A process-dissociation examination of the cognitive processes underlying unconscious thought

Abstract

Conscious and unconscious thought have been previously found to differentially impact decision-making quality. However, little research has directly measured the processes underlying these modes of thinking. We propose that both thinking modes are characterized by rule-based and intuitive processing. In two experiments, we used the Process Dissociation Procedure to independently measure these cognitive processes. We tested three competing hypotheses: (a) conscious thinking evokes both increased rule-based and decreased intuitive processing compared to unconscious thinking; (b) conscious and unconscious thinking evoke similar levels of intuitive processing but conscious thinking enhances rule-based processing; and (c) conscious and unconscious thinking evoke similar levels of rule-based processing but unconscious thinking enhances intuitive processing. Experiment 1 used base-rate and law-of-large-numbers decision-making problems, whereas Experiment 2 used decision-making problems similar to the “apartment” problem that is often used in unconscious thought studies. In both experiments we found support for hypothesis (b).

Introduction

Research has shown that changes in attention allocation—the extent to which we devote our conscious awareness to the task at hand—may impact the quality of our decisions (e.g., Dijksterhuis, 2004, Wilson and Schooler, 1991). When tasks require the application of rules, deliberation with attention (conscious thought) leads to better decisions. In contrast, when tasks require the holistic integration of a large amount of information that is not rule-bound, deliberation without attention (unconscious thought) leads to better decisions. This finding is referred to as the Unconscious Thought Effect (Dijksterhuis, 2004, Dijksterhuis and Nordgren, 2006). Finally, when tasks require both the application of a rule and the holistic integration of information, a combination of conscious and unconscious thought leads to better decisions (Nordgren, Bos, & Dijksterhuis, 2011). Although these effects have been supported by a large number of studies, little research has investigated the cognitive processes that underlie conscious and unconscious thought. What happens during deliberation with or without attention at a cognitive level?

In explaining these effects, researchers have focused on the roles of rule-based and intuitive processes.¹ Rule-based processing is a top-down process that affects decisions via the application of rules that define the determinant features of quality. In contrast, intuitive processing is a bottom-up process that affects decisions via the holistic integration of information about different options. Not surprisingly, rule-based processing is believed to facilitate decisions that require the application of a rule (Dijksterhuis & Nordgren, 2006). In contrast, intuitive processing has been argued to be critical for making decisions that require the integration of many features of different options (Dijksterhuis, 2004, Dijksterhuis and Nordgren, 2006). Given that conscious thinking improves performance on rule-based decisions and that unconscious thinking improves performance on holistic decisions, the expectation follows that conscious thinking evokes rule-based processing, whereas unconscious thinking evokes intuitive processing (e.g., Dijksterhuis & Nordgren, 2006).

One complication with this account of decision quality under conscious and unconscious thought is that experimental tasks are not process-pure. Rather, any given task reflects the operation of multiple component processes (e.g., Ferreira et al., 2006, Jacoby, 1991). As such, instructions to think consciously or unconsciously likely evoke both rule-based and intuitive thought processes. Thus, conscious thinking may both enhance rule-based processing and diminish intuitive processing compared to unconscious thinking. If rule-based processing improves rule-based decisions, whereas intuitive processing hinders the effective use of rules, then effects on either processing component may explain why conscious thinking produces superior decisions on rule-based tasks. That is, conscious thinking may promote quality rule-based decisions by increasing rule-based processing, decreasing intuitive processing, or both, compared to unconscious thinking. Likewise, if intuitive processing improves decisions requiring the integration of information, whereas rule-based processing interferes with such decisions, then effects on either processing component may explain why unconscious thinking produces superior decisions on integrative tasks. That is, unconscious thinking may promote quality integrative decisions by increasing intuitive processing, decreasing rule-based processing, or both, compared to conscious thinking.

