ABSTRACT
Research has shown that observers often spontaneously extract a mean representation from multiple faces/objects in a scene even when this is not required by the task. This phenomenon, now known as ensemble coding, has so far mainly been based on data from Western populations. This study compared East Asian and Western participants in an implicit ensemble-coding task, where the explicit task was to judge whether a test face was present in a briefly exposed set of faces. Although both groups showed a tendency to mistake an average of the presented faces as target, thus confirming the universality of ensemble coding, East Asian participants displayed a higher averaging tendency relative to the Westerners. To further examine how a cultural default can be adapted to global or local processing demand, our second experiment tested the effects of priming global or local processing orientation on ensemble coding via a Navon task procedure. Results revealed a reduced tendency for ensemble coding following the priming of local processing orientation. Together, these results suggest that culture can influence the proneness to ensemble coding, and the default cultural mode is malleable to a temporary processing demand.
Data statement
The data and materials for all experiments are available upon request and none of the experiments was preregistered.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 The kind of repeated presentation of the same stimuli was adopted in previous studies (de Fockert & Wolfenstein, Citation2009; Haberman et al., Citation2015). See, for example, Study 1 of Haberman et al. (Citation2015), where “observers saw a total of 20 unique face sets and 20 unique gabor sets 8 times each”. Haberman et al. (Citation2015) computed the reliability of the ensemble coding task and found that the Cronbach’s alpha was .84 in their ensemble face task, suggesting that averaging of multiple face identities was highly consistent. However, according to the results of the ANOVA based on the first block in Experiment 2, we found that though the result pattern did keep similar, the effect size of the Priming Condition × Test-image Type interaction based on the first block (F (1, 58) = 8.55, p = .005, η2 = .13) was smaller than that based on the all trials (η2 = .35) as shown in the main text.