Experiment 1: between-subjects manipulation of face masks

The mean perceived ages of masked and unmasked neutral and smiling faces are presented in Fig. 2. As can be seen in the figure, the results of the unmasked condition replicate our previous findings of an aging effect of smiling (AES) for female and male faces in the young adults, and for male faces in the middle-aged adults group. There was no effect of smiling on the age evaluations of the elderly group. The pattern of results for masked faces was similar, but now there was an indication for AES for middle-aged adult females as well.

Fig. 2
figure 2

The aging effect of smiling (AES) for masked and unmasked faces in Experiment 1. For unmasked faces, AES was found for male and for female faces in young adults, and for male faces in middle-aged adults. A similar pattern of results was found for masked faces, but now AES was also found for middle-aged female faces. Error bars represent standard errors of the mean

A mixed ANOVA design with the gender of the photographed person, expression, and age group as the within-subject independent variables and the presentation format (masked, unmasked) as a between-subjects independent variable was used to analyse the data, with perceived age (years) serving as the dependent variable. Preliminary analysis that included the gender of the participants did not show a main effect of gender or interactions with presentation format or expression and therefore, the participant’s gender was not included here or in further analyses.

Main effects were found for age group [F(1.33,109.27) = 3313.57, p < 0.001, ηp2 = 0.98] and for expression [F(1,82) = 81.67, p < 0.001, ηp2 = 0.49], indicating that, overall, smiling faces were perceived as older than neutral faces. The main effect of gender (of the photographed person) was significant with female faces perceived as younger than male faces [F(1,82) = 13.8, p < 0.001, ηp2 = 0.14]. More importantly, there was no effect of mask on perceived age, as indicated by a nonsignificant main effect of format [F(1,82) = 0.02, p > 0.05]. This result showed that face masks do not produce biases along perceived age across the different conditions.

A significant interaction between gender and expression [F(1,82) = 12.43, p < 0.001, ηp2 = 0.13] indicated larger AES for male compared to female faces (for similar results, see Ganel & Goodale, 2021). A significant interaction between age group and gender [F(2,164) = 99.79, p < 0.001, ηp2 = 0.55], indicated that the effect of gender was different in the three age groups. This interaction was qualified, however, by a three-way interaction with presentation format [F(2,164) = 6.42, p < 0.001, ηp2 = 0.07] and, as described below, was further explored using specific comparisons. A significant interaction was also found between age group and expression [F(2,164) = 34.39, p < 0.001, ηp2 = 0.29], indicating different effects of smiling on perceived age in the different age groups (Ganel & Goodale, 2021). Again, this interaction was qualified by a three-way interaction with format [F(2,164) = 3.78, p < 0.05, ηp2 = 0.04]. The two-way interactions between age group and format [F(2,164) < 1, p > 0.05], between gender and format [F(1,82) = 2.49, p > 0.05, ηp2 = 0.03], and between expression and format [F(1,82) = 1.07, p > 0.05, ηp2 = 0.01] were not significant. The interactions between gender, expression, and format [F(1,82) < 1, p > 0.05], between gender, expression, and age group [F(2,164) = 1.02, p > 0.05, ηp2 = 0.01], and the four-way interaction [F(2,164) = 1.31, p > 0.05, ηp2 = 0.01] were not significant as well. To better understand the pattern of results and to test for the presence of AES in the different conditions, we performed planned comparisons between smiling and neutral female and male faces within each age group.

The results of the unmasked condition replicated those of our recent study with unmasked faces (Ganel & Goodale, 2021). Planned comparisons showed an aging effect of smiling (AES) for female [F(1,82) = 23.72, p < 0.001, ηp2 = 0.22] and for male faces [F(1,82) = 46.76, p < 0.001, ηp2 = 0.36] of young adults, but only for male faces of middle-aged adults [F(1,82) = 53.69, p < 0.001, ηp2 = 0.19]. AES was not found for either male or female elderly faces.

