Working Papers

Under Review

de Freitas, J.*, & Johnson, S.G.B.* Optimality bias in moral judgment.

We often make decisions with incomplete knowledge of their consequences. Might people nonetheless expect others to make optimal choices, despite this ignorance? Here, we show that people are sensitive to moral efficiency: that people hold moral agents accountable depending on whether they make optimal choices, even when there is no way that the agent could know which choice was optimal. This result held up across judgments of wrongness, punishment, and blame; whether the outcome was positive, negative, or unknown; whether the agent had a positive or neutral intention; across within-subjects and between-subjects designs; and even when the relative quality of the agent’s choices was unknowable. Participants consistently distinguished between optimal and suboptimal choices, but not between suboptimal choices of varying quality. We argue that moral efficiency operates largely out of awareness, reflects broader tendencies in how humans understand one another’s behavior, and has real-world implications

Johnson, S.G.B., Valenti, J.J., & Keil, F.C. Simplicity and complexity preferences in explanation: An opponent heuristic account.

People often prefer simple to complex explanations because they generally have higher prior probability. However, simpler explanations are not always normatively superior because they often do not account for the data as well as complex explanations. How do people negotiate this trade-off between prior probability (favoring simple explanations) and goodness-of-fit (favoring complex explanations)? Here, we argue that people use opponent heuristics to simplify this problem—that people use simplicity as a cue to prior probability but complexity as a cue to goodness-of-fit. Experiment 1 finds direct evidence for this claim. In subsequent studies, we examine factors that lead one or the other heuristic to predominate in a given context. Experiments 2 and 3 find that people have a stronger simplicity preference in deterministic rather than stochastic contexts, while Experiments 4 and 5 find that people have a stronger simplicity preference for physical rather than social causal systems, suggesting that people use abstract expectations about causal texture to modulate their explanatory inferences. Together, we argue that these cues and contextual moderators act as powerful constraints that can help to specify the otherwise ill-defined problem of what distributions to use in Bayesian hypothesis comparison.

Working Papers

Johnson, S.G.B. Evaluating explanations: The lovely as a guide to the likely.

Many everyday inference problems take the form of diagnostic reasoning: Evaluating potential explanations (such as causes, categories, or mental states) for observed data (such as events, features, or behavior). Here, I review studies across these very different explanatory tasks, which converge on the view that people evaluate explanations by assessing their 'loveliness', using a set of explanatory heuristics. I further claim that while these heuristics can lead to error in particular cases, they are generally effective means of approximating the truth given the noisy data and computational complexity of probabilistic reasoning. Thus, I claim that we value lovely explanations because they typically guide us toward likely explanations.

Johnson, S.G.B. Explanatory logic in the pages of the Wall Street Journal.

Johnson, S.G.B. Scrupulous stopping: Peeking at the data to reduce false positives.

Psychology is a science in crisis. Prominent failures to replicate have damaged our field’s reputation and led many to doubt the validity of published findings. One contributing factor is optional stopping—collecting an initial sample, checking the p-value, and collecting additional data only if the significance threshold has not been met. Here, I propose that this problem suggests its own solution: scrupulous stopping, the practice of collecting an initial sample, checking the p-value, and collecting additional data until significant p-values have been observed multiple times. I report simulations demonstrating that this practice can control Type I errors while preserving statistical power. I argue, more generally, that any stopping rule is acceptable, so long as it is selected a priori, reported publicly, and has reasonable Type I and Type II error rates. This philosophy of statistics allows scientists to trade off error rates against resource consumption without sacrificing integrity.

Johnson, S.G.B., Jin, A., & Keil, F.C. Complexity bias in a visual task: Abductive heuristics used in fitting curves to scatterplot data.

To perceive the world, we must infer more structure than is immediately registered in the senses, such as recovering the three-dimensional world from a two-dimensional array of information—selecting the best explanation for the data. Here, we argue that behavior in visual tasks may rely on some of the same explanatory heuristics used in high-level cognition for evaluating hypotheses. We explore the use of an opponent heuristic system that uses simplicity to estimate the prior probability of a hypothesis and complexity to estimate its fit to the data. Four experiments examine this idea by testing participants’ inferences about the best fit curves for scatter plot data. Whereas previous research has revealed strong preferences for simple perceptual organizations, our Experiment 1 revealed that people tend to overfit scatter plot data, selecting curves that are more complex than normatively warranted. Experiment 2 found that people no longer overfit when the more complex curves were equally good fits to the data, and Experiment 3 revealed that more complex curves are mistakenly thought to be better fits to the data even when they are not. Finally, Experiment 4 looked at top-down influences on curve-fitting, finding that people calibrate their preferences based on the domain from which the data are drawn. Together, these experiments reveal that simplicity and complexity both play roles in visual tasks, in ways that are analogous to high-level cognition. Implications for explanatory reasoning and science education are discussed.

