In their short abstract, Newell & Shanks (N&S) state: “Recommendations to ‘stop thinking’ and rely on ‘gut instincts’ reflect widely held beliefs that our decisions can be influenced by unconscious processes.” (N&S) predominantly focus on the second part of this phrase and show that there is no convincing evidence these processes are indeed unconscious. We take their argument one step further by addressing the first part of their phrase. That is, we discuss whether there is evidence that these decisions, presumably based on the unconscious, are superior to those based on thinking.
To determine whether presumed unconscious decisions are superior to conscious ones, we first need to define what constitutes a good decision. To this end, we use the distinction between compensatory and non-compensatory decisions. In compensatory decisions, options are compared on their probability weighted sum of all attributes, in which probability and attributes are evaluated objectively or subjectively (e.g., Tversky & Kahneman Reference Tversky and Kahneman1992). Non-compensatory decisions, however, are not based on a weighted sum of all attributes. For example, in Dawes' strategy (e.g., Bröder & Schiffer Reference Bröder and Schiffer2003a) decisions are based on the number of positive attributes, and in a lexicographic strategy (e.g., Tversky & Slovic Reference Tversky and Slovic1988), decisions are based on a sequential comparison of attributes, in which a decision is made if options differ sufficiently on an attribute under consideration. There seems to be general consensus that compensatory decisions are superior to non-compensatory ones, as all attributes are taken into account (yet see Payne et al. Reference Payne, Bettman and Johnson1988 for an interesting counterexample).
Dual-process theories (e.g., Kahneman Reference Kahneman2011; Stanovich & West Reference Stanovich and West2000) posit that non-deliberative processes often yield non-compensatory decisions, whereas deliberative processes generate compensatory ones. This hypothesis is supported by evidence showing that non-compensatory decisions are common in case of mental overload, which is assumed to hinder full reliance on the deliberative system (Bröder & Schiffer Reference Bröder and Schiffer2003b; cf. Pohl et al. Reference Pohl, Erdfelder, Hilbig, Liebke and Stahlberg2013). Interestingly, proponents of unconscious decision making argue the opposite: They state that the non-deliberative system facilitates compensatory decisions, whereas the deliberative system facilitates non-compensatory ones (e.g., Dijksterhuis et al. Reference Dijksterhuis, Bos, Nordgren and van Baaren2006b). In the following we determine whether the two decision-making paradigms discussed by (N&S), the Iowa Gambling Task (IGT) and the paradigm of Unconscious Thought Theory, the Unconscious Thought Paradigm (UTP), offer the possibility to test this alternative claim.
In the IGT, decision makers presumably relying on unconscious processes would opt for the two options (C & D) with the highest expected value (Bechara et al. Reference Bechara, Damasio, Damasio and Anderson1994); that is, they would use an objective compensatory strategy. However, IGT studies often do not allow for a test of this claim, as choices for specific options are generally not reported. The few studies that did include an analysis of specific options support a different conclusion (e.g., Duijvenvoorde et al. Reference Duijvenvoorde, Jansen, Visser and Huizenga2010; Horstmann et al. Reference Horstmann, Villringer and Neumann2012; Lin et al. Reference Lin, Song, Lin and Chiu2012). That is, decision makers generally prefer options with low probability of losses (B & D), and some, but certainly not all, decision makers gradually develop a preference of D (low losses, low gains) over B (high losses, high gains) (Huizenga et al. Reference Huizenga, Crone and Jansen2007). It is not very likely that the latter decision makers adopted a compensatory strategy, as they did not prefer both optimal options (C & D). It is more likely that these decision makers adopted a non-compensatory lexicographic strategy, in which they first considered probability of losses and then losses themselves. These findings show that in the IGT, participants using non-compensatory and compensatory strategies may arrive at similar decisions. We therefore conclude that the IGT is not suitable to differentiate decision strategies.
According to Unconscious-Thought Theory (Dijksterhuis et al. Reference Dijksterhuis, Bos, Nordgren and van Baaren2006b), decision makers who presumably rely on unconscious processes would prefer the option with the highest compensatory value over all attributes. However, using importance ratings of attributes, it was shown that the compensatory strategy, Dawes's strategy, and a lexicographic strategy all converged on the same choice (Huizenga et al. Reference Huizenga, Wetzels, Van Ravenzwaaij and Wagenmakers2012). Therefore we conclude that the UTP also does not allow for a differentiation of decision strategies.
