Face masks degrade our ability to remember face-name associations more than predicted by judgments of learning

ABSTRACT

In response to the COVID-19 pandemic, face masks became required attire. Face masks obstruct the bottom portion of faces, restricting face processing. The present study examined the influence face masks have on memory predictions and memory performance for new face-name associations. Participants studied face-name pairs presented for 8 s (Experiment 1) or 10 s (Experiment 2). Half of the face-name pairs included a face mask obstructing the nose and mouth of the pictured face, counterbalanced across participants. Participants provided item-by-item judgements of learning (JOLs) and completed subsequent cued recall and associative recognition memory tests. Both experiments demonstrated that face masks impaired memory for newly-learned names, however, the magnitude of the mask impact was under-predicted by JOLs. The presence of a face mask negatively influenced memory performance to a greater degree than participants’ JOLs predicted. Results have implications for name learning during pandemics, as well as in settings where face masks are common (e.g., medical field).

KEYWORDS:

• Face mask
• face processing
• face-name associations
• memory
• memory predictions

Open Scholarship

This article has earned the Center for Open Science badge for Open Data. The data are openly accessible at https://doi.org/10.48497/CZP9-MX51.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are publicly available on the University of Tampa’s Institutional Repository: https://doi.org/10.48497/CZP9-MX51.

Notes

1 Mask condition was split within each gender-ethnicity subtype (e.g., of the 4 white males, 2 were presented with a mask and 2 were presented without a mask during encoding).

2 On the associative recognition test, a correct response includes responding “yes” to a studied face-name pair and responding “no” to a face that was paired with an incorrect name. Participants received instructions to “Click yes if that was the person’s name you studied” and “Click no if you studied a different name with that person’s face”.

3 A 2 (JOL1 vs. JOL2) × 2 (No Mask vs. Mask) repeated measures ANOVA confirmed JOLs were significantly higher for JOL1 than JOL2, p = .026, and that JOL predictions were significantly lower for the mask condition, p < .001. There was a significant interaction, p = .012.

4 Gamma correlations are a common index of metamemory resolution (e.g., see Rhodes, 2016; Zhao et al., 2023; Undorf et al., 2022; Giffard et al., 2020), but have important limitations (e.g., Bröder & Undorf, 2019; Higham & Higham, 2019).

5 Note that we opted to calculate gamma correlations using participant’s memory predictions during the first study phase (JOL1) because this corresponds most closely to the majority of other reported gamma correlations in the literature with only one study phase. Further, we would expect that gamma correlations would only get stronger for JOL2, as Tauber and Rhodes (2012) demonstrated that repeated study phases resulted in stronger gamma correlations.

6 Note: 3 participants skipped this question.

7 Due to the small variability in responses, no further analyses were conducted for these questions.

8 A 2 (JOL1 vs. JOL2) × 2 (No Mask vs. Mask) repeated measures ANOVA confirmed JOLs were significantly higher for JOL1 than JOL2, p = .005, and that JOL predictions were significantly lower for the mask condition, p < .001. There was no significant interaction, p = .838.

9 Note: 5 participants skipped this question.

10 Due to the small variability of responses, no further analyses were conducted for these questions.

11 One participant did not provide aggregate predictions and was excluded from the analyses.

Additional information

Funding

This work was supported by The University of Tampa.