There is precedent for expecting that rule-based processing may interfere with integrative decisions. Applying rules in the context of integrative decisions, for which a large number of attributes must be integrated, could lead to mental fixation on a limited number of attributes. In turn, this mental fixation could cause people to overlook other relevant attributes and lead to inferior decisions. In other words, by using rules, people may end up focusing on the “wrong” aspects of the problem, or on an incomplete set of information, which may cause them to make poor decisions. As a result, conscious thought would be detrimental when making integrative decisions (Schooler and Melcher, 1995, Schooler et al., 1993, Wilson et al., 1993, Wilson and Schooler, 1991). Similarly, there is precedent for expecting that intuitive processing may interfere with rule-bound decisions, as in the case of almost all the standard heuristics. The base-rate and law-of-large-numbers problems are examples of cases in which intuition leads to poor decisions, as are most of the examples of representativeness and availability (Kahneman & Tversky, 1972).

Of course, it also is possible that conscious and unconscious thinking do not differ in the extent to which they evoke rule-based and/or intuitive processing. For example, intuitive processing may not differ between the conditions, and variation in decision quality may be related solely to differences in the extent of rule-based processes. Alternatively, rule-based processing may not differ between the conditions, and variation in decision quality may be related solely to differences in the extent of intuitive processing.

Because the extent of rule-based and intuitive thinking cannot be inferred from the manipulation of conscious and unconscious thinking, these competing possibilities cannot be examined within the standard methods of studying conscious and unconscious thought. To exemplify this point, it is instructive to consider previous studies that have examined the processes underlying conscious and unconscious thinking.

Dijksterhuis (2004) showed that, compared to conscious thought, unconscious thought leads to more polarization and clustering of information. That is, when considering decision alternatives unconsciously (as opposed to consciously), positive alternatives become more positive and negative alternatives become more negative over time, and pieces of information that load on the same dimension become more clustered. These findings seem to suggest that unconscious thought leads to more “intuitive processing,” a type of holistic processing that favors a better organization of large amounts of information. But this is not the only possibility. It is also possible, for instance, that unconscious thought (as opposed to conscious thought) leads to better organization due to decreased rule-based processing. According to previous research (Wilson & Schooler, 1991), rule-based processing may hinder a good and efficient organization of large amounts of information, because it focuses people too much on a limited number of rules. Thus, although this research showed important outcomes of unconscious thought, it did not independently measure the underlying processes.

Further characterizing the outcomes of unconscious thinking, Bos, Dijksterhuis, and van Baaren (2011) showed that unconscious thought, as opposed to immediate decision-making, leads to an automatic weighting process in which important decision attributes receive more weight and unimportant decision attributes receive less weight. However, the same problem described earlier applies: evidence for effective weighting following a period of unconscious thought does not unequivocally support the idea that unconscious thought is characterized by more intuitive processing, as opposed to less rule-based processing. Better weighting under unconscious thought may result from increased intuitive processing, decreased rule-based processing, or both.

Another attempt to identify the underlying processes of conscious and unconscious thought was made by Usher, Russo, Weyers, Brauner, and Zakay (2011), who showed that deliberation without attention (i.e., the distraction manipulation that is regularly used to elicit unconscious thought) has the same effect on decision-making as instructing participants to “use their intuition or gut feeling.” To be precise, participants thinking unconsciously and participants instructed to “use their intuition” made better decisions than participants who thought consciously. The authors conclude that unconscious thought is characterized by increased intuitive processing, but in so doing, they assume that telling people to use their intuition has only one outcome; that is, to increase people's intuitive processing. However, it also is possible that both unconscious thought and the instruction to use one's intuition lead to decreased rule-based processing, and this might be the process driving the effects.

Although the above findings have undoubtedly enriched our understanding of unconscious thought and its effects on decision-making, they do not reveal the extent to which manipulations of conscious and unconscious thinking evoke rule-based and intuitive thinking. To address this issue directly, we need a means to derive independent estimates of rule-based and intuitive processing following the imposition of conscious and unconscious thought. This was the goal of the present research.

