Although we agree that the proposed model is innovative and coherent, and that it undoubtedly contributes to the field, its further development may require addressing two problems regarding the switching mechanism. The first problem is that the model explains switching only in a particular type of situation: When two or more competing intuitions are available for system 1, or there are no intuitions at all. The model, however, does not explain the spontaneous activation of system 2 in situations when the “alleged system 1 response” is automatic, but the “alleged system 2 response” cannot be available for system 1 without system 2 activation because it must be first produced by the deliberation or calculation of system 2. For example, in the commonly known cognitive reflective task (CRT) bat and ball task, the answer “$0.10” can be available to system 1 because $1.10 − $1.00 is a very easy equation that gets calculated automatically. However, to find the correct answer, more complex calculations are required: finding x if (a) x + y = $1.10 and (b) y = $1.00 + x. It is unlikely that this response can be automatically available to system 1.
De Neys's answer to this problem is the non-exclusivity assumption and the idea that system 1 can produce the correct answer itself. However, if system 1 can conduct such complex operations as the calculations presented above, then why do we even need system 2? De Neys further proposes that the correct intuition may be available to system 1, not thanks to its ability to conduct the calculations, but because of previous exposure to this riddle (or similar riddles) and learning the correct answer. However, by applying such reasoning, the model still does not explain switching when one encounters a completely new problem that requires calculating the correct answer first. It does not even explain switching in the case of a well-known problem presented with different numbers (even if the schema of a riddle is the same, but the numbers are changed, the response still needs to be calculated from scratch). Therefore, non-exclusivity does not fully resolve the problem of switching paradoxes.
Moreover, the proposed switching mechanism does not explain various manipulations that trigger system 2. For example, system 2 deliberation may be turned on by priming (Gervais & Norenzayan, Reference Gervais and Norenzayan2012), presenting questions in a way that makes them difficult to read (Alter, Oppenheimer, Epley, & Eyre, Reference Alter, Oppenheimer, Epley and Eyre2007; Song & Schwarz, Reference Song and Schwarz2008), asking participants to frown during the study (Alter et al., Reference Alter, Oppenheimer, Epley and Eyre2007), explicitly asking participants to deliberate on the questions before answering, and other interventions (see Horstmann, Hausmann, & Ryf [Reference Horstmann, Hausmann, Ryf, Glöckner and Witteman2009] for an overview). None of these effects can be explained by means of the proposed model.
The second problem is that the switching mechanism seems not to work in a considerable number of situations. First, a similar activation of competing intuitions should trigger system 2, but this does not always happen. For example, in logical riddles (e.g., syllogisms, base-rate problems) or moral dilemmas, we simultaneously present several alternative answers from which the participant may choose. Therefore, all alternatives should be equally strongly activated and trigger system 2. However, an explicit presentation of alternative responses does not make people more reflective or may even enhance more intuitive processing (Sirota & Juanchich, Reference Sirota and Juanchich2018).
Furthermore, the proposed feedback loop seems fallible. If system 2 deliberation leads to choosing an answer, then it should decrease the relative activation of the rejected intuition. For example, when one solves base-rate problems or assesses probability or randomness (e.g., what is more probable, six heads in a row or head–tail–tail–head–tail–head), deliberation lets them use the probability distributions to find the right answer. However, even though deliberation and formal knowledge allow giving the correct answer with high certainty, the “homunculus” keeps jumping and shouting the intuitive response (Gould, Reference Gould1992, p. 469; see also Kahneman [Reference Kahneman2011] and Thompson [Reference Thompson, Evans and Frankish2009] for discussions on the subjective feeling that the intuitive answer is correct). This suggests that the relative activation of the initial, intuitive response is still very high.
The solution to these problems is either to clarify the nature of competing intuitions or to propose a more all-encompassing switching trigger. Regarding the first possibility, the competing intuitions should not be pictured as alternative responses (p vs. q) but rather as either an intuitive response and its negation (p vs. not-p) or as two alternative ways of solving the problem (e.g., the simple equation $1.10 −$1.00 vs. applying formal algebra). With competing intuitions defined this way, the non-exclusivity assumption holds: System 1 is still able to generate the alternative intuition. It is impossible for system 1 to have access to certain responses (e.g., based on advanced calculations), but it may have intuitions that the initial response is incorrect (i.e., not-p intuition) or that there are different ways to approach the problem. To put it differently, it is impossible for system 1 to do complex calculations, however, people can still automatise the reaction of suspiciousness in response to logical riddles or the belief that it is better to rely on formal algebra than on a gut feeling.
Nevertheless, even after clarifying the nature of intuitions, the model still does not explain why manipulations such as priming, frowning, explicit instructions, or difficult-to-read font trigger system 2. Therefore, we suggest the incorporation of a more universal switching mechanism into the model. It is not a new idea to identify processing disfluency as the ultimate intuitive trigger of system 2 (e.g., Alter et al., Reference Alter, Oppenheimer, Epley and Eyre2007; Thompson, Reference Thompson, Evans and Frankish2009). The general idea behind disfluency is that system 2 is activated if processing within system 1 does not go smoothly. Including a disfluency-triggered switch does not rule out the possibility of the switching being caused by uncertainty, as disfluency should lead to uncertainty (Gill, Swann, & Silvera, Reference Gill, Swann and Silvera1998). However, this modification will allow the model to explain a wider range of phenomena: Disfluency may be caused by difficult-to-read font, the instruction to think twice before answering, a lack of faith in the intuitive answer, and so on.
