The fan effect influences face recognition but does not moderate the own-age bias

ABSTRACT

The fan effect shows that memory is superior for information associated with few contexts relative to information associated with many contexts. The current study examined the fan effect within face recognition. Participants studied faces in which eye regions were associated with only one face (low-fan) or several faces (high-fan). Additionally, we examined whether featural fan might moderate in-group biases (i.e. in-group faces are better remembered than out-group faces). To this end, we manipulated occupation status (Experiment 1) and age (Experiment 2) of studied faces, and presented high- and low-fan faces. Results showed that low-fan faces were better remembered than high-fan faces. We did not detect an in-group bias as a function of occupation status, but there was a robust own-age bias. Fan type did not moderate the own-age bias, however. Although face recognition is sensitive to featural fan, the effect does not appear to differentially impact the own-age bias..

KEYWORDS:

• Relative distinctiveness
• own-age bias
• face recognition
• fan effect
• memory

Acknowledgements

We thank Daniel Dickison and Almas Kebekbayev for their programming help with stimuli generation.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 Only females were tested in both experiments in order not to introduce a potential own-gender recognition bias.

2 Participants were screened for age and gender to match the age and gender of the people depicted in the stimulus pictures, but they were not screened for race. The purpose was to determine if an ad-hoc grouping by same- or other-race might reveal any further moderating effects. An exit survey revealed that 37 participants were of the same race as the people depicted in the stimulus pictures whereas 28 were of a different race.

3 All repeated measures effect sizes reflect Cohen’s d (see Lakens, 2013, for details of the calculation.)

4 We note that, although different numbers of items in the two conditions were not ideal, the discrepancy was the same for both the “old” and “new” items. Given that there were different numbers of high- and low- fan faces, there was a possibility that the low-fan faces might have been given a memorial boost because there were less of them. Therefore we addressed this issue by making the numbers of high- and low- fan faces equivalent in Experiment 2.

5 To test for any moderating effects of participant race we initially conducted a 2 (Fan Type: low vs high) × 2 (Occupation Status: high vs. low) × 2 (Race: same vs. other) repeated-measures ANOVA. Since there was no effect of race, F < 1, and no interactions between race and fan type or occupation status, F(1, 63) = 1.6, p > .10, and F < 1, respectively, we removed race from the model. The possible effect of race was also tested in Experiment 2, but once again neither the main effect nor interactions were significant, ps > .10, so race was again removed from the model.

6 For the sake of transparency, we felt it important to publish the null results for occupation status. In that way, the data allow for a more accurate effect size for that particular in-group/out-group manipulation. Indeed, psychology has a “replication crisis” (Lilienfeld, 2017). In addition, we used a student population similar to the student population that was used in the original Ratcliff et al. (2011) paper and at the time we had no reason to suspect we might not replicate.

7 The electrophysiological results have yet to be reported.

Additional information

Funding

Participant payment was funded by the Doctoral Student Research Award of American University’s College of Arts and Sciences to Joyce Oates.