The advantages of thinking in the mode of System 2 are not hard to appreciate: this mode is characterized as deliberative, logical, and critical thinking. This stands in contrast to thinking faster in the mode of System 1 which is associated with intuitive, automatic and associative thinking. The thinking style of System 2 is more effortful than System 1’s but it allows more time for reflection on a judgement or decision problem, and therefore one is less prone to jump to wrong conclusions. But is it necessarily beneficial to instruct or encourage individuals to think more slowly than they would normally do? The answer may be a bit more complicated and not straightforward; it seems to require some careful qualifications.

Lawson, Larrick, and Soll (2020) were intrigued about how and when it really helps individuals to intentionally think more slowly, taking more time to think than they usually do. The researchers studied the implications of fast and slow thinking through a sophisticated multi-layered experiment that they constructed [1]. They raise a crucial point: thinking more slowly while reflecting on a problem may be useless if one lacks familiarity with and does not have access to the rules that should help him or her to solve it correctly. Thinking more slowly is not a solution by itself.

Lawson and colleagues devised twelve (12) judgement or decision-making (JDM) problems, two problems in each of six categories, all related to statistical inferences (this choice of problem domain is of significance and we will return to it later). They instructed participants to take their time, at their discretion (control) to give their best answer to each problem posed to them; to take more time than they would normally do to think about the problem more carefully, actively reflecting on it, before answering (Slow); or to think more quickly, thus giving the initial answer that intuitively comes to mind (Fast) {between-subjects, BS}. Yet, an additional group was asked to give initially an intuitive answer they come up with, and then take more time to reflect on the problem more deliberately before answering again {within-subjects, WS}.

Optimising and rational decision rules that humans are more likely to employ in a type of System 2 thinking process resemble a more complicated model that utilises more data (e.g., factors or attributes). They are more expensive in time and cognitive effort one needs to invest. Simpler heuristic rules often recalled in System 1 type of thinking can provide close approximations that are useful and adequate. But thinking in the mode of System 2 should not necessarily imply the recruitment of more complicated rules and computations; efficient and practical rules may be recalled in System 2 (deliberative) as well as in System 1 (intuitive) types of thinking. Lawson and colleagues point out that the desired outcome, particularly when thinking more thoroughly, should be in choosing the more relevant and suitable rule for solving a problem one faces.

• The researchers composed JDM problems which answering them correctly relies on the application of canonical statistical rules (i.e., general, fundamental rules, at level of principle, e.g., the conjunction rule, avoiding base rate neglect or default bias). The advantage in using such problems is ensuring that solving them does not rely merely on logic or common sense. The weakness is that the problems may present participants with a relatively high level of difficulty (i.e., more sophisticated, require some understanding of probabilities and statistics, even if in a more basic intuitive level rather than in a formal manner).

Lawson, Larrick, and Soll demonstrate that pushing individuals to think faster is potentially more damaging than they might be helped by inducing them to think more slowly in giving correct answers. The evidence is more consistent (across a number of models estimated with different specifications) that thinking faster lowers the probability of answering problems correctly (relative to control), whereas the support for the positive contribution of thinking more slowly is equivocal. In just one model they find a moderate support for such contributive effect of slow thinking. The researchers provide another piece of evidence based on their models about the asymmetric effects of fast versus slow thinking: based on the magnitude of coefficients they infer that “whereas the intervention Fast harmed performance accuracy by 6.7 percentage points, Slow yielded a gain of only 2.7 percentage points, relative to the control” (p. 673).

The study entails three individual-level measures regarding reflective thinking: Cognitive Reflective Test (CRT) is the main measure, a version of it excluding numerical computations, and a measure of numerical competence. A negative effect of thinking faster prevails whether any of these metrics is included in the model or none of them. On the other hand, slow thinking exhibits a positive effect only when all three measures of reflective thinking are included, and actually their effects are all positive (and statistically stronger). Therefore, the propensity to think reflectively seems to overshadow the explicit effect of the Slow intervention.

