donderdag 24 april 2014

All Accidents Are Preventable?


I think for the statement "all accidents are preventable" to become REMOTELY true (ish) we will need to add some words. For example:

All accidents are preventable... with the benefit of hindsight.


All accidents are preventable... in theory.


All accidents are preventable... given unlimited knowledge, resources, perfect prediction (and quite some luck).

All of which, regrettably makes it a rather useless statment in my everyday job. And besides, do we really want to prevent absolutely everything? Really??

dinsdag 1 april 2014

Gerd Gigerenzer - Gut Feelings - Short Cuts To Better Decision Making

This book was on my “things to do” list for quite a while (triggered by Gladwell’s “Blink” and Gigerenzer’s “Risk” book) and went substantially up to the top of the list after “Antifragile” and Taleb’s promotion of heuristics. It’s very compact, almost 230 sides of easy to read (and often even slightly humorous) text.

The first part of the book is titled “Unconscious Intelligence”. We think of intelligence as deliberate, conscious activities guided by the laws of logic. Much of our mental activities, however, are unconscious and driven by gut feelings and intuitions without involvement of formal logic. Our mind often relies on the unconscious and uses rules of thumb in order to adapt and economize. One reason it needs to do so is because there is only so much information we can digest at one time.

One advantage is that these simple rules are less prone to estimation and calculation error and are intuitively transparent as opposed to complex models. So, an intuitive shortcut, or heuristic, often gets us where we want without a smaller chance of big errors, and with less effort. Gigerenzer says therefore that it’s not a question if but when we can trust our intuitions.

One example for this that an experienced chess player will usually generate the best solution first and he will not do better with more time to reflect and reconsider - rather in contrary. Inexperienced players on the other hand will most of the time benefit from long deliberation. So, stop thinking when you are skilled. Thinking too much about processes we master (expertly) will usually slow down or disrupt performance as everyone who has tried to think about going down the stairs can confirm. These things run better outside our conscious awareness, so more is not always better.

Heuristics try to react on the most important information, ignore the rest and lead to fast action. Heuristics are a result of evolved capacities of our brain: simple rules are developed over time and thanks to practice we are able to perform an action really quickly and well - effective and efficient. The selection of the applicable rules is unconscious. Gigerenzer thus views our mind as an adaptive toolbox with rules of thumb that can be transferred culturally or genetically and also developed by ourselves, or adapted from existing rules.

One function of intuition is also to help us master one of our main challenges: we don’t have all the information, so we have to go beyond the information that is given to us. Out brain often “sees” more than our eyes by “inventing” things in addition to what in fact is seen, like depth (a ‘third dimension’) in a (2D) drawing.

If a gut feeling will have the wanted/correct outcome depends upon the context it’s used in. So a heuristic is neither good nor bad, this depends upon the environmental structures. Selection of rules of thumbs can be triggered by the environment (‘automatic rules’) or selected after a quick evaluation; conscious or (often) not. If the first chosen of the latter category ‘flexible rules’ doesn’t work another is selected. Gut feelings may seem simplistic, but their underlying intelligence is selecting the right rule of thumb for the right situation (depending on circumstances, environment, etc).

It’s perceived wisdom that complex problems demand complex solutions. In fact, however, in unpredictable situations the opposite is true. As things are, our world has limited predictability. Keeping that in mind one may consider that we should spend less resources and money on making complex predictions (and on consultants who make them for us). In hindsight a lot of information may help to explain things and it’s easy to fit information to past events. In order to predict the future, however, much of the information one gets is not helpful in predicting and thus it’s important to ignore the information that is not of value. The art of intuition is to ignore everything except the best clue that has a good chance on hitting that useful information. Psychological research suggests that people often (but not always!) base intuitive judgments on one single good reason. (do check pages 150 and 151!).