A similar pattern of results was found for masked faces. Planned comparisons showed significant AES for female [F(1,82) = 14.09, p < 0.001, ηp2 = 0.15] and male faces [F(1,82) = 78.66, p < 0.001, ηp2 = 0.49] of young adults, and for male faces of middle-aged adults [F(1,82) = 35.45, p < 0.001, ηp2 = 0.30]. Interestingly, the AES was now present for female faces of middle-aged adults [F(1,82) = 6.31, p < 0.05, ηp2 = 0.07]. For old adults, there was an unexpected trend in the opposite direction, with smiling female [F(1,82) = 4.22, p < 0.05, ηp2 = 0.05] and male faces [F(1,82) = 4.91, p < 0.05, ηp2 = 0.06] perceived as younger than neutral faces. Due to the unpredicted small effect sizes, however, these results should be interpreted with caution.

To further test for possible mask-induced biases in age perception, we performed post hoc comparisons of the effect of mask within each combination of gender and expression in each age group. None of the comparisons were significant (all F’s < 1). These results coincide with the non-significant main effect of presentation format and suggest that masks do not lead to directional biases in age evaluations.

Accuracy in age evaluation

To test if masks interfered with the accuracy of age evaluations, we computed accuracy scores by calculating the average absolute difference between the perceived and real age of each of the faces in each combination of age group, gender, and expression. Accuracy scores for the different conditions are shown in Table 1. As can be seen in the table, accuracy decreased with age group and was overall lower for smiling compared to neutral faces (see Ganel & Goodale, 2021; Voelkle et al., 2012).

Table 1 Mean accuracy (absolute errors in years) of age evaluations in Experiment 1 (standard deviations in brackets). Note that larger numbers indicate lower accuracy

A mixed ANOVA design with gender, expression, and age group as the within-subject independent variables and with presentation format as a between-subjects independent variable was used to analyse the accuracy data.

A main effect was found for age group [F(1.67,136.9) = 4.76, p < 0.05, ηp2 = 0.05], reflecting higher accuracy in age judgments for young compared to middle-aged and old adults. A main effect was also found for expression [F(1,82) = 96.68, p < 0.001, ηp2 = 0.54], indicating reduced accuracy for smiling compared to neutral faces (Ganel & Goodale, 2021; Voelkle et al., 2012). The main effect of gender was significant [F(1,82) = 14.48, p < 0.001, ηp2 = 0.15], indicating overall reduced accuracy for female faces. This effect was qualified by a significant age group X gender interaction [F(1,82) = 12.81, p < 0.001, ηp2 = 0.14]. In addition, there was a significant interaction between age group and expression [F(2,164) = 20.68, p < 0.001, ηp2 = 0.19], resulting from a reduced effect of expression on accuracy scores in the old age group. As in the main analysis, there was no effect of masks on accuracy, indicated by a non-significant main effect of presentation format [F(1,82) = 0.01, p > 0.05]. The two-way interactions between age group and format [F(2,164) < 1, p > 0.05], between gender and format [F(1,82) < 1, p > 0.05], and between expression and format [F(1,82) < 1, p > 0.05] were not significant. The interactions between gender, expression, and format [F(1,82) = 1.98, p > 0.05, ηp2 = 0.02], between gender, expression, and age group [F(2,164) = 2.62, p > 0.05, ηp2 = 0.03], between gender, age group and format [F(2,164) = 2.95, p > 0.05, ηp2 = 0.03], and the four-way interaction [F(2,164) < 1, p > 0.05] were all not significant.

Response times

Response times were not the main dependent variable in our design (participants were not required to complete their age estimation in a speeded manner). Still, we analysed response times data to account for the possibility of speed-accuracy trade-off in accuracy of age evaluations. Response times were measured from the time of the presentation of the face until participants pressed the “Continue” button after they have completed to type their age evaluation in years. Outliers larger or smaller than 3 standard deviations above or below the mean were excluded from the response times analysis. Mean response times are presented in Table 2. For sake of brevity, we did not include the gender of the face in this analysis.

Table 2 Mean response times (in ms) to complete age evaluations of unmasked and masked neutral and smiling faces in the different age groups in Experiments 1 and 2 (standard errors of the mean in brackets)

A mixed ANOVA with expression and age group as the within-subject independent variables and with presentation format as a between-subjects independent variable showed a main effect of age group [F(2,168) = 12.74, p < 0.001, ηp2 = 0.13]. This effect resulted from longer response times to evaluate the ages of middle-aged adults faces compared to the other two age groups. The effect of presentation format was not significant [F(1,84) < 1], excluding the possibility of a speed-accuracy trade-off. All other effects and interactions were not significant and are not reported for sake of brevity.