You have
Access
Although we agree that the proposed model is innovative and coherent, and that it undoubtedly contributes to the field, its further development may require addressing two problems regarding the switching mechanism. The first problem is that the model explains switching only in a particular type of situation: When two or more competing intuitions are available for system 1, or there are no intuitions at all. The model, however, does not explain the spontaneous activation of system 2 in situations when the “alleged system 1 response” is automatic, but the “alleged system 2 response” cannot be available for system 1 without system 2 activation because it must be first produced by the deliberation or calculation of system 2. For example, in the commonly known cognitive reflective task (CRT) bat and ball task, the answer “$0.10” can be available to system 1 because $1.10 − $1.00 is a very easy equation that gets calculated automatically. However, to find the correct answer, more complex calculations are required: finding x if (a) x + y = $1.10 and (b) y = $1.00 + x. It is unlikely that this response can be automatically available to system 1.
De Neys's answer to this problem is the non-exclusivity assumption and the idea that system 1 can produce the correct answer itself. However, if system 1 can conduct such complex operations as the calculations presented above, then why do we even need system 2? De Neys further proposes that the correct intuition may be available to system 1, not thanks to its ability to conduct the calculations, but because of previous exposure to this riddle (or similar riddles) and learning the correct answer. However, by applying such reasoning, the model still does not explain switching when one encounters a completely new problem that requires calculating the correct answer first. It does not even explain switching in the case of a well-known problem presented with different numbers (even if the schema of a riddle is the same, but the numbers are changed, the response still needs to be calculated from scratch). Therefore, non-exclusivity does not fully resolve the problem of switching paradoxes.
Moreover, the proposed switching mechanism does not explain various manipulations that trigger system 2. For example, system 2 deliberation may be turned on by priming (Gervais & Norenzayan, Reference Gervais and Norenzayan2012), presenting questions in a way that makes them difficult to read (Alter, Oppenheimer, Epley, & Eyre, Reference Alter, Oppenheimer, Epley and Eyre2007; Song & Schwarz, Reference Song and Schwarz2008), asking participants to frown during the study (Alter et al., Reference Alter, Oppenheimer, Epley and Eyre2007), explicitly asking participants to deliberate on the questions before answering, and other interventions (see Horstmann, Hausmann, & Ryf [Reference Horstmann, Hausmann, Ryf, Glöckner and Witteman2009] for an overview). None of these effects can be explained by means of the proposed model.
The second problem is that the switching mechanism seems not to work in a considerable number of situations. First, a similar activation of competing intuitions should trigger system 2, but this does not always happen. For example, in logical riddles (e.g., syllogisms, base-rate problems) or moral dilemmas, we simultaneously present several alternative answers from which the participant may choose. Therefore, all alternatives should be equally strongly activated and trigger system 2. However, an explicit presentation of alternative responses does not make people more reflective or may even enhance more intuitive processing (Sirota & Juanchich, Reference Sirota and Juanchich2018).
Furthermore, the proposed feedback loop seems fallible. If system 2 deliberation leads to choosing an answer, then it should decrease the relative activation of the rejected intuition. For example, when one solves base-rate problems or assesses probability or randomness (e.g., what is more probable, six heads in a row or head–tail–tail–head–tail–head), deliberation lets them use the probability distributions to find the right answer. However, even though deliberation and formal knowledge allow giving the correct answer with high certainty, the “homunculus” keeps jumping and shouting the intuitive response (Gould, Reference Gould1992, p. 469; see also Kahneman [Reference Kahneman2011] and Thompson [Reference Thompson, Evans and Frankish2009] for discussions on the subjective feeling that the intuitive answer is correct). This suggests that the relative activation of the initial, intuitive response is still very high.
The solution to these problems is either to clarify the nature of competing intuitions or to propose a more all-encompassing switching trigger. Regarding the first possibility, the competing intuitions should not be pictured as alternative responses (p vs. q) but rather as either an intuitive response and its negation (p vs. not-p) or as two alternative ways of solving the problem (e.g., the simple equation $1.10 −$1.00 vs. applying formal algebra). With competing intuitions defined this way, the non-exclusivity assumption holds: System 1 is still able to generate the alternative intuition. It is impossible for system 1 to have access to certain responses (e.g., based on advanced calculations), but it may have intuitions that the initial response is incorrect (i.e., not-p intuition) or that there are different ways to approach the problem. To put it differently, it is impossible for system 1 to do complex calculations, however, people can still automatise the reaction of suspiciousness in response to logical riddles or the belief that it is better to rely on formal algebra than on a gut feeling.
Nevertheless, even after clarifying the nature of intuitions, the model still does not explain why manipulations such as priming, frowning, explicit instructions, or difficult-to-read font trigger system 2. Therefore, we suggest the incorporation of a more universal switching mechanism into the model. It is not a new idea to identify processing disfluency as the ultimate intuitive trigger of system 2 (e.g., Alter et al., Reference Alter, Oppenheimer, Epley and Eyre2007; Thompson, Reference Thompson, Evans and Frankish2009). The general idea behind disfluency is that system 2 is activated if processing within system 1 does not go smoothly. Including a disfluency-triggered switch does not rule out the possibility of the switching being caused by uncertainty, as disfluency should lead to uncertainty (Gill, Swann, & Silvera, Reference Gill, Swann and Silvera1998). However, this modification will allow the model to explain a wider range of phenomena: Disfluency may be caused by difficult-to-read font, the instruction to think twice before answering, a lack of faith in the intuitive answer, and so on.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interest
None.