On RATIONALITY (the book)

The author of this book is a cognitive psychologist who has written, among other topics, about The Language Instinct, How the Mind Works, The Blank Slate, The Better Angels of Our Nature, and Enlightenment Now, and who has become a public intellectual of sorts (having to show for it two nominations for the Pulitzer Prize and an entry into Time’s 100 Most Influential People in the World Today list.)

We are promised to learn WHAT IT IS, WHY IT SEEMS SCARCE, and WHY IT MATTERS (one of the subtitles and also the title of the concluding chapter). Rationality, that is. The book grew out of a university course that explored “the nature of rationality and the puzzle of why it seems to be so scarce.” (p. xiv) Pinker admits that he loves “to teach the arresting, Nobel Prize-winning discoveries of the infirmities that afflict human reason, and consider them among the deepest gifts to knowledge that our science has contributed. … Yet as a cognitive scientist I cannot accept the cynical view the human brain is a basket of delusions.” (p. xiv) In fact, we are assured that “RATIONALITY (the book) is, above all, an affirmation of rationality.” (p. xvi) He got me at that. Almost.

The book is uneven, and the unevenness springs from five sources. First, Pinker’s knowledge of cognitive psychology is reflected in the (sub)titles of many of the chapters that constitute it (logic and critical thinking, probability and randomness, Bayesian reasoning, correlation and causation, etc.). This is the whole catalogue of “the intellectual tools of sound reasoning” (p. xv), or of the classic enlightenment apparatus, and surely it is no coincidence that his previous book was about (the) ENLIGHTENMENT. Second, Pinker’s knowledge of the meaning of rationality in economics is wanting. (His understanding is a very simplistic version of modern economics). Third, Pinker’s knowledge of alternative conceptions of rationality (broadly delineated by the term ecological rationality) is wanting. (He seems to be mostly unaware of ongoing debates between those that are firm believers in the biases-and-heuristics paradigm and those that aren’t). Fourth, Pinker seems to have a major blind-spot regarding the evidentiary value of the findings that psychology has produced. Granted, cognitive psychologists have to worry about that less than social psychologists but replicability has become a major issue across all social sciences including the various psychology sub-discipines. Fifth, there is little indication that Pinker is across current debates about the proper way to do experiments and the proper use of statistics, not withstanding some perfunctory — but questionable — references (Ioannidis 2005; see Schimmack on that classic). This fifth point is intriguing because Pinker started out as an experimental cognitive psychologist. Alas, that was apparently many books back. The price you pay, I guess, for being a public intellectual of sorts for too many years and feel tempted, or prompted, to comment on God and the world.

Let’s illustrate the blind-spots with some examples.

According to Pinker, chapters 3 to 9 are dedicated to explain “the most widely applicable normative tools of reason” (p. 7) Normative means in this context “the ways an intelligent agent ought to reason, given its goals and the world in which it lives.” (ibid.) Note that the point of departure is hence very much small-world-ish. And sure enough that is where Pinker starts out in chapter 3. He reviews Cognitive-Reflection-Test examples (including the alleged failure to understand exponential growth), the Wason selection task, the Monty-Hall-Let’s-Make-A-Deal example, and the conjunction fallacy. All these are well-worn brain teasers that have been used to demonstrate people’s fundamental inability to deal with simple logic and probability problems, logic and probability being among those normative tools of reason that Pinker is fond of. All of these brain-teasers have been used to demonstrate how much people struggle even with allegedly simple problems of logic and probability. Pinker calls such errors in reasoning, like many others, cognitive illusions and compares them, like many others, to optical illusions. The problem here is two-fold. First, while optical illusions are a fairly robust phenomenon, cognitive illusions are not. Second, that cognitive illusions are not that robust a phenomenon and that in effect they can be made to appears, disappear, or even reverse, is a well-established fact and has been a bone of contention at least since the very public 1996 pissing match that Kahneman & Tversky (KT) had with Gigerenzer, in The Psychological Review no less. (Pinker mentions these papers by but does not elaborate clearly on the issues, or that they are ongoing.)

Pinker spends several pages in this chapter on that old KT chestnut called the conjunction fallacy, made famous by the “Linda problem” that supposedly has shown that people are not able to understand, and reason according to, basic laws of probability. Yet, people like Gigerenzer and Hertwig have pointed out that wording and framing of the problem matters for the outcome, and Charness, Karni, & Levin demonstrated, more than a decade ago, that mild incentivization and the ability to communicate with others makes a dramatic difference, too. While KT found an error rate of 85 percent for unincentivized UBC students, Charness and his colleagues found, for modestly incentivized trios of UCSB students, an error rate of 10 percent. Not perfect, I grant you that, but I’d argue pretty darn good.

