity with new technological devices; (ii) on the validity of the ecological character of virtual environments; and (iii) on the question of whether the results obtained in a specific context can be generalized to all dual-task situations typical of daily life. To overcome the limitations associated with virtual reality, we propose: (i) to include a standardized familiarization phase in assessment protocols so as to limit the interference caused by the use of new technologies; (ii) to systematically compare virtual reality performance with conventional tests or real-life tests; and (iii) to design dual-task scenarios that are independent from the patient's expertise on one of the two tasks. We conclude that virtual reality appears to constitute a useful tool when used in combination with more conventional tests.