Augmenting Autobiographical Memory: An Approach Based on Cognitive Psychology

This thesis investigates how an interactive software system can support a person in remembering their past experiences and information related to these experiences. It proposes design recommendations for augmented autobiographical memory systems derived from Cognitive Psychology research into human memory – a perspective missing from prior work. Based on these recommendations, a conceptual design of an augmented autobiographical memory system is developed that aims to support users in retrieving cues and factual information related to experiences as well as in reconstructing those experiences. The retrieval aspects of this design are operationalised in an interactive software system called the Digital Parrot. Three important factors in the design and implementation are the context of an experience, semantic information about items in the system and associations between items. Two user studies evaluated the design and implementation of the Digital Parrot. The first study focused on the system's usability. It showed that the participants could use the Digital Parrot to accurately answer questions about an example memory data set and revealed a number of usability issues in the Digital Parrot's user interface. The second study embodied a novel approach to evaluating systems of this type and tested how an improved version of the Digital Parrot supported the participants in remembering experiences after an extended time period of two years. The study found that the Digital Parrot allowed the participants to answer questions about their own past experiences more completely and more correctly than unaided memory and that it allowed them to answer questions for which the participants' established strategies to counteract memory failures were likely to be unsuccessful. In the studies, associations between items were the most helpful factor for accessing memory-related information. The inclusion of semantic information was found to be promising especially in combination with textual search. Context was used to access information by the participants in both studies less often than expected, which suggests the need for further research. Identifying how to appropriately augment autobiographical memory is an important goal given the increasing volume of information to which users are exposed. This thesis contributes to achievement of this goal by stating the problem in Cognitive Psychology terms and by making design recommendations for augmented autobiographical memory systems. The recommendations are confirmed by the design and implementation of such a system and by empirical evaluations using an evaluation method appropriate for the field