In addition, the researchers show how the average level of accuracy (rate of correct response) across the 12 items-problems increases with the tendency to reflect: From a 0 score on CRT (no tendency to reflect shown) to a maximum of 3 (strongest reflection exhibited), the accuracy level rises on average by 9.5 percentage points with a one-point increase on the CRT scale. Overall, the accuracy level of respondents (across conditions) mounts from 42% at score 0 to 70% at score 3 on the CRT scale.

We may learn from these findings that while it is hard to establish a benefit of thinking more slowly, thinking in a reflective way does seem to be important for solving (statistical-oriented) JDM problems successfully. It is not the slow pace of reflective thinking elicited by the intervention that consequently matters as much as the personal tendency to apply a reflective, deliberative style of thinking. Nonetheless, it may be even more beneficial to refrain from hurrying individuals as this may harm more greatly their judgements and the quality of the decisions they make. As we will see next, there is yet an additional qualification required for getting a true and fuller benefit from thinking more reflectively or slowly — it concerns the utilisation of appropriate inferential rules.

• One of the contentions made by Lawson et al. is that comparing between a fast condition and a slow condition enables measuring the difference in performance in favour of the latter, but it does not allow to identify the source of the difference — whether it arises from the benefit of thinking more slowly or the harm that may be caused by speeding up individuals to think faster. They argue that the fast and slow conditions have to be disentangled, such as by comparing each one of them to a control condition, as done in their study (previous manipulations, they note, compared a fast condition with a slow condition or control, which was not sufficient). The within-subjects treatment still focuses on contrasting fast-versus-slow, but it offers an additional advantage of detecting if participants change and correct responses per person when transitioning from fast intuitive thinking to slower reflective thinking.
• The researchers display a graphic summary of a number of comparisons between the conditions they tested: The average accuracy level across 12 problem-questions rises from 52.2% in the Fast_BS condition to 61.6% in Slow_BS condition; yet the latter is very close to accuracy of 58.9% in the control condition. A similar result is obtained in the within-subjects treatment (Fast_WS 50.6%, Slow_WS 59.0%). The Fast intervention turns out to be more consequential than the Slow intervention.

For answering correctly the questions on JDM problems as articulated by the researchers one has to identify and apply a relevant inferential rule — it is a necessary tool one has to know of in advance for solving the problem. When thinking fast and intuitively a participant may retrieve a rule that seems applicable at first sight, but one which could actually be less suitable (e.g., chosen on the basis of a wrong criterion or less relevant information included in the problem description). More observant, deliberative or reflective thinking about the problem that takes a little longer may help the individual to realise and access from memory the appropriate inferential rule to apply. In order to certify whom of their participants have the ability to access the relevant inferential rule for each of the six types of problems, the researchers presented them in the first stage of the study simplified, more transparent versions of problems from each category (i.e., a form that may cue the relevant rule more easily). The problems in Stage 1 were matching the ‘full’ problems presented in Stage 2 with respect to the rule that should be applied (i.e., same problem category).

Reflective thinking (CRT) together with access to the specific inferential rule applicable to a problem of a given type contribute positively to the likelihood of answering correctly the matching problem-questions, relative to a situation when one is thinking fast. The researchers therefore emphasise that knowledge of the specific rule relevant to the problem context is essential and is more important than knowing ‘candidate’ inferential rules in general. Yet, they note that general knowledge of rules in the domain also benefits performance (as well as numeracy). When compared relative to Fast thinking, Slow thinking has a positive effect on accuracy, but thinking ‘freely’ in the control condition also has a positive effect, and the Slow intervention is not superior in its contribution over the control condition. Again, it should be noted that the effect of Slow thinking is positive and significant when accounting also for the propensity to think reflectively and the use of the specific rule relevant — accentuating what makes the slow thinking condition conducive.

