Further investigations into post-completion error : the effects of interruption position and duration

Further Investigations into Post-completion Error: the Effects of Interruption Position and Duration Simon Y. W. Li, Anna L. Cox, Ann Blandford, Paul Cairns, Richard M. Young and Aliza Abeles {simon.li, anna.cox, a.blandford, p.cairns, r.m.young, a.abeles}@ucl.ac.uk UCLIC, UCL Interaction Centre 31/32 Alfred Place, London WC1E 7DP UK Abstract Edwards & Gronlund, 1998), the role of retrieval cues after an interruption (e.g. Cutrell, Czerwinski & Horvitz, 2001), control over the interruption (McFarlane & Latorella, 2002), and preparation before engaging the interruption (Trafton, Altmann, Brock & Mintz, 2003). However, there has been no work looking at the effect of interruption on a specific kind of error, namely PCE. A particularly useful theoretical framework for the current study is Altmann & Trafton’s (2002) activation- based goal memory (AGM) model. The AGM model has its origin in explaining goal suspensions and resumptions in problem-solving and has been applied to investigating the disruptiveness of interruption on primary task performance and explaining the occurrence of PCE. Using the construct of activation, the AGM model suggests that, just like other memory elements in the cognitive system, goals have associated activation levels and cognition is directed by the most active goal retrieved at any given time. The amount of activation associated with a memory item is subject to decay, and this decay process is time-based and gradual. If the cognitive system needs to refocus attention to (or resume) an old goal then this old goal needs to undergo a priming process to become active again. The priming process is possible through associative links between retrieval cues and the to-be-resumed goal. The retrieval cues can be internal or external to the cognitive system. The AGM model suggests that task steps in a learned procedural skill can be viewed as a sequence of associative links, each action step acting as a retrieval cue for the next. This procedural cueing mechanism explains how PCE is usually avoided; hence, people usually manage to carry out PC tasks, such as photocopying, without committing the PCE most of the time. Consideration of the AGM model enables us to predict that different interruption positions during procedural tasks will result in different effects on the rate of PCE. An interruption occurring just before the post-completion (PC) step is more likely to lead to a PCE than an interruption occurring at any other point within a task structure. An interruption just before the PC step will have disrupted the associative priming from the preceding step, whereas other interruption points will still allow the pre-PC step to be carried out (after the correct goal is resumed successfully), and once it is carried out, it would cue the PC step. However, the gradual decay process of the AGM model suggests that the disruptiveness of an interruption also depends on the duration of the interruption. An interruption Two experiments were carried out to examine the effect of interruption position and interruption duration on post- completion error (PCE) occurrences in a game-like procedural task. Experiment 1 showed a significant main effect of interruption position on PCE rate; significantly more PCEs were obtained when the interruption occurred just before the PC step than interruptions at any other positions in the task. The same effect was also obtained for other non-PCEs suggesting that PCEs are no different to other non-PCEs in terms of the interruption position effect. Experiment 2 replicated the interruption position effect but did not show a reliable difference in PCE rates between a 45-sec and a 15-sec interruption. However, the trend of the differences in PCE rates between the two interruption durations is consistent with our initial prediction. The results of both experiments were explained in terms of the activation-based goal memory model (Altmann & Trafton, 2002). Introduction Post-completion error (PCE) is a specific kind of omission error, which occurs after the completion of a main task, for example, forgetting to collect your change after purchasing from a vending machine. Recent research has identified various factors that affect the error rate in routine procedural tasks such as working memory load (Byrne & Bovair, 1997) and dynamic visual cues (Lee, 1992; Chung & Byrne, 2004). It has also been shown that this robust error phenomenon occurs in non- routine problem-solving situations (Li, Blandford, Cairns & Young, 2005). Although PCE has received more attention recently, there are still many open questions about which factors provoke or mitigate the occurrence of the error. Given that interruptions are pervasive in most workplace environments and have been shown to lead to increased levels of overall error, their direct consequences in safety critical domains can be serious. Therefore, it seems a logical route to investigate what effect interruptions might have (or not have) on the occurrences of post-completion error. This study set out to investigate the effect of interruption position and duration on the rate of post-completion error and is motivated by the activation-based goal memory model (Altmann & Trafton, 2002). Background Several dimensions of interruptions have been investigated and are thought to affect performance on the primary task, e.g. the complexity of the interruption and its similarity to the primary task (Gillie & Broadbent, 1989;