A Bubble of Theory
Apr. 3rd, 2013 01:10 am![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
mtvesselAug 2012
Thinking Fast and Slow - Daniel Kahneman - Penguin, 2012 (Kindle edition)
* * *
One of the most attractive features of the collection of just-so stories that we call the physical sciences is its self-consistency. Biology derives from chemistry, chemistry from physics, physics from maths and maths from logic. Each supports the other. A chemical model that requires electrons to behave in ways that contradict the laws of quantum mechanics is (almost certainly) false. Likewise, if Stephen Hawking were to propose a form of string theory that predicts atomic behaviour that prevents, say, the formation of DNA strands, we would be perfectly justified in telling him that he is wrong - his theory is refuted by millions of biological observations. The self-reinforcing and interlocking nature of the standard biological, chemical and physical models and the surprising absence of obvious counter-examples (when searched for) strongly suggest that, imperfect and provisional as they are, we can regard them as about as near as humans can get to objective knowledge.
But what of the disciplines that study aspects of the human world - medicine, economics and psychology? Here things are more opaque. There are, certainly, models that are solidly based in biology and chemistry, particularly in medicine. But a number of theories in these areas seem to exist in their own little conceptual bubbles,only loosely connected (if at all) to the mainstream of the physical sciences. Freud's theory of the superego, ego and id is one example. Macroeconomic theories such as Keynsianism or Monetarism are another - they are nice mathematical models, but their assumptions about human behaviour and biology are hopelessly simplistic (as Kahneman shows). Such bubbles of theory can be beautiful, they can elegantly explain observations, they can even have been tested with experiments that follow good scientific practice, but without clear consistency with the models of the physical sciences, one has to be suspicious, for many such beautiful ideas - phlogiston, the aether, the four humours, vitalism - have not stood the test of time.
I fear that the model of thinking that Kahneman promotes in this book is a similar bubble of theory. It may merge into the mainstream of science, or it may drift away and pop. Certainly it should not be hailed as true. Not yet, anyway.
Which is not to say that it is uninteresting. Kahneman posits two forms of thinking for which he unhelpfully uses the unmemorable and easily confusable labels "System 1" and "System 2". System 1 is fast thinking - instinctive, quick and effortless. Our instant assessments of people and rapid recall of simple facts (such as the capital city of France) are system 1-mediated. By contrast, system 2 thinking is slow, effortful and requires conscious attention - an example would be working out a non-trivial sum. Or writing the next sentence.
Kahneman has spent much of his career teasing out the characteristics of these systems by means of clever experiments involving common biases of judgement, perception and memory. For example, System 1 is dominated by the principle of "what you see is all there is", the tendency to consider only those factors that are immediately apparent. System 2 is less susceptible to bias - it is Sherlock Holmes' system 2 that leads him to consider the ramifications of the dog that didn't bark in the night time - but in most people it is lazy. We only use it when we have to, preferring to rely on rule-of-thumb estimates made by system 1.
This explanatory framework is used to examine many common biases and failures of intuition such as anchoring effects (where priming a person with a number causes them to shift their subsequent numerical estimates) and other well-known phenomena such as the failure to take Regression to the Mean into account when evaluating statistical results. It is then applied to examples of overconfidence such as the tendency to underestimate how long things are going to take (all too common in my line of work), or the hopelessly inaccurate predictions of chief financial officers about the movements of common financial indicators. This leads into a discussion of how people make choices and the failure of conventional economic theories - based on totally rational beings that Kahneman calls "Econs" - to explain them. A more psychological approach - prospect theory - assumes that people start from an anchor point and are more averse to losses than they are eager for gains, which explains a number of otherwise puzzling phenomena such as the sensitivity to the expression of risk (an operation with a one month survival rate of 90% sounds better than an operation with a 10% risk of death in the first month) or the tendency to overweight the salience of rare but bad events in risk analysis (this means, for example, that you will make worse investment choices if you check your share prices every day rather than, say, once every three months).
The system 1/system 2 framework provides a handy way of uniting some disparate observations and helpfully suggests approaches to avoiding some of the more obvious mistakes which seem to work in practice. However, I am still not convinced by the science. Very early on, Kahneman states that system 1 and system 2 do not have obvious locations in the brain, and my immediate response was "why not?" If they are so fundamental to the way in which we think, there should be some very obvious distinction between them. At the very least, MRI scans or EEGs should show significant differences in brain activity when people engage in system 1- and system 2-based cognitive tasks. Surely these experiments have been done?
The psychological experiments that Kahneman describes, elegant and striking though they are, have some obvious potential confounders. Are effects like aversion to risk and anchoring seen in all cultures, or are they a feature of the WEIRD (Western, educated, industrialised, rich and democratic) undergraduates on which the experiments were performed? Could some of the more striking responses be due to the fact that a good proportion of subjects misinterpreted an ambiguous question rather than demonstrating a fundamental feature of the way in which we think? As with Daniel Gilbert's Stumbling on Happiness, the descriptions of the experiments are lacking.
And I do think that Kahneman goes out on a limb when he moves on to discuss the distinction between the "remembering" and "experiencing" selves in the final section. He illustrates this with a thought experiment. Imagine that you were offered a fabulous holiday but that on return, all your memories and pictures of it would be wiped. How much would you be prepared to pay for such an experience? To me, this question is nonsensical. An experience is your memory of it; there is no fundamental distinction between the self that experienced the event and the self that remembers it. But Kahneman thinks there is. No, I don't know why either.
Still, if I were a betting man, I would wager that something like Kahneman's systems of thought will turn out to be represented in the structures or processes of the brain. They make evolutionary sense (one area where they do accord with the physical sciences) and neatly explain some otherwise puzzling psychological effects. But truth? Not yet.