The evidence reviewed above suggests that the IGT and UTP are not suited to identify decision strategies and therefore are not suited to test whether presumably unconscious decision processes facilitate compensatory decision making. To test this claim of compensatory decision making, the field requires new paradigms that allow assessment of decision strategies, namely, paradigms in which compensatory and non-compensatory strategies result in different decisions. Fortunately, both within, as well as outside, the IGT and UTP literature, paradigms are being developed that suit this purpose. In the IGT-related field there exists a paradigm that allows a further study of lexicographic versus compensatory strategies (Lin et al. Reference Lin, Chiu and Huang2009). In the UTP literature, there are paradigms to delineate Dawes and compensatory strategies (Payne et al. Reference Payne, Samper, Bettman and Luce2008; Usher et al. Reference Usher, Russo, Weyers, Brauner and Zakay2011) and to delineate lexicographic and compensatory strategies (Huizenga et al. Reference Huizenga, Wetzels, Van Ravenzwaaij and Wagenmakers2012). Outside these fields it has been shown that process-tracing techniques (Bröder & Schiffer Reference Bröder and Schiffer2003a; Payne et al. Reference Payne, Bettman and Johnson1988) provide valuable tools to study decision strategies. In addition, it has been shown that modern statistical techniques like mixture analyses offer the possibility to differentiate decision strategies (Duijvenvoorde et al. Reference Duijvenvoorde, Jansen, Visser and Huizenga2010; Jansen et al. Reference Jansen, Van Duijvenvoorde and Huizenga2012).
To conclude, the evidence in favor of the superiority of unconscious decisions is not convincing, as paradigms like the IGT and UTP do not allow for an assessment of decision strategies. However, there do exist new paradigms, experimental approaches, and statistical techniques that provide a detailed assessment of decision strategies and therefore allow for a more convincing test of the superiority of – presumed– unconscious processes.
In their short abstract, Newell & Shanks (N&S) state: “Recommendations to ‘stop thinking’ and rely on ‘gut instincts’ reflect widely held beliefs that our decisions can be influenced by unconscious processes.” (N&S) predominantly focus on the second part of this phrase and show that there is no convincing evidence these processes are indeed unconscious. We take their argument one step further by addressing the first part of their phrase. That is, we discuss whether there is evidence that these decisions, presumably based on the unconscious, are superior to those based on thinking.
To determine whether presumed unconscious decisions are superior to conscious ones, we first need to define what constitutes a good decision. To this end, we use the distinction between compensatory and non-compensatory decisions. In compensatory decisions, options are compared on their probability weighted sum of all attributes, in which probability and attributes are evaluated objectively or subjectively (e.g., Tversky & Kahneman Reference Tversky and Kahneman1992). Non-compensatory decisions, however, are not based on a weighted sum of all attributes. For example, in Dawes' strategy (e.g., Bröder & Schiffer Reference Bröder and Schiffer2003a) decisions are based on the number of positive attributes, and in a lexicographic strategy (e.g., Tversky & Slovic Reference Tversky and Slovic1988), decisions are based on a sequential comparison of attributes, in which a decision is made if options differ sufficiently on an attribute under consideration. There seems to be general consensus that compensatory decisions are superior to non-compensatory ones, as all attributes are taken into account (yet see Payne et al. Reference Payne, Bettman and Johnson1988 for an interesting counterexample).
Dual-process theories (e.g., Kahneman Reference Kahneman2011; Stanovich & West Reference Stanovich and West2000) posit that non-deliberative processes often yield non-compensatory decisions, whereas deliberative processes generate compensatory ones. This hypothesis is supported by evidence showing that non-compensatory decisions are common in case of mental overload, which is assumed to hinder full reliance on the deliberative system (Bröder & Schiffer Reference Bröder and Schiffer2003b; cf. Pohl et al. Reference Pohl, Erdfelder, Hilbig, Liebke and Stahlberg2013). Interestingly, proponents of unconscious decision making argue the opposite: They state that the non-deliberative system facilitates compensatory decisions, whereas the deliberative system facilitates non-compensatory ones (e.g., Dijksterhuis et al. Reference Dijksterhuis, Bos, Nordgren and van Baaren2006b). In the following we determine whether the two decision-making paradigms discussed by (N&S), the Iowa Gambling Task (IGT) and the paradigm of Unconscious Thought Theory, the Unconscious Thought Paradigm (UTP), offer the possibility to test this alternative claim.
