Transitivity

intransitiveimagesThis post is based on a paper that does a good job of providing a general picture of some of the big questions in decision making.  The experiments with their small samples (even though the results are statistically significant) seem unlikely to be definitive, but the overall measure of transitivity, I think is a good one.  I usually think about transitivity only once a year, when I am getting my eyes tested–Which one is clearer is it a or b? Now is it b or c? …Transitivity is usually considered required for rationality. In this paper, they use it as a measure of both intuition and analysis. Of course, transitivity does not work in rock, paper, scissors, and humans seem to be able to be quite irrational in certain of their preferences.

The researchers, Zohar Rusou, Dan Zakay and  Marius Usher note that there is considerable agreement among researchers that information in decision making involves two qualitatively different thinking modes: (1) an intuitive mode characterized by fast and parallel processes that are affective, holistic, and associative in nature and (2) a deliberative/analytical mode characterized by slower processes that are rule based in nature. They note that despite the above mentioned agreement, there are different viewpoints regarding the ways in which these two thinking modes interact. According to Rusou et al some like Kahneman have argued that the two modes operate sequentially, so that intuitive answers to judgment problems are generated rapidly and automatically, and then the analytical mode may initiate a process of monitoring, endorsing, correcting, or overriding the initial intuitive response. Yet other researchers have suggested that the two thinking modes work in parallel and are used to different extents depending on the decision environment.

Rusou et al carried out experiments to help resolve the ongoing debate concerning the relative advantage of each mode of thinking in decision making. Studies show contradictory results. Some studies show that the intuitive mode is inferior to the analytical mode, so that intuition and/or emotion may be a source of faulty decisions. Other studies demonstrate that the intuitive mode yields equal or better decisions, in comparison with the analytical mode.

Identifying the circumstances under which each thinking mode is preferable might help in understanding the advantages of each mode. Hammond et al.  were the first to suggest that performance is best when the mode of a person’s thinking matches the task. Hammond proposed that the nature of the task could be more intuitive or more analytical, depending
on task characteristics. The characteristics of intuitive tasks include high familiarity, pictorial presentation, subjective measure, and unavailability of an organizing principle or algorithm to integrate cues. Analytical tasks are characterized by quantitative presentation, objective measures, and an organizing principle readily available. Hogarth proposed that tasks are more likely to be processed through the intuitive mode when their context and form promote visual reasoning, whereas stated that tasks in which participants must follow strict rules are performed best through analytical deliberation. Despite these extensive theoretical hypotheses, relatively little experimental work has directly examined the association between thinking modes and task characteristics.

Rusou et al assume that the relative advantage of each thinking mode in decision making depends on the compatibility between thinking mode and task. To test this assumption, they pitted intuition and analytical thinking against each other within either intuitive or analytical tasks, comparing decision making in a factorial design.

Transitivity  was used as the dependent measure in pairwise choices. Transitivity is one of the major principles of rationality that stands at the basis of the development of decision theory.
This principle implies that for any three alternatives (A, B, C), if A is judged as better than B and B is judged as better than C, then A should also be judged as better than C. If individuals are inconsistent or prone to error in decision making, they will commit many violations of
transitivity. In contrast, if the error rate is very low and if individuals evaluate the different options consistently, there will be few violations of transitivity. Here, the researchers looked for more consistency when thinking mode matches task, thus expecting fewer transitivity violations in such cases.

Since previous results were subject to alternative explanations, the present study was designed to avoid similar pitfalls. (1) Information was presented in a uniform format across conditions. (2) A uniform measure was used for both the intuitive and the analytical tasks.  (3) Looking at transitivity relations and violations across choices offers a measure of consistency that was missing in previous studies.

Experiment 1 example (above).  There were 36 pairs to be evaluated in each of the “faces” and “multiplications”. For the analytical mode, subjects were told that they would need to provide reasons for their choices. For the “faces”, the mean number of transitivity errors was 4 for the analytical mode and 2 for the intuitive mode. For the “multiplications”, the mean number of transitivity errors was 7 for the analytical mode and 14 for the intuitive mode.

Experiment 2 example (above).  There were 28 pairs to be evaluated in each of the numerical averages and photos. For the analytical mode, subjects were asked to choose which “is” better or larger, while in the intuitive mode, subjects were asked to choose which “feels” better or larger. For the “averages” the mean number of transitivity errors was 4 for the analytical mode and 8 for the intuitive mode. For the “photos” the mean number of transitivity errors was 3 for the analytical mode and 1 for the intuitive mode.

The experiments  provide support for the hypothesis that the compatibility between thinking mode and task characteristics determines which mode is optimal in each case. On pictorial tasks, intuitive thinking resulted in fewer transitivity violations  than did analytical thinking, whereas on numerical tasks, analytical deliberation led to more consistency than did intuitive thinking. To conclude, the authors propose that the debate as to whether decision quality is best when using intuitive or analytical thinking might be clarified once each thinking mode is examined within a task that favors such thinking.

Rusou, Z., Zakay, D., & Usher, M. (published online February 22, 2013) “Pitting intuitive and analytical thinking against each other:  The case of transitivity.” Psychon Bull Rev.

Please note if you look at this paper that the analytical and intuitive columns in figures 3 and 5 are reversed. The darker color is intuition.

 

 

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.