Another set of models, where Fast and Slow conditions are compared relative to control, provide more supportive findings, consistent with the stated above: The effect of Fast thinking is negative and statistically significant whereas the effect of Slow thinking is positive but weaker and not significant; yet, reflective thinking, access to the specific inferential rule relevant, as well as knowledge of rules applicable to other problems, would improve accuracy of response over the control condition. Moreover, the interaction of reflective thinking (e.g., CRT) with access to the specific rule relevant to a matching problem further enhances the likelihood of solving the JDM problems correctly.

Lawson and colleagues consider three types of persons: (1) experts who are able to detect and access the relevant inferential rule even when thinking intuitively; (2) novice individuals who do not have the ability to access the rule simply because they lack knowledge of those rules; and (3) knowledgeable but not expert individuals who are familiar with such inferential rules but may need more time to deliberate and reflect in order to identify and access the appropriate rule to apply. Experts and novices are not expected to be affected by an intervention how to think: experts can perform well in either mode of thinking, and novices cannot be harmed or helped much either way. Individuals of the third type stand a higher risk to lose from an intervention if they are speeded up and not allowed to deliberate, thus being deprived of the chance to access the relevant inferential rule from their memory-based knowledge. The implications of this distinction between decision-makers can be better understood through the function of reflective thinking jointly with access and application of the specific rule relevant to a problem, as demonstrated in the research of Lawson, Larrick, and Soll.

We may need to add a reservation here. For the deliberative kind of thinking of System 2 to roll-in and have an effect, two conditions have to be met: (1) one has to recognise (consciously or unconsciously) that an answer given by System 1 is missing, inadequate or has to be corrected; (2) one is familiar with and can access relevant judgement and decision tools such as inferential rules that he or she can apply to provide a better or correct answer. There is no assurance that these conditions will be met. Kahneman (2012) noted that System 2 tends to be lazy and needs a ‘persuasive’ trigger to roll-in (e.g., by a call for help from System 1). The ‘expert’ could be right to rely on an intuitive answer without having to continue thinking. The ‘novice’ may not recognise anything wrong with an intuitive answer, and even if he or she did, it may not help much to reflect and deliberate without access to relevant rules.

A person of the ‘intermediate’ (and probably most frequent) type may sense that something could be inadequate or wrong with the rule he or she is trying to apply intuitively and that will trigger slower and more reflective thinking; based on prior knowledge of inferential rules, that individual has a good chance of identifying and applying the appropriate rule after some additional deliberation. But a ‘warning’ signal must first be successfully triggered, and the relevant rule should indeed be accessible for the reflective and deliberative thinking to prove beneficial.

Lawson, Larrick, and Soll offer a sober and insightful perspective on the relations between fast intuitive thinking and slow reflective thinking. Merely instructing consumers to think more slowly may not be so effective and helpful; it is more important that the consumer has a tendency and willingness to reflect, and the necessary tools accessible and ready to apply (e.g., inferential and choice rules). This may have implications for relations between sellers and consumers: when consumers demonstrate the willingness and skill to deliberate and think more carefully, especially in situations that require greater background knowledge and comprehension, a seller should not try to press and speed-up the consumer as this could be damaging to both. A consumer who needs and seeks the time to think more, should be given the opportunity to deliberate, so he or she may not end up resentful and frustrated but instead satisfied and relieved.

Ron Ventura, Ph.D. (Marketing)

References:

[1] Comparing Fast Thinking and Slow Thinking: The Relative Benefits of Interventions, Individual Differences, and Inferential Rules; M. Asher Lawson, Richard P. Larrick, & Jack B. Soll, 2020; Journal of Judgment and Decision Making, Vol. 15, No. 5 (September), pp. 660-684. (The journal provides open access, see the article link fourth on the list for this issue).

[2] Thinking, Fast and Slow; Daniel Kahneman, 2012, Penguin Books