Age-related accessibility biases in pass-face recognition

Abstract

Accessibility and security are often depicted as conflicting aspirations. Accessible systems may be less secure and secure systems may be less accessible. The search is on for greater security for logging onto systems, whilst achieving acceptably high levels of accessibility. Pass-faces are based on the twin axioms of greater accessibility and security. A new user of a pass face system is asked to select "n" faces from an array of faces, where n is at least two and usually more. The user is required to memorize those faces and to recognize them again when represented to you as part of larger display. It has been suggested that this approach is less susceptible to poaching than are alphanumeric methods. There has been a considerable volume of work to evaluate the usage of pass face systems, but little work on the psychology of pass faces. Equally, pass face systems have received little attention from researchers in accessibility. In the present study, two previously unrelated themes were investigated in two experiments. First, are pass face systems acceptably usable? Second, how do pass face systems rely on the reliability of human face recognition memory? In two experiments, two types of pass face system consisting of (a) older faces; over 50 years of age and (b) younger faces; under 30 years of age were created. It turns out that younger participants are often better at recognizing younger faces than older faces in the context of pass face security, whilst older participants are sometimes better at recognizing older faces than younger faces in the context of pass face security. Thus an experiment that used only younger faces would falsely conclude that younger participants were better at face recognition memory than older participants in general. These results were checked and confirmed by literatures reviews of pass face security and human recognition memory for faces. These results show that universal access cannot be applied on a one-size-fits-all basis. They also suggest that the security-related disciplines of HCI and psychology would benefit from greater interaction between them. © 2011 Springer-Verlag.