ABSTRACT
What is a paper on the Wason selection task (Wason, 1966) doing in a spe-
cial issue on the “new paradigm”? As we see it, an important goal of the new
paradigm is to examine our reasoning tasks from a different perspective, one
that shifts our focus from a relatively impoverished view of our participants
as either “correct” or “incorrect” and incorporates a more meaningful
toolbox of measures to understand their reasoning processes. A second key
objective for the new paradigm is to expand our theoretical analysis of rea-
soning performance by drawing from cognate literatures in cognitive sci-
ence. This paper aims to fulfil both of these goals with respect to one of our
most venerable and most studied reasoning tasks. Here the Wason selection task is employed not just as an object of study, but as a means to study
another set of processes. Specifically, our goal was to apply what we know
about how participants respond to the selection task as a means to study
metacognitive processes in reasoning. In particular we believe that metacog-
nitive processes we investigated in this paper will help to address one of the
most fundamental questions of reasoning, namely why people who have the
skills to do otherwise, nonetheless make decisions such as the following:
Presumably, the people who employ the fortunate/unfortunate executive
know that, in the long run, the good decisions depend on good reasoning
processes. This tendency to evaluate decisions based on the valence of the
outcome rather than the probability of success is known as “outcome bias”
(Baron & Hershey, 1988). From a common dual process theory perspective
(Evans, 2007; Kahneman, 2011; Stanovich, 2011), such decisions arise when
autonomous, Type 1 processes give rise to an initial response that may or
may not be subject to more deliberate Type 2 analysis (Stanovich & West, 2008). Strictly speaking this applies to a class of dual process theories known
as “default-interventionist” (Evans, 2007) as distinct from those which pro-
pose parallel operation of the two kinds of reasoning (e.g., Sloman, 1996). A
key issue for such theories is to identify the conditions which trigger inter-
vention with more effortful Type 2 reasoning on the default, intuitive
response (Evans, 2009; Stanovich, 2011; Thompson, Prowse Turner &
Pennycook, 2011; Thompson et al., 2013). We know already some of the
factors which influence this. For example, (a) there are individual differences in the ability and inclination to engage such processing (Stanovich, 1999,
2011), (b) limiting the time available for a decision decreases the probability
of analytic thinking (e.g., Evans & Curtis Holmes, 2005) as does concurrent
working memory load (De Neys, 2006) and (c) providing strong or elabo-
rated instructions to reason logically can increase analytic thinking and
reduce belief bias (e.g., Evans, Newstead, Allen, & Pollard, 1994; Newstead,
Pollard, Evans, & Allen, 1992). However, the fact remains that, for a given
individual operating under a constant set of constraints, there is still vari-
ability from decision to decision in terms of the probability that analytic pro-
cesses are engaged. How can this be explained?