Decision Science News

2009 BRUNSWIK SOCIETY MEETING CANCELLED, THOUGHTFUL ESSAY EXPLAINS

Egon Brunswik

Ken Hammond wrote this conference cancellation notice / thoughtful essay. If you are concerned with current directions in Brunswikian research, it may be the most interesting conference cancellation notice you ever read.

[Edited for length – Ed]

Dear Participants in the Brunswik Society

As many of you will have noticed, the 2009 meeting has been cancelled. This was done because of my dissatisfaction with the program and the manner in which it has been carried out. An explanation follows some brief history.

The Brunswik Society was organized 25 years ago in Boston during a discussion with several colleagues in which we expressed our dissatisfaction with the absence of Brunswikian concepts and research (about which more below) in the Judgment and Decision Making Society meetings. It may be worthwhile to remind readers that the J/DM Society that exists today emerged from two separate meetings; (a) the ones that had taken place to encourage further work on the Bayesian approach that had been developed by Ward Edwards and colleagues and (b) the meetings that had taken place in Boulder, Colorado that included Bayesian work and Brunswikian work developed by me and my students as well as other colleagues . Ward’s meetings continue (despite his death), but the meetings in Boulder were discontinued many years ago. Fortunately, Jim Shanteau, Gary McClelland, John Castellan (dec) and others took up the task and formed the current J/DM Society. It was the absence of Brunswikian discussion in the J/DM meetings that led to the formation of the Brunswik Society 25 years ago. The meetings were generally considered successful in that the attendance grew so much that it became necessary to expand the meetings to one and onehalf days; this was astonishing to me. In view of this apparent success, why has the 2009 meeting been cancelled?

A little more history; when the Society was formed I made all the arrangements and set up the one day program. Initially, about 10 people attended, and attendance gradually grew to about 45 participants, literally, from all over the world. As the Society grew, Tom Stewart set up a web page, a newsletter was created, and other attributes of a successful scientific Society appeared. After about 10 years of organizing meetings I asked Tom to take over this task, which he did very successfully (the attendance grew). Tom asked Jim Holzworth to organize the meetings after about eight years or so and Jim has been doing so. My dissatisfaction with the recent meetings grew out of my judgment that the meetings did not critically discuss Brunswikian concepts or methodology, but merely gave thin examples of some of his ideas. I decided that many of these papers might have been interesting 20 years ago, but did not advance our understanding of Brunswikian principles.

I tried to convey that criticism to the Program committee but failed. I concluded that the Committee would never see my point and that the only way to change course and restore the original purpose of the formation of the Society was to cancel the 2009 meeting and to start over with a new Program committee. Jason Beckstead and Elke Kurz-Mickle accepted appointment for the future. Because of my advanced age, after this message I will no longer participate in the affairs of the society, although of course my interest will continue. In what follows I indicate some topics that I believe warrant discussion at our meetings.

Topics

a. Uncertainty. Despite the central place of this concept there has been little or no discussion of uncertainty in the last 20 years of Brunswikian meetings. This despite the disputes between Gigerenzer and Kahneman and their followers, each of whom has based an entire research program on different ideas about uncertainty. “Probability” is now prominent in the theories of a few cognitive psychologists (e.g., Anderson, 1991) but has yet to achieve a prominent place in the modern neuroscience.