The results of Experiment 1 extend our previous findings and show that the AES continues to be present in masked faces. This finding, together with the similar pattern of results found for unmasked and masked faces, suggests that age evaluations rely on visual information from regions of the face that are not covered by masks. It is still possible that the design of Experiment 1, which was focused on the effect of smiling in masked (and unmasked) faces, was not sensitive enough to detect possible effects of masks on age perception. Experiment 2 was designed to resolve this concern using a within-subject manipulation of masking that can provide a more sensitive measure for detecting possible effects of masks on the perception of age of neutral and smiling faces.

Experiment 2: within-subject manipulation of face masks

The mean perceived ages in the different categories are presented in Fig. 3. As can be seen in the figure, the results provide a close replication of the results of Experiment 1. In particular, in the unmasked condition, AES were found for female and male faces for young adults, and for male faces for middle-aged adults. AES was found in these groups and also for middle-aged adult females in the masked condition. There was no AES present in either the unmasked or the masked elderly faces.

Fig. 3
figure 3

The aging effect of smiling (AES) for masked and unmasked faces in Experiment 2. The results provide close replication of the results in Experiment 1. For masked faces, AES was found in young and middle-aged adults male and female photographs. Error bars represent standard errors of the mean

A repeated-measures ANOVA with presentation format (masked, unmasked), gender (of the face), expression, and age group as within-subject independent variables was used to analyse the data of the perceived age (in years). As in Experiment 1, main effects were found for age group [F(1.34,100.4) = 3486.48, p < 0.001, ηp2 = 0.98] for expression [F(1,75) = 41.88, p < 0.001, ηp2 = 0.36], and for gender [F(1,75) = 32.39, p < 0.001, ηp2 = 0.3]. As in Experiment 1, the main effect of mask was not significant [F(1,75) = 0.29, p > 0.05]. Again, this result shows that across all age groups and conditions, perceived age is not biased by the presence of a mask.

As in Experiment 1, there was a significant interaction between gender and expression [F(1,75) = 15.26, p < 0.001, ηp2 = 0.17], with a larger AES for male compared to female faces. Again, the significant interaction between age group and gender [F(2,150) = 142.1, p < 0.001, ηp2 = 0.66] was qualified by a three-way interaction with presentation format (masked vs. unmasked) [F(2,150) = 11.23, p < 0.001, ηp2 = 0.13]. A significant interaction was again found between age group and expression [F(2,150) = 20.26, p < 0.001, ηp2 = 0.21], indicating different AES in the different age groups. This interaction was qualified by a three-way interaction with gender [F(2,150) = 4.94, p < 0.05, ηp2 = 0.06]. The four-way interaction was also significant [F(1.65,123.38) = 5.33, p < 0.001, ηp2 = 0.07]. The two-way interactions between gender and format [F(1,75) = 2.17, p > 0.05, ηp2 = 0.03], between expression and format [F(1,75 < 1, p > 0.05] and the three-way interactions between gender, expression, and format [F(1,75) < 1, p > 0.05], between age group, expression, and format [F(1.58, 118.16) = 2.05, p > 0.05, ηp2 = 0.03] were all not significant.

Planned comparisons between smiling and neutral faces were performed to the test the presence of AES in the different conditions. In the unmasked condition, AES was again found for female [F(1,75) = 29.57, p < 0.001, ηp2 = 0.28] and male faces [F(1,75) = 15.91, p < 0.001, ηp2 = 0.16] of young adults, but only for male faces of middle-aged adults [F(1,75) = 31.17, p < 0.001, ηp2 = 0.29]. AES was not found for the faces of elderly male and female individuals. The pattern of results for masked faces was also similar to the one obtained in Experiment 1. A significant AES was found for female [F(1,75) = 5.31, p < 0.05, ηp2 = 0.07] and male faces [F(1,75) = 34.23, p < 0.001, ηp2 = 0.31] of young adults and for female [F(1,75) = 6.9, p < 0.05, ηp2 = 0.08] and male faces [F(1,75) = 18.58, p < 0.001, ηp2 = 0.19] of middle-aged adults. Unlike in Experiment 1, there was no difference between smiling and neutral faces of old adult males [F(1,75) < 1, p < 0.05]. Similar to the results of Experiment 1, smiling faces of old adult females were perceived as younger than faces with a neutral expression [F(1,75) = 8.19, p < 0.001, ηp2 = 0.09].

