Dissociation Between Judgments and Outcome-Expectancy Measures in Covariation Learning: A Signal Detection Theory Approach.

Abstract

A number of studies using trial-by-trial learning tasks have shown that judgments of covariation between a cue c and an outcome o deviate from normative metrics. Parameters based on trial-by-trial predictions were estimated from signal detection theory (SDT) in a standard causal learning task. Results showed that manipulations of P(c) when contingency (deltaP) was held constant did not affect participants' ability to predict the appearance of the outcome (d') but had a significant effect on response criterion (c) and numerical causal judgments. The association between criterion c and judgment was further demonstrated in 2 experiments in which the criterion was directly manipulated by linking payoffs to the predictive responses made by learners. In all cases, the more liberal the criterion c was, the higher judgments were. The results imply that the mechanisms underlying the elaboration of judgments and those involved in the elaboration of predictive responses are partially dissociable.

Evidence for an illusion of causality when using the Implicit Association Test to measure learning

“…In contrast to this dominant view, a recent study by Ratcliff and Nosek (2010) obtained support for a dual-process performance interpretation quite similar to the one proposed by Allan et al (2005) and Perales et al (2005) in the area of contingency learning. Unlike previous experiments on illusory correlation, in the study conducted by Ratliff and Nosek (2010) participants had to report their explicit attitudes towards the majority and the minority groups, and their attitudes also were measured by means of an Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998).…”

“…Although their theoretical interpretation of this finding is framed in different terms, it is surprisingly similar to the dual-process account provided by Allan et al (2005Allan et al ( , 2007 and Perales et al (2005) for cue-and outcome-density biases in contingency learning. In line with the popular dual-process framework proposed by Gawronski and Bodenhausen (2006), Ratliff and Nosek concluded that the fact that only explicit evaluations showed the illusory correlation effect suggests that the bias is not due to the way associations are encoded in memory (which is assumed to be indexed by the IAT), but to deliberate reasoning processes that qualify the expression of these associations (which is assumed to influence only explicit measures).…”

“…In other words, the outcome-density effect is assumed to be a performance phenomenon, not a learning phenomenon. Quite interestingly, Perales, Catena, Shanks, and González (2005) have proposed a similar explanation for a related bias that happens when it is the cue, instead of the outcome, that occurs very frequently, namely the cue-density effect (see also Matute, Yarritu, & Vadillo, 2011;Vadillo, Musca, Blanco, & Matute, 2011).…”

“…The purpose of the present experiment is to find out whether performance in the IAT is sensitive enough to measure the outcome-density effect that usually develops in the standard experimental paradigm used in contingency learning and causal judgment studies. On the basis of dual-process accounts such as those provided by Allan et al (2005Allan et al ( , 2007 and Perales et al (2005), one would not expect to observe outcome-density effects in the IAT because the effect is assumed to occur not during association formation (i.e., learning) but when making explicit contingency judgments (i.e., performance). An association formation model, on the other hand, does attribute the outcome-density effect to learning processes and therefore predicts an effect on any measure of learning, regardless of whether the measure requires an elaborated, explicit judgment or not.…”

“…Although the theoretical framework proposed by Allan et al (2005) and Perales et al (2005) is certainly inspiring and sheds new light on these effects, it is based on limited IAT and illusion of causality 6 evidence. Dissociations between dependent variables can sometimes reflect the action of independent cognitive systems.…”