Abstract
The role of the word predictability from sentence context for reality monitoring and external source monitoring was examined in two experiments. In a reality-monitoring task, discrimination of an internal source was better in the hard than in the easy condition. It is probable that extra cognitive operations engaged during word generation in the hard condition were effective cues for reality-monitoring judgements. In contrast, in an external source-monitoring task (recognition memory of item's colour), the hard condition resulted in worse source memory for generated words than did the easy condition. This result is consistent both with an item–context trade-off hypothesis and a processing hypothesis. Greater effort involved at the time of generation might limit resources available for encoding of an external source. It is also possible that for generated words, the hard condition promoted conceptual processing instead of perceptual processing; therefore the item's colour was not effectively encoded.
Acknowledgments
A portion of the data from the paper was presented at Cognitive Psychology Colloquium, Chair of Psychology III, Mannheim, 29 October 2009. My thanks go to Edgar Erdfelder, Sandra Mattern, and other participants of this meeting for their helpful comments and suggestions on my research. The author is also grateful to Neil Mulligan, Jeroen Raaijmakers, David Riefer, and an anonymous reviewer for their critical comments and suggestions on an earlier draft of the paper.
Notes
1 The number of letters may potentially influence generation difficulty, which may raise some doubts about accuracy of results. In order to explore this possibility, additional multinomial processing tree (MPT) model analyses were conducted when the number of letters in words used in hard and easy conditions was equated. This balance was achieved by excluding 6 shortest words from the easy condition and 6 longest words from the hard condition (overall 216 trials were excluded from the original data set). The result of main interest in this experiment—that is, better source memory for generated items in the hard condition than in the easy condition—was confirmed on this reduced data set; however, it showed only a trend toward significance (p = .07), probably due to the decrease in statistical power.
2 In retrospect, the way the sources were defined at test (i.e., “intact” or “with blanks”) could have directed participants' attention to the form of presentation instead of cognitive operations, which was not intended in this experiment. However, all items that were “with blanks” were generated, and all “intact” items were read at study, so when answering the question of how the target item looked, participants could anyway refer to their records of cognitive operations (see for comparison a study by Mattern, Citation2010, who found no effect of the kind of used instruction in a source memory task). In future research, it would be interesting to investigate a case when an external source is systematically related to a particular kind of cognitive operation.