We agree with the author's concerns about the flaws that are the focus of the target article. They are part of the ecological validity problem that plagues a variety of experimental approaches in the behavioral sciences (Salmon, Reference Salmon2020). While some psychologists have acknowledged concerns about generalizing from WEIRD (western, educated, industrialized, rich, and democratic) samples to the wider human population (Baumard & Sperber, Reference Baumard and Sperber2010; Henrich, Heine, & Norenzayan, Reference Henrich, Heine and Norenzayan2010), researchers have often failed to recognize the differences between the lab and the “real world” which challenges applicability to real world problems. The concern is whether there is a good fit between the task and the ecological problem it is approximating (Adolph, Reference Adolph and Sternberg2019; Salmon, Reference Salmon2020). While the target article focuses on decision-making, research examining effects of pornography on violence has also been plagued by experimental results that don't correspond to those of more ecologically valid studies (Diamond, Jozifkova, & Weiss, Reference Diamond, Jozifkova and Weiss2011; Ferguson & Hartley, Reference Ferguson and Hartley2020; Hatch et al., Reference Hatch, Esplin, Aaron, Dowdle, Fincham, Hatch and Braithwaite2020)

When designing tasks to test hypotheses, we must be mindful of all contexts and factors the mechanism of interest is theoretically sensitive to and what aspects are likely to be more general across contexts. Consider deception detection research. The tasks participants face in detecting deception in the lab are quite different from those faced in everyday life, where they have a great deal more information about base rates of lying in their environment, specific individuals (reputation, non-verbal cues, and motivation), and ways to test if they are lying such as other sources of information (Levine, Reference Levine2018). Lab studies are useful are in that they can provide information about the unique effect of an isolated factor on some particular outcome with “all else being equal.” They are less helpful, however, for understanding how the system works when other relevant factors are included. Contrary to the strict experimental control employed in lab studies, in the real world, never is “all else equal.” Individual and group differences exist, which likely influence behavior, attitudes, and cognition. Therefore, more multifactorial studies are needed to examine not only the effect of multiple factors at the same time, but also the potential combined effect of factors on some outcome.

Good science starts with thorough observations of the behavior of interest and then moves to hypothesis testing. The author highlights this in the police shooting case by pointing out the value of conducting task analysis of actual shootings first to get a more complete understanding of (a) all the relevant variables and (b) which may be most critical to include in experimental hypothesis testing. Assuming the variable you are interested in is the most relevant one, without comprehensive descriptive observational work, is likely to lead to erroneous conclusions for applications outside the lab. This is especially concerning when findings are used to inform public policy.

We disagree with the author that “distal causes of group differences are irrelevant” because they are “separable” from questions about specific outcomes. Rather than separate issues, they represent different levels of analysis. We would argue that another inherent problem of social psychological research has been ignoring ultimate factors (e.g., evolved psychological adaptations for solving recurring problems across our ancestral past) and focusing exclusively on the proximate mechanisms. Knowledge of why a particular behavior or outcome is occurring, informs our understanding of what is occurring as well as other factors that may be relevant.

The importance of understanding distal factors can be seen in the different patterns of stereotyping for different social groups. Although stereotype bias can be diminished by reducing ambiguity and providing more individuating information for some social groups, the reduction of ambiguity does not dramatically reduce stereotype bias against all social groups. Failure to acknowledge the ultimate origins of different types of stigma results in the error of assuming that all biased stereotyping processes occur under the same circumstances, in the same way, and thus may be mitigated in the same way. Consider differences between ageism and racism.

Consistent with age being a natural social category (i.e., one that would have existed across our evolutionary past), stereotypes about age tend to have cross-cultural similarities (Fiske, Reference Fiske2017). Across cultures, older adults are commonly described as being doddering but dear, incompetent but warm (Cuddy, Norton, & Fiske, Reference Cuddy, Norton and Fiske2005; North & Fiske, Reference North and Fiske2015). The only cultures found to admire their older adults were Native Americans (Burkley, Durante, Fiske, Burkley, & Andrade, Reference Burkley, Durante, Fiske, Burkley and Andrade2017) and African Americans (Fiske et al., Reference Fiske, Bergsieker, Russell and Williams2009). Inconsistent with the popular belief that older adults are revered in Eastern cultures, older adults were found to be more derogated in Asian cultures than in Western cultures (North & Fiske, Reference North and Fiske2015). This is important in that it highlights a disconnect between attitudes toward older adults and cultural practices/expectations that aging parents are cared for by their adult children in Asian cultures. This disconnect suggests that stereotype beliefs about older adults do not necessarily map onto behavioral outcomes.

Stereotypes about social groups not considered to represent natural social categories (e.g., race), tend to be much more variable across cultures (Fiske, Reference Fiske2017). Different beliefs attached to varying ethnic groups depend on the specific nation and its history, including factors such as immigration history, income equality, political systems, and amount of conflict versus peace (Durante et al., Reference Durante, Fiske, Gelfand, Crippa, Suttora, Stillwell and Teymoori2017). Therefore, racial stereotypes appear to be determined by historical events (Fiske, Reference Fiske2017). The different patterns of racial stereotypes, for example, suggest race is being used as a proxy for something else (e.g., coalition membership). As pointed out by the author, racial stereotypes tend to be used in decision-making in the absence of other cues. A recent meta-analysis of “erasing race” studies in the United States confirmed a robust effect of reducing race-encoding by providing other cues to group membership such as team membership (Woodley of Menie et al., Reference Heeney, Peñaherrera-Aguirre, Sarraf, Banner and Rindermann2020). This suggests that racial biases may be activated under conditions of ambiguity, but not activated when other (perhaps more reliable) cues of coalitional membership are present. The same, however, does not apply to natural category biases. Those stereotypes appear to be consistently, unconsciously activated, even in unambiguous situations.