ABSTRACT
Multiple experimental, neuropsychological, and individual differences studies indicated that faces are processed (e.g. perceived and memorized) differently from non-face objects. It was suggested that face specificity is a result of configural processing which is different from a rather part-based processing of other objects. In this research, we investigated the specificity of face cognition in terms of processing style. To this aim, we estimated Time-on-Task Effects (ToTEs) that allow inferring the level of automaticity of face and house processing. We collected data from 219 participants by applying four perception and recognition tasks. Generalized linear mixed-effects modelling was used to estimate fixed and random effects. Random slopes were interpreted as individual differences in ToTEs. Findings suggested that in the majority of tasks, face processing was dissociable from the processing of non-face objects. Thereby, these results offer a new perspective on the nature of differences in face and object processing.
Acknowledgements
The tasks used in this study were a part of a larger test battery described by Nowparast Rostami et al. (Citation2017). Some data used in this research were analyzed beforehand with a different aim by Nowparast Rostami et al. (Citation2017) and by Meyer et al. (Citation2019). We thank Hadiseh Nowparast Rostami, Laura Kaltwasser, and many student assistants for data collection. We are grateful to these colleagues who shared part of their data with us.
Disclosure statement
No potential conflict of interest was reported by the authors.
Data availability statement
The data is available within the supplementary materials.
Notes
1 According to the recommendation of Matuschek et al. (Citation2017), we raised the alpha level for the likelihood ratio test to αLRT = .20, in order to avoid Type II errors in the process of model reduction.
2 In order to test H3 more directly and robustly, we additionally estimated separate models for face and house stimuli in each task, extracted the random slopes representing subject-level ToTEs and estimated their correlation. Next, we used bootstrapping (1000 samples) to estimate the 95% confidence intervals, and corrected the correlation along with its confidence intervals for attenuation (unreliability). Estimated correlations were consistently lower than the ones from joint models, and supported our conclusions about the separability of face and house ToTEs. Specifically, for LR the estimated R was .331 (95% CI: .199–.454).
3 The bootstrap test of the correlation between face and house ToTE for DR yielded R = .480 (95% CI: .352–.589).
4 The bootstrap test of correlation between face and house ToTE for IM yielded R = .604 (95% CI: .427–.754).
5 The bootstrap test of correlation between face and house ToTE for SM yielded R = .249 (95% CI: .104–.389).