In the IGT, decision makers presumably relying on unconscious processes would opt for the two options (C & D) with the highest expected value (Bechara et al. Reference Bechara, Damasio, Damasio and Anderson1994); that is, they would use an objective compensatory strategy. However, IGT studies often do not allow for a test of this claim, as choices for specific options are generally not reported. The few studies that did include an analysis of specific options support a different conclusion (e.g., Duijvenvoorde et al. Reference Duijvenvoorde, Jansen, Visser and Huizenga2010; Horstmann et al. Reference Horstmann, Villringer and Neumann2012; Lin et al. Reference Lin, Song, Lin and Chiu2012). That is, decision makers generally prefer options with low probability of losses (B & D), and some, but certainly not all, decision makers gradually develop a preference of D (low losses, low gains) over B (high losses, high gains) (Huizenga et al. Reference Huizenga, Crone and Jansen2007). It is not very likely that the latter decision makers adopted a compensatory strategy, as they did not prefer both optimal options (C & D). It is more likely that these decision makers adopted a non-compensatory lexicographic strategy, in which they first considered probability of losses and then losses themselves. These findings show that in the IGT, participants using non-compensatory and compensatory strategies may arrive at similar decisions. We therefore conclude that the IGT is not suitable to differentiate decision strategies.
According to Unconscious-Thought Theory (Dijksterhuis et al. Reference Dijksterhuis, Bos, Nordgren and van Baaren2006b), decision makers who presumably rely on unconscious processes would prefer the option with the highest compensatory value over all attributes. However, using importance ratings of attributes, it was shown that the compensatory strategy, Dawes's strategy, and a lexicographic strategy all converged on the same choice (Huizenga et al. Reference Huizenga, Wetzels, Van Ravenzwaaij and Wagenmakers2012). Therefore we conclude that the UTP also does not allow for a differentiation of decision strategies.
The evidence reviewed above suggests that the IGT and UTP are not suited to identify decision strategies and therefore are not suited to test whether presumably unconscious decision processes facilitate compensatory decision making. To test this claim of compensatory decision making, the field requires new paradigms that allow assessment of decision strategies, namely, paradigms in which compensatory and non-compensatory strategies result in different decisions. Fortunately, both within, as well as outside, the IGT and UTP literature, paradigms are being developed that suit this purpose. In the IGT-related field there exists a paradigm that allows a further study of lexicographic versus compensatory strategies (Lin et al. Reference Lin, Chiu and Huang2009). In the UTP literature, there are paradigms to delineate Dawes and compensatory strategies (Payne et al. Reference Payne, Samper, Bettman and Luce2008; Usher et al. Reference Usher, Russo, Weyers, Brauner and Zakay2011) and to delineate lexicographic and compensatory strategies (Huizenga et al. Reference Huizenga, Wetzels, Van Ravenzwaaij and Wagenmakers2012). Outside these fields it has been shown that process-tracing techniques (Bröder & Schiffer Reference Bröder and Schiffer2003a; Payne et al. Reference Payne, Bettman and Johnson1988) provide valuable tools to study decision strategies. In addition, it has been shown that modern statistical techniques like mixture analyses offer the possibility to differentiate decision strategies (Duijvenvoorde et al. Reference Duijvenvoorde, Jansen, Visser and Huizenga2010; Jansen et al. Reference Jansen, Van Duijvenvoorde and Huizenga2012).
To conclude, the evidence in favor of the superiority of unconscious decisions is not convincing, as paradigms like the IGT and UTP do not allow for an assessment of decision strategies. However, there do exist new paradigms, experimental approaches, and statistical techniques that provide a detailed assessment of decision strategies and therefore allow for a more convincing test of the superiority of – presumed– unconscious processes.