Johnson, S.G.B., Johnston, A.M., & Keil, F.C. Explanatory power as a substitute for statistical reasoning.

People judge the strength of cause-and-effect relationships as a matter of routine, and often do so in the absence of evidence about the covariation between cause and effect. Here, we examine the possibility that explanatory power is used as a heuristic for making these judgments. Our argument proceeds in three steps. First, we show that explanatory power and causal strength judgments for sets of historical events are almost perfectly correlated (Experiment 1). Next, we intervene on explanatory power without changing the target causal relation by manipulating explanatory scope—the number of effects predicted by an explanation. Scope manipulations lead to downstream consequences for causal strength judgments (Experiment 2), supported by item-by-item correlations between causal strength and explanatory power (Experiment 3). Finally, we show that explanatory power has causal signatures even in the non-causal domain of mathematics (Experiment 4). These results suggest that explanatory power may be a useful heuristic for estimating causal strength in the absence of statistical evidence.

Johnson, S.G.B., Kim, H.S., & Keil, F.C. Stereotyping as sense-making.

Johnson, S.G.B., Kim, K., & Keil, F.C. Perceptions of the unknown.

Johnson, S.G.B., Merchant, T., & Keil, F.C. Belief digitization.

Humans are often characterized as Bayesian reasoners. Here, we call into question the core Bayesian assumption that probabilities reflect degrees of belief. Across 11 studies, we found that people instead reason in a digital manner, assuming that uncertain information is either true or false when using that information to make further inferences. Participants learned about two hypotheses, both consistent with some information but one more plausible than the other. Although people explicitly acknowledged that the less-plausible hypothesis had positive probability, they ignored this hypothesis when using the hypotheses to make predictions. This was true across a variety of ways of manipulating plausibility (simplicity, base rates, and posterior probability) and task demands. Explicitly quantifying all relevant probabilities was the only boundary condition we could find, but even then participants under-utilized the less-plausible hypothesis compared to normative standards. We discuss implications for philosophy of science and for the organization of the mind.

Johnson, S.G.B., Merchant, T., & Keil, F.C. Toward an abductive account of inductive reasoning.

Johnson, S.G.B., Rajeev-Kumar, G., & Keil, F.C. Decisions influence beliefs.

Johnson, S.G.B., & Rips, L.J. The efficiency heuristic: Overextending the assumption of optimal choice in predicting and explaining decisions.

We often do not have access to all relevant information when making decisions. Here, we show that people nonetheless expect others to behave optimally even when they lack critical information. Four experiments test judgments about agents’ decision-making under uncertainty, comparing inferences about agents who were knowledgeable about the quality of each decision option to inferences about agents who were ignorant. Participants believed that even ignorant agents would choose optimally, both when the participants were predicting (Experiments 1, 2, and 4) and explaining (Experiment 3) the agents’ behavior. This effect was robust in both between-subjects (Experiments 1 and 2) and within-subjects (Experiments 3 and 4) designs, and it occurred in predictions of both one’s own behavior (Experiment 1) and others’ behavior (Experiment 2). We argue that this heuristic reflects a kind of behaviorist thinking that is pervasive in social judgment.

Johnson, S.G.B., Zhang, M., & Keil, F.C. Choice contexts attenuate explanatory reasoning biases.

Articles/Projects in Progress

Ahl., R.E., Johnson, S.G.B., Johnston, A.M., Dunham, Y., & Keil, F.C. The conceptual structure of mathematics across culture and development.

Burke, N., Johnson, S.G.B., & Keil, F.C. Folk reductionism.

Johnson, S.G.B. Necessary causation and sufficient causation.

Some causes are necessary for their effects to come about because the effect could come about no other way (e.g., sex is necessary for pregnancy). Other causes are sufficient for their effects because they lead to them with certainty (e.g., the explosion of the sun is sufficient for the destruction of the Earth). Three experiments demonstrate that people use these modal concepts of necessity and sufficiency over-and-above the probabilistic implications of causes on their effects. These intuitions can lead to intransitive causal judgment (Experiments 1 and 2) and to causal strength judgments that are U-shaped over the space of probabilities (Experiment 3), so that events are seen as more causal if they raise a probability from 0 to positive, or from positive to 1, even if the size of the probabilistic difference is the same. Implications for probabilistic theories of causal cognition, such as Bayesian theories, are discussed.

