During a family dinner, a chat sways into politics. Most guests have an opinion about politicians and their conduct, labeling some politicians as irrational or immoral. How did the guests form these opinions? Unless the family belongs to political elites, these opinions are likely based on news reports or commentaries on social media. Such news posts are brief, decontextualized, and ill-defined, with uncertainty about politicians' interests and preferences – caricatured abstractions, often far off from reality. However, even if guests are aware of these limits to their knowledge, it hardly stops them from expressing rather strong opinions about politicians.

We use politicians, but many everyday judgments can be characterized this way – fast and based on limited information. Another feature of judging others is that such judgments – as long as they don't trigger a response – typically bear little consequence to the person expressing them.

In contrast, consider how guests at the same family dinner would decide on particular policies. Here, much more information is likely to be incorporated in their decision: Goals and preferences are concrete, and the decision is put in a rich context of their political and social background. Notably, the consequences of their decisions will have a big real-world impact via the outcomes of their choice. Here, the decision about policies appears qualitatively different from the judgment about politicians proposing these policies, even though both judgment and decision-making processes can inform each other (Ariely & Norton, Reference Ariely and Norton2008).

To further unpack this distinction between decision and judgment, consider a visit to a restaurant. You notice pulled pork and smoked ribs on the menu. Though both are your favorites, you like pulled pork a bit more, say 8 versus 9 points on a 10-point scale. When ranking the two meals, you will always put pulled pork before smoked ribs. But this does not mean you will always order pulled pork and will never order ribs. There is more for you to consider when deciding about a particular order than merely the judgment in a form of ranking your preferences. Judgments tell us about values, beliefs, and preferences, whereas decisions also tell us how judgments are (mis)applied into action. The strength of external factors in judgment and decision making is illustrated by the decoy effect – that is, a phenomenon whereby people who value option A over B may sometimes chose option B when presented with a third option C (Huber, Payne, & Puto, Reference Huber, Payne and Puto1982). Note, we do not claim decision is just distorted judgment. We simply highlight that judgments and decisions differ in several ways.

These insights have direct implications to social bias research. In a typical study, participants are presented with experimental stimuli (words and pictures) and are asked to choose, often in a narrow time slot, between the two options. For example, a picture of a black or white man is presented with an item, and participants are asked to quickly choose to “shoot” or not to. The experimental set-up aims to create a simplified, bare-bone representation of behavior. For ethical reasons, other experimental paradigms may not always be possible. In pursuit of experimental control, these experiments are stripped of much context information, and consequently diverge from real-world decisions. Cesario highlights these differences, raising a provocative question about the reduced utility of such experiments. When considering predictions of real-world behavior, we agree with this sentiment. At the same time, we note that such decontextualized studies may, in fact, resemble conditions in which judgments are made. As we outlined in examples above, judgments are often decontextualized, are made on the fly, and bear little consequences to the agent. Thus, social bias experiments may well mimics conditions in which people make judgments, especially judgments of the conduct of others, despite lack of predictive power for the real-world decisions.

If judgments and decisions are two different but interrelated phenomena, there are implications for how to interpret the results of social bias experiments. Cesario notes that having observed an implicit bias in an agent, one cannot infer that this agent will necessarily make biased real-life decisions. More likely, such person will make biased judgments.

How people make judgments is not only relevant to the study of social bias, but also for other areas of psychology, including the closely related research on moral psychology. Humans judge each other's morality at a much greater rate that they make morally charged decisions. Hence, learning about processes underlying moral judgments seems at least as important for moral psychology to answer than processes underlying moral decisions (Białek, Turpin, & Fugelsang, Reference Białek, Turpin and Fugelsang2019). In a similar vein, people can judge others daily, but only sometimes actively discriminate against others. Although the latter may have more severe consequences, biased judgment can also be devastating. For example, biased judgment can support system justification (Jost, Banaji, & Nosek, Reference Jost, Banaji and Nosek2004), and maintenance of discriminatory laws harming the group one is biased against. Bias can lead disadvantages groups to favor privileged outgroup (dos Santos & Pereira, Reference dos Santos and Pereira2021). Hence, discovering how biased judgments are formed allows us to understand psychological reasons for support of existing inequalities.

We believe that these criticized experiments on bias have an enormous value for social sciences: They inform us about core beliefs and preferences of particular social groups. Whether a person belonging to a given group will act on these beliefs is a distinct, and arguably more complex question. It may be unreasonable to expect researchers to comprehensively answer both in one project. After all, social scientists are not (well) trained in prediction modeling of social issues (Hofman, Sharma, & Watts, Reference Hofman, Sharma and Watts2017; Yarkoni & Westfall, Reference Yarkoni and Westfall2017), often ignore the broader cultural (Henrich, Heine, & Norenzayan, Reference Henrich, Heine and Norenzayan2010) and cross-temporal factors (Grossmann & Varnum, Reference Grossmann and Varnum2015; Varnum & Grossmann, Reference Varnum and Grossmann2017), and consequently appear as inaccurate in their assessments of broader societal issues as an average person on the street (Hutcherson et al., Reference Hutcherson, Sharpinskyi, Varnum, Rotella, Wormley, Tay and Grossmann2021).

Instead, it may be prudent to establish robust scientific evidence one step at a time. For instance, researchers may start by focusing on the study of agent's preferences and biases, prior to scaling up a model to predict real-life decisions. The second step will require extending the experiments by considering contextual cues. The third step will benefit from greater integration of insights from computational social sciences and complex systems for predictive modeling of human behavior (e.g., Hofman et al., Reference Hofman, Sharma and Watts2017; Yarkoni & Westfall, Reference Yarkoni and Westfall2017). Only by integrating these steps together, social scientists can start translating insights from experiments about social judgment biases into the study of real-world behavior.