Thinking Fast and Slow - Daniel Kahneman - Penguin, 2012 (Kindle edition)
* * *
One of the most attractive features of the collection of just-so stories that we call the physical sciences is its self-consistency. Biology derives from chemistry, chemistry from physics, physics from maths and maths from logic. Each supports the other. A chemical model that requires electrons to behave in ways that contradict the laws of quantum mechanics is (almost certainly) false. Likewise, if Stephen Hawking were to propose a form of string theory that predicts atomic behaviour that prevents, say, the formation of DNA strands, we would be perfectly justified in telling him that he is wrong - his theory is refuted by millions of biological observations. The self-reinforcing and interlocking nature of the standard biological, chemical and physical models and the surprising absence of obvious counter-examples (when searched for) strongly suggest that, imperfect and provisional as they are, we can regard them as about as near as humans can get to objective knowledge.
But what of the disciplines that study aspects of the human world - medicine, economics and psychology? Here things are more opaque. There are, certainly, models that are solidly based in biology and chemistry, particularly in medicine. But a number of theories in these areas seem to exist in their own little conceptual bubbles,only loosely connected (if at all) to the mainstream of the physical sciences. Freud's theory of the superego, ego and id is one example. Macroeconomic theories such as Keynsianism or Monetarism are another - they are nice mathematical models, but their assumptions about human behaviour and biology are hopelessly simplistic (as Kahneman shows). Such bubbles of theory can be beautiful, they can elegantly explain observations, they can even have been tested with experiments that follow good scientific practice, but without clear consistency with the models of the physical sciences, one has to be suspicious, for many such beautiful ideas - phlogiston, the aether, the four humours, vitalism - have not stood the test of time.
I fear that the model of thinking that Kahneman promotes in this book is a similar bubble of theory. It may merge into the mainstream of science, or it may drift away and pop. Certainly it should not be hailed as true. Not yet, anyway.
Which is not to say that it is uninteresting. Kahneman posits two forms of thinking for which he unhelpfully uses the unmemorable and easily confusable labels "System 1" and "System 2". System 1 is fast thinking - instinctive, quick and effortless. Our instant assessments of people and rapid recall of simple facts (such as the capital city of France) are system 1-mediated. By contrast, system 2 thinking is slow, effortful and requires conscious attention - an example would be working out a non-trivial sum. Or writing the next sentence.
Kahneman has spent much of his career teasing out the characteristics of these systems by means of clever experiments involving common biases of judgement, perception and memory. For example, System 1 is dominated by the principle of "what you see is all there is", the tendency to consider only those factors that are immediately apparent. System 2 is less susceptible to bias - it is Sherlock Holmes' system 2 that leads him to consider the ramifications of the dog that didn't bark in the night time - but in most people it is lazy. We only use it when we have to, preferring to rely on rule-of-thumb estimates made by system 1.
This explanatory framework is used to examine many common biases and failures of intuition such as anchoring effects (where priming a person with a number causes them to shift their subsequent numerical estimates) and other well-known phenomena such as the failure to take Regression to the Mean into account when evaluating statistical results. It is then applied to examples of overconfidence such as the tendency to underestimate how long things are going to take (all too common in my line of work), or the hopelessly inaccurate predictions of chief financial officers about the movements of common financial indicators. This leads into a discussion of how people make choices and the failure of conventional economic theories - based on totally rational beings that Kahneman calls "Econs" - to explain them. A more psychological approach - prospect theory - assumes that people start from an anchor point and are more averse to losses than they are eager for gains, which explains a number of otherwise puzzling phenomena such as the sensitivity to the expression of risk (an operation with a one month survival rate of 90% sounds better than an operation with a 10% risk of death in the first month) or the tendency to overweight the salience of rare but bad events in risk analysis (this means, for example, that you will make worse investment choices if you check your share prices every day rather than, say, once every three months).
The system 1/system 2 framework provides a handy way of uniting some disparate observations and helpfully suggests approaches to avoiding some of the more obvious mistakes which seem to work in practice. However, I am still not convinced by the science. Very early on, Kahneman states that system 1 and system 2 do not have obvious locations in the brain, and my immediate response was "why not?" If they are so fundamental to the way in which we think, there should be some very obvious distinction between them. At the very least, MRI scans or EEGs should show significant differences in brain activity when people engage in system 1- and system 2-based cognitive tasks. Surely these experiments have been done?
The psychological experiments that Kahneman describes, elegant and striking though they are, have some obvious potential confounders. Are effects like aversion to risk and anchoring seen in all cultures, or are they a feature of the WEIRD (Western, educated, industrialised, rich and democratic) undergraduates on which the experiments were performed? Could some of the more striking responses be due to the fact that a good proportion of subjects misinterpreted an ambiguous question rather than demonstrating a fundamental feature of the way in which we think? As with Daniel Gilbert's Stumbling on Happiness, the descriptions of the experiments are lacking.
And I do think that Kahneman goes out on a limb when he moves on to discuss the distinction between the "remembering" and "experiencing" selves in the final section. He illustrates this with a thought experiment. Imagine that you were offered a fabulous holiday but that on return, all your memories and pictures of it would be wiped. How much would you be prepared to pay for such an experience? To me, this question is nonsensical. An experience is your memory of it; there is no fundamental distinction between the self that experienced the event and the self that remembers it. But Kahneman thinks there is. No, I don't know why either.
Still, if I were a betting man, I would wager that something like Kahneman's systems of thought will turn out to be represented in the structures or processes of the brain. They make evolutionary sense (one area where they do accord with the physical sciences) and neatly explain some otherwise puzzling psychological effects. But truth? Not yet.
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
no subject
Date: 2013-04-07 11:18 pm (UTC)