One technique for measuring the operation of multiple processes within a single task is the Process Dissociation (PD) procedure developed by Jacoby (1991; Jacoby and Dallas, 1981, Jacoby et al., 1993). Though originally developed for estimating the contributions of recollection and familiarity to memory performance, a variation of the PD procedure has been developed to estimate the contributions of rule-based and intuitive processes to decision making (Ferreira et al., 2006). The fundamental logic of PD is to design experiments that include both congruent conditions, in which the two processes of interest act in concert, and incongruent conditions, in which the two processes oppose each other. In the PD procedure that Ferreira et al. (2006) developed for use in the context of decision-making, a congruent problem is one in which both rule-based processing and intuitive processing lead to the same correct answer. In this case, the probability of a correct response is the probability of using rule-based processing (RB), plus the probability of using intuitive processing (I) when the use of rules fails, I(l − RB):

$$\text{Congruent}=\text{RB}+\text{I}\left(\text{l}-\text{RB}\right)\text{.}$$

In contrast, an incongruent problem is one in which rule-based processing predicts the correct answer and intuitive processing predicts the wrong answer. In this case, the probability of an incorrect response is the probability of using intuitive processing when the use of the rule fails:

$$\text{Incongruent}=\text{I}\left(\text{l}-\text{RB}\right)\text{.}$$

Given these two equations, we can now estimate intuitive and rule-based processing independently.

$$\text{RB}=\text{Congruent}–\text{Incongruent}$$

$$\text{I}=\text{Incongruent}/\left(1–\text{RB}\right)\text{.}$$

This procedure was validated and applied across four experiments by Ferreira et al. (2006). The authors showed that rule-based and intuitive processing are independent cognitive processes that may be separately enhanced through various manipulations in the context of decision-making. For example, dividing attention while participants made decisions reduced estimates of rule-based processing, but did not affect intuitive processing. However, encouraging participants to use intuitive processing during practice trials increased intuitive processing during test trials, but did not affect rule-based processing.

Similar dissociations have been observed when using PD in the context of human memory and implicit attitudes. For example, studies have shown that a controlled use of memory (similar to rule-based processing) is strongly affected by variables such as divided attention, depth of processing, speeded responding, and aging. In contrast, automatic influences of memory (similar to intuitive processing) tend to be affected by variables such as perceptual fluency, conceptual fluency, and habitual responses (for reviews, see Payne, 2008, Yonelinas, 2002).

The present research used the PD procedure to directly and independently measure the extent to which rule-based and intuitive processing characterize conscious and unconscious thought in the context of decision-making. Given the existing assumptions regarding the processes underlying conscious and unconscious thought (e.g., Dijksterhuis, 2004, Wilson and Schooler, 1991), three competing hypotheses were identified. First, it may be that conscious thinking evokes both increased rule-based and decreased intuitive processing compared to unconscious thinking. Second, it may be that conscious and unconscious thinking evoke similar levels of intuitive processing but that conscious thinking enhances rule-based processing. Finally, it may be that conscious and unconscious thinking evoke similar levels of rule-based processing but that unconscious thinking enhances intuitive processing.

As mentioned previously, the PD procedure requires the use of complex decision-making problems in which intuitive and rule-based processes are working either in concert or in opposition. In other words, rules are presented in both the unconscious and conscious thought conditions, adding to the complexity of the problems. In such cases, unconscious versus conscious thought does not produce better decision-making (in fact, a combination of the two appears to lead to optimal decisions; see Nordgren et al., 2011). We therefore did not predict that unconscious thought would lead to better decision-making. If anything, because correct answers were defined in terms of the use of rules, conscious thought might be expected to lead to better decisions. However, this does not affect our ability to measure the extent to which conscious and unconscious thought elicit rule-based and intuitive processing, the goal of our research.