Johnson, S.G.B., & Hill, F. Inferred evidence and pessimism in consumer and managerial decision-making.

Decision-makers often lack critical pieces of information. For instance, consumers may not know how a new product would taste, and managers may not know how a job applicant would perform. In five experiments, we show that missing information leads to pessimistic judgments and choices, even when the information is missing for innocuous reasons (e.g., a consumer is unable to taste a product in the store, or a manager has misplaced an applicant's resumé). These pessimistic choices occur because missing information leads positive attributes to be seen as less likely, and negative attributes to be seen as no more likely, as predicted by recent findings in cognitive psychology. Implications for marketing and management are discussed.

Johnson, S.G.B., & Hill, F. Explanatory logic in theory of mind.

Johnson, S.G.B., & Keil, F.C. Explanatory scope: An asymmetry between positive and negative evidence.

Johnson, S.G.B., McNally, P., Royka, A., & Keil, F.C. What makes science interesting and important?

Scientists often speak of counterintuitive findings as "cute" or "sexy," and journal reviewers and editors often favor such results because they are thought to be more interesting to the public. In two studies using real scientific abstracts, we assess this claim empirically. We show that laypeople do find results to be more interesting and more important to the extent that they are counterintuitive, perceived as changing scientific understanding of a topic, and of practical importance. However, judgments of both interest and importance had no correlation whatsoever with the actual scientific importance of findings, as measured by citation counts three years post-publication. Thus, although scientists and editors may achieve popular fame by chasing after the counterintuitive, this strategy has poor returns for promoting truly innovative scientific breakthroughs.

Johnson, S.G.B., & Tuckett, D. Narrative decision-making in investment choices: How investors use news about company performance.

Story-telling helps to define the human experience. Do people also use narratives to make sense of, and to act on, financial information? Three studies demonstrate that people’s investment predictions and choices instead are by narrative thinking. Whereas neoclassical financial theory maintains that past public information cannot be used to predict future prices, participants used company performance information revealed before a base price quotation to project future price trends after that quotation (Experiment 1), as though they are using affectively laden information to predict the ending of a story. Importantly, these projections were stronger when information concerned predictions about a company’s future performance rather than actual data about its past performance, suggesting that people not only rely on financially irrelevant (but narratively relevant) information for making predictions, but erroneously impose temporal order on that information. These biased predictions had downstream consequences for asset allocation choices (Experiment 2) and these choices were driven in part by affective reactions to the company performance news (Experiment 3). These results could not be explained either by expectations conforming to neoclassical financial models or by expectations informed by behavioral anomalies. We conclude by discussing the prospects for a narrative theory of choice.

Johnson, S.G.B., Zhang, J., & Keil, F.C. Do consumers apply a zero-sum economic model to their own transactions?

People often speak of economic transactions having "winners" and "losers." Do consumers apply this zero-sum model of exchange to their own purchases? Eight studies demonstrate that they do indeed. When describing their own actual past and future purchasing choices, consumers frequently described themselves as worse off after the transaction. Since past work has identified perspective-taking errors and mercantilist thinking (valuing money over-and-above what it can purchase) as key drivers of zero-sum thinking, we attempted to intervene on these two mechanisms. Interventions designed to recognize the voluntariness and purposefulness of their transactions did not attenuate zero-sum thinking, but interventions designed to equip consumers to look through the "veil of money" were more effective. Marketing and policy implications are discussed.

Johnson, S.G.B., Zhang, J., & Keil, F.C. Zero-sum thinking in international consumer transactions.

Global trade has led to unprecedented standards of living in much of the world, yet recent political developments show that the idea of globalization has been called into question. The current studies examined how zero-sum thinking contributes to skepticism about international trade. Six studies find that equivalent transactions across international boundaries are seen as more harmful to the buyer's country compared to domestic transactions; even transactions across U.S. state lines led to similar perceptions. This harm is seen as collective rather than individual, has downstream consequences for the moralization of consumer transactions, and can be partly combatted by instructing consumers about the circular flow of currency among nations. Implications for political economy and for the marketing of international products are discussed.

Sheskin, M., Johnson, S.G.B., & Keil, F.C. Principled abstention as a political strategy for minimizing exposure to controversial social issues.

Sheskin, M., Johnson, S.G.B., Olash, C., & Keil, F.C. The ideological Turing test.