Previous knowledge can induce an illusion of causality through actively biasing behavior.

Ion Yarritu,Helena Matute

doi:10.3389/fpsyg.2015.00389

Abstract

It is generally assumed that the way people assess the relationship between a cause and an outcome is closely related to the actual evidence existing about the co-occurrence of these events. However, people's estimations are often biased, and this usually translates into illusions of causality. Some have suggested that such illusions could be the result of previous knowledge-based expectations. In the present research we explored the role that previous knowledge has in the development of illusions of causality. We propose that previous knowledge influences the assessment of causality by influencing the decisions about responding or not (i.e., presence or absence of the potential cause), which biases the information people are exposed to, and this in turn produces illusions congruent with such biased information. In a non-contingent situation in which participants decided whether the potential cause was present or absent (Experiment 1), the influence of expectations on participants' judgments was mediated by the probability of occurrence of the potential cause (determined by participants' responses). However, in an identical situation, except that the participants were not allowed to decide the occurrence of the potential cause (Experiment 2), only the probability of the cause was significant, not the expectations or the interaction. Together, these results support our hypothesis that knowledge-based expectations affect the development of causal illusions by the mediation of behavior, which biases the information received.

Highlights

IntroductionExisting evidence questions the ability of humans to make accurate assessments of causality (e.g., Allan and Jenkins, 1983; Perales and Shanks, 2007; Hannah and Beneteau, 2009; Blanco et al, 2011; Matute et al, 2011)
Neither the main effect of the expectations nor the interaction were significant [largest F(1, 110) = 0.83]. These results provide evidence that supports the mediational hypothesis, as they suggest that when the effect of the previous expectations over participants’ behavior [p (C)] is controlled, their effect on the participants’ judgments disappears
The results support the hypothesis that people’s previous expectations about the causal relationship between two events affect their assessment of that relationship by modulating their behavioral patterns when they have the opportunity to decide about the presence, or absence, of the potential cause (Experiment 1)

Summary

Introduction

Existing evidence questions the ability of humans to make accurate assessments of causality (e.g., Allan and Jenkins, 1983; Perales and Shanks, 2007; Hannah and Beneteau, 2009; Blanco et al, 2011; Matute et al, 2011). Even though there are conditions in which people are perfectly capable of making precise estimations about causal relationships between events (e.g., Ward and Jenkins, 1965; Peterson, 1980; Shanks and Dickinson, 1987; Allan, 1993), there are well-established situations that invariably lead to a biased interpretation of the available evidence, resulting in systematic deviations from the objective assessment of causality (e.g., Jenkins and Ward, 1965; Allan and Jenkins, 1983; Blanco et al, 2009; Hannah and Beneteau, 2009). The concept of contingency has been regarded as the cornerstone of most of the models proposed to explain causal learning (e.g., Rescorla, 1968; Rescorla and Wagner, 1972; Cheng, 1997)

Objectives

Methods

Results

Conclusion