Daniel Kahneman has been practically ignored in this blog. His 2011 book: Thinking, Fast and Slow, is well written and an excellent resource. I certainly do not hold winning the Nobel Prize for Economics against him. I do wish he was more like Ken Hammond and gave me more background and more perspective on what questions that are likely to be answered in the future or what research he believes is interesting, or especially why Gigerenzer is wrong. System 1 and System 2 seem outdated and Kahneman seems to just ignore research like that of Glockner and Betsch Intuition in J/DM that sees the systems as more holistic.
This post is based on an interesting paper published in the November 2011 Information Systems Security Association Journal. “A Call to Arms: It’s Time to Learn Like Experts,” is authored by Jay Jacobs. Try to base your interest in the article on its well done ties to the JDM literature and not on any passive aggressive feelings that you may have toward the black box of your employer’s security efforts and subsequent requirements on you. Forget that the system administrators can look at anything with impunity, and you have to change your password every week or so it seems. Try to avoid any thoughts of Edward Snowden as an information security expert, and focus on the author who is clearly one of the good guys and several pay grades above those system administrators who might occasionally torture you.
Ski guides who use helicopters or tracked vehicles to get skiers into the treasured deep powder must evaluate avalanche possibilities. Iain Stewart-Patterson of Thompson Rivers University, Kamloops, BC, examines avalanche expertise in a JDM context in “What Does Your Gut Say and Should You Listen? The Intuitive-Analytical Decision Making Continuum in Mechanized Ski Guiding.” The paper was presented at the 2010 Snow Science International Workshop.
British Columbia does seem to be the obvious place to do such a study. Even a non-skier does not forget the concrete avalanche sheds that protect the highway that come seemingly one after another in southern British Columbia. Stewart-Patterson tells us that ski guides are trained in the decision process at each step in the certification process. There are three levels of avalanche training and four levels of guide training and certification representing a total of 50-60 days. Stewart-Patterson interviewed 32 accomplished guides with an average of about 9600 hours of experience. He also had them complete questionnaires.
Findings on Expertise
A. Expertise is limited to specific domains acquired through considerable experience.
B. Outstanding performance in any domain takes years of dedication usually including demanding regimes of deliberate practice benefiting from good teachers. It also seems to be necessary to have many hours of experience in what might be considered an unstructured manner. Learning from experience is key. We all develop intuitive abilities, but once skills have been over learned and made automatic–intuitive–self insight is difficult, and a third-party perspective can be helpful. True experts tend to restructure their performance with experience and acquire new methods and skills that others do not.
C. Experts are not much better in predicting the future than novices though this may be to an extent due to the uncertain and even random environments some of them work in. Hogarth also sees a social phenomenon at work here. He believes that people seek certainties in life and that experts are expected to provide these. Since people expect experts to provide accurate predictions, experts want to meet these expectations so they do their best to explain away all their mistakes and maintain the illusion that they are correct.
D. Experts and novices process information differently. Experts tend to make more use of their well-educated intuitions to counteract limitations in short-term memory. Novices tend to identify a specific goal and then using much analytic thinking work backward through the details toward a solution. Experts tend to take in the details of the problems they face, and then by recognizing patterns and similarities create a general framework that fits the data that allows them to explore possible solutions. Experts learn what patterns to expect in different situations, and their diagnostic activities involves comparisons with expectations. Do the data match or is something missing? Expertise involves the encoding of intuitive patterns and the more expert a person becomes, the more patterns available in memory.
My recent post Minimizing Diagnostic Error touched this subject, but this post will look at it in a more general way. Feedback is an important part of learning and accordingly decision making. Feedback can have varying levels of relevance. Relevance depends on which measure is selected and then measuring it well. Learning is further advanced if the consequences of error are greater.
Cass Sunstein is one of the more accomplished writers on judgment and decision making. He is an attorney and was recently the head of OIRA, the Office of Information and Regulatory Affairs at the White House. He seems to be able to write a book in about a day. He has written about twenty books including Nudge with Richard Thaler. One of the reasons is that the books each overlap a good deal, which should make it easier for me to get the gist of his ideology. Ken Hammond wrote of him as one of the coherence school of judgment and decision making.
He often speaks of conformity, social cascades, and group polarization. Group polarization seems to me to relate to the ideas of supersense, sacred values, and moral imagination that were presented in my previous post. I enjoy his references to the founding fathers and these references are often the most persuasive part of his work. This post looks at the two books referred to in the title of the post. The post will not be a good summary, but includes a few things I found interesting.
Bruce Hood is an experimental psychologist and in Supersense he argues that beliefs in the supernatural are a consequence of reasoning processes about natural properties and events in our world. This includes a mind design for detecting patterns and inferring structures where there may be none. Our naive theories form the basis of our supernatural beliefs, and religion, culture and experience simply work to reinforce what we intuitively hold to be correct. As an example, one of these is the common belief that we can tell when someone is staring at us. Hood says that supernatural thinking is simply the natural consequence of failing to match our intuitions with the true reality of the world.
According to Glockner and Betsch, deliberate constructions (DCs) are the opportunity for the deliberate/analytical system to provide input into decision making. The Parallel Constraint Satisfaction rule holistically considers the information contained in a network. The network consists of all pieces of information that comprise the decision problem (cues, goals, options, evaluations, etc.). In many mundane situations, the constitution of the network does not require any sort of active information search. Relevant features of the environment and currently activated memory entries provide the input to the network. Glockner and Betsch refer to the network installed spontaneously when encountering a decision situation as the primary network. Deliberate processes are activated if the consistency of a resulting mental representation is below a threshold θ. I think this is interesting because it is the intuitive/automatic system requiring coherency of itself. Typically, we think of the deliberate/analytical system as being rational and coherent, while the intuitive system requires correspondence. In this situation, the intuitive/automatic system finds answers that are not consistent, incoherent, and thus seeks input from the deliberate/analytical system.
Parallel Constraint Satisfaction Theory (this was discussed to some extent in the post “Intuition in J/DM“) is a descriptive theory of decision making whose main proponents are Andreas Glockner and Tilman Betsch. They propose that decision making uses analytic processes for information search and production and intuitive (automatic) processes for combining information and making the decisions. To provide this basic outline of PCS, I am using the preprints “Experts and Decision Making: First Steps Towards a Unifying Theory of Decision Making in Novices, Intermediates and Experts,” by Britta Herbig / Andreas Glöckner and published by the Max Planck Institute for Research on Collective Goods, Bonn 2009/2, and “How Evolution Outwits Bounded Rationality The Efficient Interaction of Automatic and Deliberate Processes in Decision Making and Implications for Institutions” by Andreas Glockner and published by the Max Planck Institute for Research on Collective Goods, Bonn 2008/8. I will likely be looking at specific parts of this theory in future posts. Continue reading