b. Compromise. Compromise is one of the most important, yet least recognized and least employed, concepts introduced by Brunswik. It is at the root of his theory of cognition. He gave it considerable prominence in his 1956 book when he contrasted the two “intentions” of perception, namely, “proximal” (retinal) size and “distal” (object) size. Actual judgments were found to be located on a continuum between these two poles of intention, and thus represented a “compromise” between the two poles (its specific location depending on conditions), although generally approaching the distal pole. When discussing the intellect, compromises were located between intuition (perception) and analysis (thinking), and the process was termed “quasirationality”, the specific form of which was also dependent on conditions. This term — quasirationality — was chosen because it indicates approximation to, but not full achievement of, rationality. “Compromise” runs through all of Brunswik’s theorizing and can be brought to bear on modern theories of judgment as well as decision making. It is most apparent when theories such as TTB that entail “maximizing” are contrasted to “matching” behavior [TTB refers to the Take The Best heuristic of Gigerenzer & Goldstein, 1996 – Ed] . In this situation matching behavior would represent the compromise between (a) maximizing (TTB) that marks the analytical pole of cognition and (b) dividing weights equally among all cues as the most “thoughtless” method of judging that would mark the intuitive pole of cognition. Yet I have never heard a discussion of this concept in any Brunswikian meeting.

c. Achievement. I believe we need a richer, more inclusive, concept than mere “accuracy”, I prefer to speak now in terms of correspondence competence, and coherence competence (see, Dunwoody, et al (2009) for a discussion of these terms). “Accuracy” simply means correspondence of the judgment or decision with an empirically correct criterion, and that, of course, is of high significance. Coherence could mean the same, but the derivation of the “answer” would in general require the competence – explicit or implicit — to include other aspects of the task or process of judgment — as the Gestaltists first showed us with their concept of the gestalt. That is, a judgment — reached intuitively or analytically — based on the coherence all the elements — and their relationships — of the judgment should match all the elements — and their relation ships — of the task.. Correspondence theorists for their part, could and should enrich their conception of correspondence competence to include error distributions, as I describe elsewhere.

d. Intersubstitutability. The terms “vicarious functioning” and “vicarious mediation” were central to Brunswik’s emphasis on the uncertainty in the environment; they served to remove the ambiguity introduced by the broad concept of uncertainty; they specified exactly where both subjective and environmental uncertainty arises, how uncertainty in the environment causes uncertainty in the subject, why some environments are more uncertain than others, and many other features of our relation to our ecology. Brunswik was very specific about both matters, particularly in the field of visual perception. He denoted the uncertainty in the environment by pointing to the differences in validity and reliability (the latter often neglected) in various indicators or cues in the environment to object size — none are fully dependable — and also denoted their intersubstitutability (their intercorrelations with other cues indicate how often substitution can be used). This intersubstitutability is one reason why our visual perception and that of other mammals is so good. Fortunately for students of judgment and decision making, all these concepts carry direct implications from perception to the study of human judgment. But they are seldom examined in empirical detail. I have never seen such an examination at the Brunswik meetings.

e. Representative design. We are gradually making progress with the recognition of the sensibility of Brunswik’s suggestions regarding experimental design. The current Editor of the Journal of Experimental Psychology: Applied has declared that she will no longer accept mss that claim that their work applies to the “real world”, and instead will require demonstration of representativeness. Most important, however, in order for the phenomena that give rise to the requirement of representativeness , the organism must be given a multi-cue environment in which to behave. Unfortunately, however, even in Gigerenzer’s demonstrations of the ubiquity of heuristics, multi-cue environments are generally avoided in favor of binary presentations (Gigerenzer, 2009, and earlier). But if multi-cue tasks are used, the organism will also have a chance to engage in vicarious functioning in response to the vicarious mediation of information. Whether it will, and under which circumstances, has become a research issue of considerable interest (see Gigerenzer & Brighton, 2009, Hogarth & Karelaia, 2007). It has become clear, however, that answering this question does require the researcher to provide an environment representative of a wide range of conditions, that is, an environment that includes causal texture (cf. Tolman & Brunswik, 1935).

These five concepts (uncertainty, compromise, achievement, intersubstitutability, representative design) are the backbone of Brunswik’s probabilistic functionalism and marked the presentation of a new kind of experimental psychology. But they seldom make their appearance at meetings of the Brunswik Society.

Of course, I strongly hope for the long life of a Brunswik Society that discusses the ideas put forward by Egon Brunswik in a fashion that is congruent with his aims for the development of a scientific discipline.

Kenneth Hammond
9/1/09