13 September 2016

Kritika kritike Kanemana, 2. deo

What is obscure? „At most, they simply demonstrate either the subjects’ ignorance of certain obscure statistics“ (p. 132) Well, ignorance may indeed be a relative term, but it is hard to construe the huge numerical preponderance of farmers over librarians as „obscure“ for anybody not living in a cave. In non-Orwellian speak, “certain obscure statistics” applies to things like the number of cane toads in Australia or the famous example of Enrico Fermi, the population of piano-tuners in Chicago. “Did it occur to you that there are more than male farmers for each male librarian in the United States?”—as if this information is intuitive or self-evident.“ (p. 134) it certainly is not self-evident, but contra BT it is intuitive, since intuition is based (on most reasonable philosophical construals, e.g., Hume’s “striking resemblance” in A Treatise of Human Nature) on actually having some prior knowledge on relationship of empirical objects. Of course, BT are free to use their own definition of what is intuitive or not – but are hardly entitled to criticize Kahneman for not sticking to their private definition in this respect.

„When respondents perform poorly, the reader is led to interpret their failure as a cognitive bias, although simple ignorance of either quantitative technique or of relevant facts suffices to explain it.“ (p. 134) Alternative explanations – which are, interestingly enough, not entertained by BT – include low IQ, iodine deficiency, serious depression, or simple laziness.

On bats and balls: “This is not at all a simple puzzle. The fact that it involves bats, balls, and small amounts of money makes it look simple and thus lulls respondents into thinking they should have the answer ready. In reality, this is a system of two equations with two unknowns—the type of mathematical problem that, in the United States, is not learned until at least the eighth grade, if ever.“ (p. 135)

Now, an interesting issue is why should the US education system (a broken one, btw, according to authors so disparate as Alan Bloom, Francis Fukuyama, Judith Butler, or Bernie Sanders) be a norm here? Even more interesting is how in the world could a person buying chocolate in two different packages in two shops for a fixed amount of money establish whether she received correct change at the counters or one shop is indeed cheaper than the other? This is also a system of two equations with two unknowns. And a person incapable of deciding so might not be irrational in the narrow sense of BT, but in reality should not be trusted very much. There are countless examples of similar problems involving two equations with two unknowns appearing in everyday contexts. Some adults take some kids to a movie, five people in all. The cost for adults is $9 and the cost for kids is $6. They paid a total of $36. How many kids are there?

„[T]he question is not only outside our normal range of experience and therefore irrelevant to the lives of most people.“ (p. 135) Only those people not doing shopping – or those not very careful with their money, or those not likely to go with kids to a movie, or... On a serious note, this assertion that what is outside our own range of experience is therefore irrelevant to our lives or to the lives of most people is, at best, outright bizarre. As someone who did some work in risk analysis, I find this claim preposterous. Rational people take into account events outside their own range of experience all the time – the insurance industry would not exist otherwise. Our own death is by definition outside our range of experience (alleged after-death experiences or spiritism notwithstanding) and yet people invest in life insurance all the time. Large natural catastrophes like tsunamies, large earthquakes, supervolcanic eruptions, etc. are outside of range of experience of most people – and yet are supremely relevant for anyone on the planet. Nuclear war has been, fortunately, outside our own range of experience so far – and yet is utterly relevant for all lives on the planet, human and animal and plant alike. In claiming this, BT show surprising narrow-mindedness and – yes, ironically, considering their parable above – large degree of self-absorption.

On geography: „There really should be no surprise that a subject asked about the total number of murders in Michigan is not able to invoke two different pieces of information never mentioned in the question: that Detroit is a high-crime city and that Detroit is in Michigan.“ (p. 135)

Consider a related statement: There really should be no surprise that a voter asked about the promise of a candidate for the governor of Michigan to reduce the number of murders in Detroit is not able to invoke two different pieces of information never mentioned in the question: that Detroit is a high-crime city and that Detroit is in Michigan. Plus, the voter was never explicitly told that reduction in the number of murders is desirable, therefore she cannot make rational judgement on the candidate’s program. And yet, and yet, the voter will be – irrationally? – asked to make a judgement on those issues at the time of elections.

The example in footnote 3 is even worse, since it is unintelligible: if drivers believe that they are better than the average and the average is being construed as “unexceptional”, what in the world could it mean for someone to be “better than unexceptional”? One could similarly argue that he is “fitter than yellow” or “more intelligent than mountainous”; it could mean something only in the context of a Derridian or a Žižekian farrago of nonsense.

