Effects of self-referencing on feeling-of-knowing accuracy and recollective experience

Abstract

The current research investigated the impact of self-referencing (SR) on feeling-of-knowing (FOK) judgements to improve our understanding of the mechanisms underlying these metamemory judgements and specifically test the relationship between recollective experiences and FOK accuracy within the accessibility framework FOK judgements are thought to be by-products of the retrieval process and are therefore closely related to memory performance. Because relating information to one's self is one of the factors enhancing memory performance, we investigated the effect of self-related encoding on FOK accuracy and recollective experience. We compared performance on this condition to a separate deep processing condition in which participants reported the frequency of occurrence of pairs of words. Participants encoded pairs of words incidentally, and following a delay interval, they attempted at retrieving each target prompted by its cue. Then, they were re-presented with all cues and asked to provide FOK ratings regarding their likelihood of recognising the targets amongst distractors. Finally, they were given a surprise recognition task in which following each response they identified whether the response was remembered, known or just guessed. Our results showed that only SR at encoding resulted in better memory, higher FOK accuracy and increased recollective experience.

Keywords:

• Self-referencing
• Feeling-of-knowing
• Accessibility
• Recollection
• Remember-know (R–K) judgements

Aysecan Boduroglu was supported by a TUBA-GEBIP grant. We would like to thank undergraduate research assistants in the Cognitive Processes Laboratory for help during data collection.

Aysecan Boduroglu was supported by a TUBA-GEBIP grant. We would like to thank undergraduate research assistants in the Cognitive Processes Laboratory for help during data collection.

Notes

¹ Hamami et al. (2011) argued that the absence of a significant difference between self and close other condition is due to the mnemonic benefit associated with intimacy. This explanation is consistent with the results from a previous meta-analysis (Symons & Johnson, 1997) showing that SRE is reduced when the target is encoded with reference to a highly familiar other.

² A_z is a highly recommended nonparametric single point measure of memory/metamemory accuracy (Masson & Rotello, 2009). In FOK studies, to calculate FOK accuracy as A_z, FOK strength (binned) is plotted against recognition accuracy and an receiver operating characteristics (ROC) curve is fitted given these data. A_z is the area underneath the ROC curve. If participants do not equally distribute their responses (i.e., use the FOK scale evenly), then for certain bins there may be insufficient observations to yield a stable point. Thus, an ROC will not be created for those participants.

³ The comparison of these 10 and with the remaining 48 participants revealed that in both conditions, the excluded group recognised significantly more items (.79 ± .13 vs. .70 ± .10 and .74 ± .17 vs. .65 ± .13 for the SR and common conditions, respectively; both ps < .05). However, their FOK strength was significantly lower than the remaining sample (approximately 35 ± 30 vs. 56 ± 24, p < .05, collapsed across conditions). The fact that the excluded participants' FOK ratings were significantly lower than the rest of the participants for incorrectly recalled items across conditions (approximately 24 ± 26 vs. 45 ± 17, p < .01) but not different for correctly recalled items suggests that the excluded participants were more able to adjust their FOK ratings by correctly monitoring their recall performance compared to the remaining participants.

⁴ For purposes of comparing our results with previously reported effects, we also calculated FOK accuracy as Gamma. The effect of SR at encoding on FOK accuracy over the common condition remained significant (t(58) = 2.39, p < .05). Gamma descriptors are presented in the tables for purposes of full disclosure.

⁵ As requested by an anonymous reviewer, we carried the same study using the “traditional” measure of FOK. All materials were identical to Experiment 1, except that participants reported FOK judgements only following omissions. We collected data from 36 participants and were able to calculate A_z scores for 23 of them; data from one participant was removed from analysis because his score was more than 2 standard deviations below the group mean. Participants whose A_z could not be calculated were similar in profile with those omitted from analyses in Experiment 1 (see footnote 3). The full set of comparisons is presented in the Appendix as Table A1. For the remaining 23 participants, a comparison of performance across the self-reference and the common condition revealed significant benefits of self-reference for memory (all t's > 2.28, p's < .03 and all d's > .21). Consistent with Experiment 1 (see performance for incorrect recognitions), there was no effect of SR on FOK strength, t < 1. Critically, there was a marginal effect of condition on FOK accuracy with accuracy higher in the self-reference as opposed to the common condition, t(22) = 1.82, p = .08, d = .25). When FOK accuracy was calculated by gamma, and the two conditions compared, there was a significant difference in the expected direction, t(35) = 2.05, p = .048, d = .52. The slight difference in the data for the two indices may be due to the reduced power when analyses are carried out with A_z (also see footnote 2). The impact of SR on recollective experience was similar to Experiment 1, with significantly higher “remember” responses reported in the self-reference as opposed to the common condition.

⁶ We would like to thank an anonymous reviewer for outlining this possibility.