Maestripieri et al. argue that attractive people receive higher financial remuneration because they are preferred as sexual partners. Although we agree that the available evidence shows that attractive people make more money than unattractive people, we suggest that differences between attractive and unattractive people do not necessarily mean that there are biases in favor of attractive people (as opposed to against unattractive people) and that there are explanations for attractiveness-related bias other than those examined by Maestripieri et al.

First, most of the research cited in the review defines attractiveness as a dichotomous variable, using only high and low levels of attractiveness. Therefore, the review cannot determine whether the effects are driven by a positive, beauty-is-good response, as the authors argue, or by a negative, ugly-is-bad response. Ugly-is-bad bias has been found in other research (e.g., Griffin & Langlois Reference Griffin and Langlois2006; Zebrowitz & Rhodes Reference Zebrowitz and Rhodes2004). Only research that includes a control or baseline group of medium attractive people can distinguish between these two alternatives.

Second, even young infants seem to prefer attractive faces (Langlois et al. Reference Langlois, Roggman, Casey, Ritter, Rieser-Danner and Jenkins1987), and adult heterosexuals choose attractive same-sex partners as friends (Langlois et al. Reference Langlois, Kalakanis, Rubenstein, Larson, Hallam and Smoot2000). Research on ugly-is-bad bias, infant research, and research on same-sex preferences for attractive others are not consistent with most versions of mating strategy.

As to explanatory mechanism, we propose that a domain general information processing system, cognitive averaging, results in preferences for attractive faces. In the initial study of cognitive averaging theory, Langlois and Roggman (Reference Langlois and Roggman1990) mathematically averaged 32 individual female faces together to create a female face morph/blend and 32 individual male faces to create a male face morph/blend. The morphed faces increased in judged attractiveness as more faces were added. Even when created with independent sets of 32 individual faces, the morphs look quite similar to one another, suggesting that a 32-face morph is a prototype of an adult face. Both averaged and attractive faces may be perceived as prototypes and, thus, seem more familiar to the viewer, even if the face is novel (Langlois et al. Reference Langlois, Roggman and Musselman1994). Faces that represent the mathematical average or central tendency of a population (e.g., male or female) also seem more typical and to be “better examples” of a face and therefore are preferred. In addition, faces whose structure approximates the mathematical average facial configuration of a population are more fluently processed than faces distant from the central tendency. Fluent processing produces positive affect, which could explain why attractive people are perceived more positively and hold better jobs with higher salaries. Humans automatically create prototypes of faces, and even infants can abstract prototypes from individual exemplars (Rubenstein et al. Reference Rubenstein, Kalakanis and Langlois1999; Strauss Reference Strauss1979).

Multiple studies with adults have provided evidence that high attractive, prototypical faces are more fluently processed than low attractive, nonprototypical faces. Averaged and high attractive faces rated low in distinctiveness (a subjective measure of typicality) are categorized faster than low attractive, high distinctive faces in a species categorization task (Trujillo et al. Reference Trujillo, Jankowitsch and Langlois2014). Attractiveness facilitates the speed and accuracy of gender-based face classification (Hoss et al. Reference Hoss, Ramsey, Griffin and Langlois2005). Moreover, prototypicality predicts perceptual fluency and increased liking for non-face stimuli as well. Dot patterns and geometric shapes are judged to be more attractive and are more rapidly categorized when they are close to the prototype (Posner & Keele Reference Posner and Keele1968; Winkielman et al. Reference Winkielman, Halberstadt, Fazendeiro and Catty2006). In addition, perceiving and processing prototypical faces and dot patterns requires fewer neural resources comparedwith perceiving nonprototypical stimuli (Leopold et al. Reference Leopold, Bondar and Giese2006; Loffler et al. Reference Loffler, Yourganov, Wilkinson and Wilson2005; P. J. Reber et al. Reference Reber, Stark and Squire1998; Trujillo et al. Reference Trujillo, Jankowitsch and Langlois2014); such a reduction in neural resource use is a hallmark of perceptual fluency.

Importantly, the fluent processing accorded by prototypicality leads to more favorable judgments of perceived stimuli (Winkielman et al. Reference Winkielman, Halberstadt, Fazendeiro and Catty2006) and also influences affective states. R. Reber et al. (Reference Reber, Winkielman and Schwarz1998) argue that fluency is in itself pleasant. Studies that have experimentally manipulated fluency (e.g., Monahan et al. Reference Monahan, Murphy and Zajonc2000; Zajonc Reference Zajonc2001) have found that increased levels of fluency augment overall mood and increase generalized positive affect.

Beyond faces and dot patterns, participants show preferences for prototypicality in many other types of stimuli, including color patches (Martindale & Moore Reference Martindale and Moore1988), music (Repp Reference Repp1997), cubist paintings (Hekkert & Van Wieringen Reference Hekkert and Van Wieringen1990), and voices (Bruckert et al. Reference Bruckert, Bestelmeyer, Latinus, Rouger, Charest, Rousselet and Belin2010), likely because of the ease in processing stimuli closest to the prototype. The wide variety of stimuli that conform to this prototypicality or averaging effect suggest that an evolved domain general mechanism such as cognitive averaging is a more likely explanation for attractiveness preferences than a domain-specific mechanism such as mate selection.