As in Experiment 1, we used post hoc comparisons to test for possible mask-induced biases within each combination of gender and expression in each age group. The pattern of results was inconsistent, both in terms of magnitude and direction, and in terms of the statistical significance. Out of the 12 specific comparisons three were significant. Out of these three comparisons, two went in one direction (faces with masks were perceived as slightly younger) and one went in the opposite direction (faces with masks were perceived as slightly older). In particular, masked faces were perceived as significantly older than unmasked faces for young adult neutral females [F(1,75) = 9.71, p < 0.05, ηp2 = 0.11] and for middle-aged adults neutral males [F(1,75) = 9.18, p < 0.05, ηp2 = 0.11]. At the same time, masked faces were perceived as marginally younger for older adult female smiling faces [F(1,75) = 8.41, p = 0.058, ηp2 = 0.1]. This inconsistent pattern of results is in agreement with the nonsignificant main effect of presentation format and show once again that masks do not impose a general (directional) bias on age evaluations.

Accuracy in age evaluations

Accuracy scores for the different conditions are shown in Table 3. As in Experiment 1, accuracy decreased with age group and was lower for smiling compared to neutral faces. Unlike Experiment 1, however, accuracy in the different age groups was lower for masked compared to unmasked faces.

Table 3 Mean accuracy (absolute errors in years) of age evaluations in Experiment 2 (standard deviations in brackets)

A repeated-measures ANOVA with presentation format, gender, expression, and age group was used to analyse the accuracy data. The main difference between the accuracy results here and in Experiment 1 was the significant reduction in accuracy for masked faces. This was indicated by a main effect of presentation format [F(1,75) = 18.54, p < 0.001, ηp2 = 0.19]. This main effect was qualified by format X gender interaction [F(1,75) = 6.83, p < 0.05, ηp2 = 0.08], which resulted from smaller effect of format for male compared to female faces. As in Experiment 1, main effects were found for age group [F(2,150) = 14.03, p < 0.001, ηp2 = 0.16], expression [F(1,75) = 101.52, p < 0.001, ηp2 = 0.58, and gender [F(1,75) = 28.22, p < 0.001, ηp2 = 0.27]. The main effect of gender was qualified by significant age group X gender interaction [F(1,75) = 24.96, p < 0.001, ηp2 = 0.25]. As was the case in Experiment 1, the interaction between age group and expression was significant [F(2,150) = 10.52, p < 0.001, ηp2 = 0.12]. The three-way interaction between gender, expression, and age group was also significant, F(2,150) = 6.61, p < 0.001, ηp2 = 0.08]. The two-way interactions between gender and expression [F(1,75) = 1.13, p > 0.05, ηp2 = 0.005], between age group and format [F(2,150 = 2.71, p > 0.05. ηp2 = 0.03], between expression and format [F(1,75) = 1.74, p > 0.05, ηp2 = 0.01] and the three-way interactions between gender, expression, and format [F(1,75) < 1, p > 0.05], and between age group, expression, and format [F(2,150) < 1, p > 0.05] were not significant.

Response times

Mean response times are presented in Table 2. As in Experiment 1, a mixed ANOVA with expression and age group as the within-subject independent variables and with presentation format as a between-subjects independent variable showed a main effect of age group [F(1.8,144.9) = 9.62, p < 0.001, ηp2 = 0.11]. As in Experiment 1, this effect was a consequence of longer response times to evaluate the ages of middle-aged adults faces compared to the other two age groups. The interaction between age group and expression was also significant [F(2,158) = 4.71, p < 0.01, ηp2 = 0.056]. All other effects and interactions were not significant.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Disclaimer:

This article is autogenerated using RSS feeds and has not been created or edited by OA JF.

Click here for Source link (https://www.springeropen.com/)