Contributions of Source-Constrained Search and Late Monitoring to Recall Accuracy

Abstract

This thesis aimed to investigate in detail the processes underlying the control of recall accuracy. It is believed that recall accuracy control comprises processes occurring at memory search and post-retrieval. There is a large body of research concerning post-retrieval monitoring processes; however, less is known about how memory search is constrained and what factors affect this process.Chapter 2 developed and tested a new paradigm for measuring the accuracy of constrained search and monitoring processes simultaneously. Participants were able to selectively retrieve one of two lists irrespective of which list was the target list, indicating successful constrained search. A key role for context reinstatement in constraining search was established through source clustering. Participants were also highly accurate in monitoring the correctness of retrieved items. Chapter 3 required participants to constrain search to one of two sources within a single list of items (Mixed-lists). Participants were able to do this, although search accuracy was poorer than for List membership (Chapter 2). Despite this, monitoring accuracy did not differ between List membership and Mixed-lists. Only source monitoring was sensitive to source manipulations within a single experiment. Chapter 4 presented an alternative method of measuring constrained search, which relies on curve fitting of recall latencies to estimate the size of a participant’s search set. This method successfully reproduced the findings from Experiments 2.3 (List membership) and 3.2 (Mixed-lists). Unfortunately due to poor curve fits, attempts to replicate findings from Experiment 3.1 were unsuccessful. Chapter 5 presents a retrieval model which attempts to replicate the typical pattern of declining search accuracy as the recall period progresses, with the assumption that participants’ ability to search for targets does not decline throughout the recall period. This model was able to produce accuracy curves which fit the search accuracy data fairly well; however, predictions for other recall metrics were poor. On the whole, this research demonstrates that in order to constrain search, participants must reinstate the context of the target source at retrieval, and that the success of constrained search is dependent upon the type of context or source which is encoded. Source monitoring data were generally in line with the predictions of the Source Monitoring Framework (Johnson et al., 1993).