Thinking about thinking: implications of the introspective error for default-interventionist type models of dual processes

Imagine sitting in a grand hall, listening to the keynote lecture of the conference you are attending. At some point your thoughts drift-off. When you look around you, half of the audience is staring intently at their smartphones. You ask yourself: what is it about this talk that makes you unable to stay focused? Do you find any aspect enjoyable? How would you behave, if it were you standing at the podium? Answering questions such as these, you are engaging in a process known as introspection. Introspection describes the ability to explicitly characterize experience. It enables one to say: “I am thinking about what I am thinking about.” In other words, introspection allows one to become meta-aware, that is, to have awareness of what one believes to be experiencing. Although agreement exists as to the fact that we all have and make experiences and therefore subjective experience seems indisputable (Schooler and Schreiber, 2004; Winkielman and Schooler, 2011), empirically gaining access to and knowledge of this subjective experience poses a great challenge. It requires us to put a subjective, internal experience into words, such as in the above-mentioned example. This raises the question, if the words we come up with are true descriptions of our experience, or confabulations. Specifically, the dissociation between experiential consciousness (the contents of experience) and meta-consciousness (the belief about the contents of experience) makes us fallible in appraising our own experiences. In some cases, this fallibility has been demonstrated to manifest in translational dissociations, that is, the distortion of experience in an attempt to recount or characterize it; this was termed the “introspective error” (Schooler, 2002). Even though—or perhaps because—the paradox of introspection has been studied extensively for a number of decades, it is almost paradoxical itself to find that the resulting implications for the ongoing debate about (dual-) process types in judgment and decision-making (JDM) and specifically for the most widely accepted and experimentally investigated defaultinterventionist model (D-I-Model), have not been considered thus far. In the present contribution we set out to fill this gap and point out the implications of the introspective error for the conceptualization of the D-I-Model. In the (neuro-) scientific community, dual-process models of intuitive and deliberate JDM currently constitute the preferred theoretical construct (e.g., Lieberman et al., 2002; Strack and Deutsch, 2004; Glöckner and Witteman, 2010; Kahneman, 2011; Evans and Stanovich, 2013). These models have been built on the assumption that judgments are formed via two qualitatively distinct process types: automatic “intuition” and controlled “deliberation” or “reflection.” In recent years, an immense influx of publications has arisen, either fervently defending or criticizing the dualistic distinction between rapid, autonomous, intuitive processes and slower, thoughtful, reasoning processes of higher order. In their most recent publication on dual-process models, Evans and Stanovich (2013)—henceforth referred to as E&S— described their concept as one, which assumes that automatic processes (Type 1, T1) yield default responses unless an intervention by higher order reasoning processes (Type 2, T2) is needed; a model, which has been termed the D-IModel. We will focus on this current-most description of dual-process theory, since it constitutes the predominant model being intensively discussed by leading authorities in the field1. E&S split the attributes of both process types into defining (necessary/sufficient) and correlated features. The defining features listed for T2 processing are working memory capacity and cognitive decoupling. These are seen to be central in order to be able to reason hypothetically and distinguish supposition from belief, thereby aiding “rational choices by running thought experiments” (E&S, p. 236). Importantly, cognitive decoupling requires a re-representation of automatic (T1) processes so as to be able to interfere with their output. In that way T2 processing allows for “metarepresentational and simulation abilities,” and is thus a form of meta-consciousness. T1 processing, in contrast, is defined as encompassing both “innately specified processing modules or procedures and experiential associations that have been learned to the point of automaticity” (E&S, p. 236). Explicitly opposing