A data-driven, hyper-realistic method for visualizing individual mental representations of faces

Abstract

Research in person and face perception has broadly focused on group-level consensus that individuals hold when making judgments of others (e.g., “X type of face looks trustworthy”). However, a growing body of research demonstrates that individual variation is larger than shared, stimulus-level variation for many social trait judgments. Despite this insight, little research to date has focused on building and explaining individual models of face perception. Studies and methodologies that have examined individual models are limited in what visualizations they can reliably produce to either noisy and blurry or computer avatar representations. Methods that produce low-fidelity visual representations inhibit generalizability by being clearly computer manipulated and produced. In the present work, we introduce a novel paradigm to visualize individual models of face judgments by leveraging state-of-the-art computer vision methods. Our proposed method can produce a set of photorealistic face images that correspond to an individual's mental representation of a specific attribute across a variety of attribute intensities. We provide a proof-of-concept study which examines perceived trustworthiness/untrustworthiness and masculinity/femininity. We close with a discussion of future work to substantiate our proposed method.

Generative models for visualizing idiosyncratic impressions

“…the interaction of perceiver and stimulus; the other component is the perceiver's variance) and to remove the error variance from the meaningful variance. Nevertheless, their estimates are comparable to other more recent studies (Albohn et al, 2022; Hester et al, 2021; Martinez et al, 2020). In cases of complex judgements such as perceived trustworthiness and competence, the idiosyncratic variance trumps shared variance.…”

“…If documenting the importance of idiosyncratic variance is the first step in the study of idiosyncratic differences in impressions, one of the final steps is building models of idiosyncratic representations of impressions. Recently, capitalizing on the power of generative face models and borrowing procedures from psychophysical reverse correlation, we have proposed and illustrated methods for building such representations (Albohn et al, 2022). But many questions remain unresolved.…”

Spontaneous Content of Impressions of Naturalistic Face Photographs

“…In fact, our results suggest that, although responses were largely reliable within participants, consensus on the content of spontaneous impressions across participants was relatively low (see Supplement). This is consistent with work on partitioning the variance of judgments from faces to variance due to the stimuli (shared) and variance due to participants (idiosyncratic) [54][55][56][57] . Thus, aggregating across participants can mask stable individual differences, resulting in an "averaging out" of evaluations.…”

Facecraft: Race Reification in Psychological Research With Faces

“…When deployed in a context-informed manner (Cikara et al, 2022), these techniques can help identify the facial features used to map race categories onto people within a context. For these models to be context-informed, people's varied understandings need to be centered in the analysis (Albohn et al, 2022;Martinez & Todorov, 2021). Attending not just to sampleaveraged representations as currently practiced but to the heterogeneity of visual features used by different people for the same category within and across contexts is critical to not reify the average representation as a generalizable racialization that operates for everyone, but to instead map topographies of racializing processes.…”