„A student who cannot solve the integral because he did not think of making the particular substitution in that case is not sufficiently proficient at calculus; but he is not biased against solving the integral.“ (pp. 135-136) Such conclusion would require telepathic superpowers to be established. In most cases, bias is defined in science as something discernible from an external perspective (perspective of observer or a statistician) – e.g., Malmquist bias, Signor-Lipps effect, various anthropic biases, etc. So, while we might indeed refrain from calling the student’s lack of performance a bias without further information, it is at least conceivable that it results from a bias, e.g., from an error in his textbook on the particular substitution, or from the teacher lacking skills in this respects, etc.

„Respondents may be wrong about the total number of murders in Detroit and Michigan, but these aggregate numbers are information that is of very little importance outside government statistical offices, just as is the question Kahneman asks.“ (p. 136) Unless one is living in Detroit or in rural Michigan or considers moving there or should judge programs of crime reduction or is deciding upon donating to charities in Michigan or some of other zillion cases BT – similarly as with the application of conservation of momentum and energy in driving – conveniently downplay and/or ignore. Which in their case, I submit, is not caused by ignorance, but by the biased view of the role of knowledge in human actions and values. However, BT do not fail to emphasize that it is outside government statistical offices, and not statistical offices or, say, private charity research like GiveWell or other private-sector statisticians, who do no less statistical research than the government does.

On the Linda experiment: „A reasonable person may, in effect, be answering a different and broader question that evaluates the entire situation rather than the textual question alone—for instance, “What has the examiner designated as the correct answer?” Or “What is the most likely answer in this experiment that I am in?”“ (p. 137)

I imagine one should use similar heuristics in filling one’s tax forms for example – not giving narrow and literal answers, but musing on what the government really wishes to achieve with existing tax rates or whether it would be better for all parties if I answer something differently, etc. The outcome of such an experiment would certainly be wondrous! Again, it is pretty ironic that BT ascribe such highly intelligent analysis to the very same set of subjects who are previously excused for failing to understand logical conjunction or for not knowing that Detroit is in Michigan.

On choices allegedly beyond cognitive biases: „it may be worth contrasting them against the costs of mistakes that clearly have nothing to do with cognitive biases: the cost of choosing a profession one ends up hating, the cost of not finding a suitable mate, the cost of having children too early in life or too late, the cost of moving to a place one ends up disliking, the cost of adopting a pet or sending children to a private school, and so on.“ (p. 140) How in the world are those choices not influenced by cognitive biases? After all, one decides about school for one’s children often on the basis of pure confirmation bias, and the profession one ends up hating is intuitively (if not self-evidently!) often a consequence of the temporal horizon bias, etc.

„Kahneman’s reliance on this model of decision making suggests that we would ideally become more literal, abstract, out-of-context thinkers and, as such, more in line with the neoclassical economic model, without making it clear how this would improve the human condition.“ (p. 142) The charge is as bizarre as what was brought, for instance, against Einstein by „Aryan physicists“ Lenard and Stark in 1930s that the theory of relativity does not contribute to the glory of Aryan race or obtaining the desired Lebensraum. On the Serbian scene, we have witnessed a dumb science minister (Dragan Domazet) who urged scientists to work „for their country instead of working for libraries“ (!). This misplaced pragmatism – reminiscent of Lysenkoism in Soviet times – has never been a signpost of scientific success.

On Kahneman’s alleged revolution: „In behavioral economics, then, we do not find the building blocks of an alternative decision making theory that could be incorporated into a more fruitful type of social science. In Kuhnian terms, it is not a scientific revolution but a strange transmogrification of an old and manifestly inadequate paradigm.“ (p. 143) While „Kuhnian terms“ should make one suspicious, it is the idea that we should consciously strive toward „fruitfulness“ of science – even social one! – which is really problematic. Arguably, Kuhnian view of revolutions in science is much less important for epistemology and philosophy of science of today than the Lakatosian notion of research programmes – and it is exactly Kahneman’s research programme (and certainly not a revolution!) which brought him the deserved success. Even under Kuhnian terms revolutions are not acts of will, as BT seem to suggest in both the title and this concluding passage – there is no perceived desire for a more fruitful kind of science which leads to a scientific revolution. In fact, the Kuhnian account does no good to BT’s concerns exactly because Kuhn insisted on incommensurability of paradigms, so there could be no way to say that a new paradigm is more or less fruitful, adequate or inadequate. Instead, as Kahneman is well aware, only a gradual competition of research programmes and reforms following their improvement lead to truly improved insight into problem situations. Kahneman is no revolutionary – and he’s among the greatest for exactly this reason