People are able to judge men’s and women’s sexual orientation with above-chance accuracy relying on no more than grossly impoverished facial photographs (i.e., grayscale, hair-removed) presented for as few as 40–50 ms 
. Despite the growing literature on reflexive, intuitive, rapid, or “snap” judgments of sexual orientation from faces, little is known about the processes by which these judgments are formed. When making accurate judgments of sexual orientation, do people rely solely on the processing of individual facial features (i.e., featural face processing)? Do people rely on the processing of relationships among facial features (i.e., configural face processing)? Or, do people rely on some combination of both? Investigating the face processing mechanisms governing sexual orientation judgments has implications for understanding whether sexual orientation is judged as category-based (e.g., male vs. female; black vs. white) or identity-based (e.g., familiar vs. unfamiliar) person information. Testing the roles of configural and featural face processing on the accuracy of snap judgments of sexual orientation – that is, rapid and intuitive judgments of sexual orientation – is the primary goal of the present research.
Existing work investigating what
type of social information underlies judgments of sexual orientation from faces has indicated that gender atypicality, whether natural or manipulated via morphing software, makes faces more likely to be perceived as gay or lesbian 
. Other research focusing on where
in the face valid cues reside has shown that the mouth and eye areas, alone, enable above-chance accuracy in sexual orientation judgments 
. Despite this growing body of research, work has yet to examine how
the face is processed to give rise to reliable judgments of sexual orientation. Research indicates that there are two routes for perceiving the human face: featural processing
primarily encodes individual facial features (e.g., an eye or nose), and configural processing
primarily encodes relationships among featural cues (e.g., distance between the eyes) 
It is important to clarify the distinction between the how
(face processing) question and the where
(in the face) question, which has been addressed in previous research 
. For example, Rule et al. 
showed that men’s sexual orientation could be judged with above-chance accuracy from the eye area alone or the mouth area alone, and that accuracy for either of these areas of the face was lower than accuracy for the whole face. At first glance, these results may suggest not only that judgments of sexual orientation involve certain facial areas, but also that configural face processing (whole faces) raised accuracy above the accuracy enabled by either of the individual areas of the face. However, an alternative explanation is that the individual facial areas provide at least partially independent sources of sexual orientation information and that when presented simultaneously (as a whole face), judgment accuracy increased (compared to accuracy for each area alone) simply because more featural information was available (but not necessarily because any configural processing occurred). Thus, to date, the role of configural face processing in judgments of sexual orientation judgments is unknown.
Configural processing can refer to several distinct ideas. Following the definition provided by Maurer et al. 
, in this paper, we refer to configural processing as any or all of the following: (a) processing the ordinal spatial relationships among individual features (e.g., eyes appear above noses), (b) processing the cardinal spatial relationships among individual features (e.g., the amount of space between the eyes), or (c) processing the face in a holistic or gestalt manner (i.e., the general shape of the face). All three subtypes of configural face processing are diminished by facial inversion; disentangling these subtypes of configural face processing is beyond the scope of this paper. Of note, specific areas of the face (e.g., pairs of eyes) can possess both featural cues (e.g., an eye) and configural cues (e.g., distance between eyes) (for a review of different types of face processing, see 
). Thus, strictly speaking, previous work showing that individual areas of the face can enable above-chance sexual orientation judgments 
does not indicate whether judgments of each face area were driven by featural processing, configural processing, or both.
Comparing the accuracy of judgments made from facial photographs presented upright vs. upside-down is one method for determining whether configural face processing contributes to a character judgment 
. Displaying photographs of faces upright allows for unimpeded processing of both featural and configural facial cues; in contrast, displaying facial photographs upside-down severely disrupts processing of configural facial cues but has little 
or no detectable effect on featural face processing 
. More concretely, when faces are manipulated so that they differ in featural information (e.g., shape of eyes, nose or mouth 
; eye color 
; combinations of eye color and hair color 
; brightness of individual facial features 
), individuals are able to distinguish faces (i.e., by making same vs. different judgments when first seeing or when recalling faces) when they are presented upright as well as upside-down 
. However, when faces are manipulated so that they differ in configural information (e.g., distance between nose and mouth 
; mouth or eye position 
; interocular distance 
), individuals are only able to distinguish faces when faces are presented upright 
; see 
, p. 257, .
Accuracy of detecting sexual orientation from upright and upside-down faces (Experiment 2).
Research and theorizing by Cloutier and colleagues suggests that understanding face processing mechanisms (i.e., featural and configural) has implication for various social inferences, with social categorization (e.g., male vs. female; black vs. white) relying heavily (but not necessarily exclusively, e.g., 
) on featural processing, and identity judgments (e.g., familiar vs. not; famous vs. not) relying heavily (but not necessarily exclusively) on configural processing. Using this facial inversion technique, Cloutier, Mason, and Macrae 
showed that judgments of sex could be accurately rendered when faces were both upright and upside down. In contrast, judgments of fame (famous vs. not famous) could be accurately rendered when perceivers viewed target faces upright but were significantly less accurate when the faces were presented upside-down–i.e., when configural face processing is dramatically impaired, e.g., 
. Because “the extraction of featural information is largely resistant to the effects of inversion” 
(p. 886), the researchers concluded that judging fame requires configural face processing – which shows reliably large effects of facial inversion.
The Present Research
Does configural face processing contribute to accuracy of sexual orientation judgments? Understanding the processing that allows sexual orientation to be read from faces may reveal how sexual orientation, as a social construct, is conceptualized. Whereas featural face processing is sufficient to enable judgments of social category information (e.g., male vs. female; black vs. white), configural face processing is necessary to enable judgments of social identity information (e.g., familiar vs. unfamiliar; famous vs. not famous) 
; see also 
. In contrast to categories such as race or gender, sexual orientation is less obvious. Thus, it is unclear whether individuals would rely on category (featural) face information, individuating (configural) face information, or both.
Additionally, we investigated the effects of stimulus gender in the ability of participants to make reliable sexual orientation judgments. Previously, snap judgments of sexual orientation have been examined separately for men’s and women’s faces. Casual comparison of accuracy rates across papers, i.e., 
, implies that women’s sexual orientation may be judged more accurately than men’s, but direct comparisons have not been performed to date. The possibility that judgments of sexual orientation differ as a function of gender is likely given the well-established gender differences in experiences of romantic love and sexual desire, neurophysiological and hormonal responses to sex and attachment, and phenomenology of sexual orientation, e.g., 
. Thus, we predict that facial markers of sexual orientation may differ by gender, as well.
Configural face processing contributes to accurate snap judgments of sexual orientation. Because sexual orientation is phenotypically ambiguous, we predicted that the deeper, more individuating type of face processing – configural face processing – would contribute to judgment accuracy. In practical terms, this means that judgment accuracy should be reduced when faces are presented upside-down (vs. upright).
The process of reading sexual orientation from faces may differ as a function of whether the stimulus person (face) is male or female. In the present experiments, participants judged both men’s and women’s faces, allowing for direct comparisons of judgments as a function of target gender. This hypothesis is exploratory in nature and does not carry a directional prediction.