The final chapter of part one discusses intuition and logic. It starts with the famous ‘Linda problem’. Criticizing Kahneman Gigerenzer argues that calling the intuitive solution of most people a fallacy is not correct because gut feelings (or humans in general) are not governed by the rules of mathematical logic. He says that the human brain has to operate in a uncertain world, not in the artificial certainty of a logical system. Our brain has to make sense of information given and go beyond it. It zooms in on certain parts that seem particularly relevant within the context and it views words (like ‘probable’) in their common use and conversational meaning rather than in a formal academic/logical sense. Gigerenzer sees heuristics as a way to success rather than as a cause for error.

I find that there are arguments for both ‘sides’. Humans are bad with probabilities and numbers (something which Gigerenzer addresses in his other book, by the way) but humans don’t operate necessarily to the rules of formal logic - as anyone can confirm who has seen humans in action. An interesting thing to think about and keep in the back of your mind.

Gigerenzer concludes that logical arguments may conflict with intuition, but that intuition is often the better guide in the real world. Nevertheless many psychologists treat formal logic as the basis of cognition and many economists use logic as the basis for ‘rational’ action. This isn’t how the world works, however, and logic is blind to content and culture and it ignores environmental structures and evolved capacities. Gigerenzer closes by stating that good intuition must go beyond information given and therefore beyond logic.

Part two is called “Gut feelings in action” and discusses a couple of real-life examples of intuitions. It starts with the functions of recognition (a very strong memory function of our brain) and recall (not so strong, especially with age). Recognition helps us to separate the old from the new. The recognition heuristic can help us making intuitive judgments, like when someone who knows nothing about football is asked what team will win, he will probably pick the best known, and more often than not be right. This means that in some cases ignorance even can be beneficial because more knowledge may mean that one cannot rely on this heuristic anymore. One remarkable fact is that during tests people who relied on the recognition heuristic made snap decisions which appeared to impress people with greater knowledge who needed time to reflect.

The recognition heuristic is an example of a flexible rule which is chosen by our unconscious intelligence after an evaluation process. It can be overridden consciously in several ways. Another example of ‘one reason’ decisions is the way that political preferences are commonly ranged on a left-right scale, even if the subject at hand is totally unrelated to the left-right opposite.

Not always do we rely on just one reason, often we make intuitive judgments based on evaluation of a sequence of cues that are evaluated, but again only one determines the final decision - so called sequential decision making. We go through a series of cues (most important first, second next, etc) and evaluate options. As long as options ‘score equal’ we continue evaluating, but at the first cue where one option is best we stop. We don’t evaluate all pros and cons to find the optimal solution rather we choose the ‘first best’. Sequential decision making based on the first good reason is very efficient, transparent and often more robust and accurate than complex models.

One tool for sequential decision making is a ‘fast and frugal tree’ which through a couple of yes/no questions leads to a quick decision, rather than working through a huge, complex complete decision tree. Fast and frugal trees have three building blocks: 1) a search rule that looks up factors in order of importance, 2) a stop rule that stops looking further if a factor allows so, and 3) decision rule that classifies an object. Among others medical services use these ‘developed intuitions’ for making diagnoses. These simple, transparent empirically informed rules help making better decisions.

The last two chapters deal with moral behavior and social instincts (e.g. imitation and trust). People do morally unbelievable things (like the example of a WW II mass murder illustrates) because of their reluctance to break group order. Peer group pressure (consciously or not) is enormous and it may even overrule deeply rooted moral instincts like “You shall not kill”.

One important heuristic to understand is that people will usually opt for the default (chose by the environment, or ‘system’) instead of making a conscious choice - as is illustrated by the differences in percentages of organ donors between various countries. By understanding the process and framing instructions or requests well, one may be able to steer things in a desired (and preferably morally just) direction.

This summary/review obviously only scratches the surface of themes treated in this most recommended book. Hope I tickled your interest, now go and read it for yourself. 

I’ve read the Penguin pocket version ISBN 978-0-141-01591-0