Non-Bayesian Inference: Causal Structure Trumps Correlation

How Do We Believe?

My first 30-odd years of research in cognitive science has been driven by an attempt to balance two facts about human thought that seem incompatible and two corresponding ways of understanding information processing. The facts are that, on one hand, human memories serve as sophisticated pattern recognition devices with great flexibility and an ability to generalize and predict as long as circumstances remain sufficiently familiar. On the other hand, we are capable of deploying an enormous variety of representational schemes that map closely onto articulable structure in the world and that support explanation even in unfamiliar circumstances. The contrasting ways of modeling such processes involve, first, more and more sophisticated associative models that capture progressively higher-order statistical structure and, second, more powerful representational languages for other sorts of structure, especially compositional and causal structure. My efforts to rectify these forces have taken me from the study of memory to induction and category knowledge to causal reasoning. In the process, I have consistently appealed to dual systems of thinking. I have come to realize that a key reason for our success as cognizers is that we rely on others for most of our information processing needs; we live in a community of knowledge. We make use of others both intuitively-by outsourcing much of our thinking without knowing we are doing it-and by deliberating with others.

How Does Explanatory Virtue Determine Probability Estimation?-Empirical Discussion on Effect of Instruction

It is important to reveal how humans evaluate an explanation of the recent development of explainable artificial intelligence. So, what makes people feel that one explanation is more likely than another? In the present study, we examine how explanatory virtues affect the process of estimating subjective posterior probability. Through systematically manipulating two virtues, Simplicity-the number of causes used to explain effects-and Scope-the number of effects predicted by causes-in three different conditions, we clarified two points in Experiment 1: (i) that Scope's effect is greater than Simplicity's; and (ii) that these virtues affect the outcome independently. In Experiment 2, we found that instruction about the explanatory structure increased the impact of both virtues' effects but especially that of Simplicity. These results suggest that Scope predominantly affects the estimation of subjective posterior probability, but that, if perspective on the explanatory structure is provided, Simplicity can also affect probability estimation.

The environmental malleability of base-rate neglect

Across two experiments (N=799) we demonstrate that people's use of quantitative information (e.g., base-rates) when making a judgment varies as the causal link of qualitative information (e.g., stereotypes) changes. That is, when a clear causal link for stereotypes is provided, people make judgments that are far more in line with them. When the causal link is heavily diminished, people readily incorporate non-causal base-rates into their judgments instead. We suggest that people use and integrate all of the information that is provided to them to make judgements, but heavily prioritize information that is causal in nature. Further, people are sensitive to the underlying causal structures in their environment and adapt their decision making as such.

The ecological rationality of explanatory reasoning

There is growing evidence that explanatory considerations influence how people change their degrees of belief in light of new information. Recent studies indicate that this influence is systematic and may result from people's following a probabilistic update rule. While formally very similar to Bayes' rule, the rule or rules people appear to follow are different from, and inconsistent with, that better-known update rule. This raises the question of the normative status of those updating procedures. Is the role explanation plays in people's updating their degrees of belief a bias? Or are people right to update on the basis of explanatory considerations, in that this offers benefits that could not be had otherwise? Various philosophers have argued that any reasoning at deviance with Bayesian principles is to be rejected, and so explanatory reasoning, insofar as it deviates from Bayes' rule, can only be fallacious. We challenge this claim by showing how the kind of explanation-based update rules to which people seem to adhere make it easier to strike the best balance between being fast learners and being accurate learners. Borrowing from the literature on ecological rationality, we argue that what counts as the best balance is intrinsically context-sensitive, and that a main advantage of explanatory update rules is that, unlike Bayes' rule, they have an adjustable parameter which can be fine-tuned per context. The main methodology to be used is agent-based optimization, which also allows us to take an evolutionary perspective on explanatory reasoning.

Stronger, sooner, and more certain climate change: A link between certainty and outcome strength in revised forecasts

What do lay people think about revised forecasts of future outcomes? A series of experiments show that when climate change forecasts are revised in an upward direction (e.g., a higher sea level rise is predicted in light of new information), the forecast is perceived as more certain in contrast to revisions in a downward direction. This association is bidirectional so that people also think a forecast that has become more certain (uncertain) indicates a stronger (weaker) outcome. Furthermore, when the timing of the outcome is revised, a predicted sooner occurrence (a “stronger” outcome) was judged to be more certain than a predicted delayed occurrence. Upward revisions may also lead to more positive impressions of the forecast and the forecaster, with clear implications for the communication of uncertainty, both for climate change and in other domains. Different theoretical explanations for the results are discussed.

Causal learning in children

How do children learn the causal structure of the environment? We first summarize a set of theories from the adult literature on causal learning, including associative models, parameter estimation theories, and causal structure learning accounts, as applicable to developmental science. We focus on causal graphical models as a description of children's causal knowledge, and the implications of this computational description for children's causal learning. We then examine the contributions of explanation and exploration to causal learning from a computational standpoint. Finally, we examine how children might learn causal knowledge from others and how computational and constructivist accounts of causal learning can be integrated. WIREs Cogn Sci 2014, 5:413-427. doi: 10.1002/wcs.1291 For further resources related to this article, please visit the WIREs website.

CONFLICT OF INTEREST

The authors have declared no conflicts of interest for this article.

Causality in thought

Causal knowledge plays a crucial role in human thought, but the nature of causal representation and inference remains a puzzle. Can human causal inference be captured by relations of probabilistic dependency, or does it draw on richer forms of representation? This article explores this question by reviewing research in reasoning, decision making, various forms of judgment, and attribution. We endorse causal Bayesian networks as the best normative framework and as a productive guide to theory building. However, it is incomplete as an account of causal thinking. On the basis of a range of experimental work, we identify three hallmarks of causal reasoning-the role of mechanism, narrative, and mental simulation-all of which go beyond mere probabilistic knowledge. We propose that the hallmarks are closely related. Mental simulations are representations over time of mechanisms. When multiple actors are involved, these simulations are aggregated into narratives.

Bayesian learning and the psychology of rule induction

In recent years, Bayesian learning models have been applied to an increasing variety of domains. While such models have been criticized on theoretical grounds, the underlying assumptions and predictions are rarely made concrete and tested experimentally. Here, I use Frank and Tenenbaum's (2011) Bayesian model of rule-learning as a case study to spell out the underlying assumptions, and to confront them with the empirical results Frank and Tenenbaum (2011) propose to simulate, as well as with novel experiments. While rule-learning is arguably well suited to rational Bayesian approaches, I show that their models are neither psychologically plausible nor ideal observer models. Further, I show that their central assumption is unfounded: humans do not always preferentially learn more specific rules, but, at least in some situations, those rules that happen to be more salient. Even when granting the unsupported assumptions, I show that all of the experiments modeled by Frank and Tenenbaum (2011) either contradict their models, or have a large number of more plausible interpretations. I provide an alternative account of the experimental data based on simple psychological mechanisms, and show that this account both describes the data better, and is easier to falsify. I conclude that, despite the recent surge in Bayesian models of cognitive phenomena, psychological phenomena are best understood by developing and testing psychological theories rather than models that can be fit to virtually any data.