What we see in this case is reflecting what happens in other parts of the book.

In Chapter 4 Pinker provides a primer on PROBABILITIES AND RANDOMNESS, including five meanings that the word “probability” has, how to represent conjunction and disjunction of events, and how to compute the conditional probability of an event, and prior and posterior probabilities. This is pretty much standard fare although there is also a discussion of how probabilities can be affected in detrimental ways by sensationalist media coverage, leading to sub-optimal decision-making (at least in expected-value terms).

In Chapter 5 Pinker provides a brief primer on BELIEFS AND EVIDENCE, subtitled BAYESIAN REASONING. Building on the concepts of prior and posterior probabilities, Pinker introduces Bayes’s Rule as the posterchild of how to do Baysian reasoning right. He discusses the importance of getting the base rate right. He argues that it is the defiance of Bayesian reasoning (and there in particular the neglect of proper priors) that is the reason for the replicability crisis. En route he notes that, “most findings in psychology, as it happens, do replicate.” (p. 160) Really? Don’t tell these folks. Or this guy. Towards the end he discusses natural frequencies and the work of people like Gigerenzer and Cosmides and Tooby. Good that.

In Chapter 6 Pinker provides a brief primer on RISK AND REWARD, subtitled RATIONAL CHOICE AND EXPECTED UTILITY. We learn nuggets such as “One of the most hated theories of our time is known in different versions as rational choice, rational actor, expected utility, and Homo economicus.” (p. 173) Stressing that “the beauty of the theory is that it takes off from a few easy-to-swallow axioms” (p. 175), Pinker then presents one version of the theory before discussing how useful utility is and how irrational various violations of axioms are. Not that much he concludes, stating — quite in line with plenty of evidence — here and here and here and here and here — that “von Neuman and Morgenstern may deserve the last laugh” (p. 197) after all.

In Chapter 7 Pinker provides a brief primer on HITS AND FALSE ALARMS, subtitled SIGNGAL DETECTION AND STATISTICAL DECISION THEORY. It’s a useful primer on false (true) positives and negatives and the consideration that go into the calculation of the appropriate cut-off. Pinker also provides a good summary of how these concepts apply to the courtroom. For another such discussion see here. While there is a discussion of the questionability of certain norms, the discussion does not get close to what it should be (see here and here). There is a useful discussion of what significance tests do deliver and what they don’t (pp. 224–6).

In Chapter 8, titled SELF AND OTHERS, the maistro introduces us very briefly to, GAME THEORY (so the subtitle). This is standard fare (zero-sum games, coordination games, games of chicken, and the prisoners’ dilemma/tragedy of the commons, and their normal-form representations). There are brief discussions of what makes game theory really interesting: the effects of playing the stage game repeatedly, for example. Which Pinker calls a “game-changer” (p. 241) without going into any detail. Likewise with “enforceable contracts and regulations” (p. 244). Areference to Hobbes is also there. Chapter 8 is surely a prime candidate for weakest chapter of the book.

In Chapter 9, Pinker provides a brief primer on CORRELATION AND CAUSATION, sans subtitle. Again, this is very much standard fare, at least for everyone who has read, for example, Cunningham’s mixtape (highly recommended!). There is a brief discussion about randomized experiments and their limits, as well as a brief discussion of causal-inference methods such as regression discontinuity and instrumental variables that allow us to tease out causal effects from natural experiments hidden in observational data. This chapter is the last of the seven chapters “intended to equip you with what I think are the most important tools of rationality.” (p. 281)

Chapters 10 (“WHAT’S WRONG WITH PEOPLE?”) and 11 (“WHY RATIONALITY MATTERS”) are the capstone chapters. Unfortunately they are only loosely informed by the reviews in the preceding chapters.

Chapter 10, we learn “is the chapter most of you have been waiting for. I know this from conversations and correspondence. As soon as I mention the topic of rationality, people ask me why humanity appears to be losing its mind.” (p. 283) Says Pinker, referring to “a glorious milestone in the history of rationality … vaccines likely to end a deadly plague are being administered less than a year after the plague emerged. Yet, in that same year, the Covid-19 pandemic set off a carnival of cockamamie conspiracy theories: that the disease was a bioweapon engineered in a Chinese lab, … “ (lit. cit.) Only of course, it is no longer that far-fetched to believe that a bio-security breach in a Chinese lab was in fact responsible for the catastrophe that followed world-wide.

But, yes, “Covid quackery, climate denial, and conspiracy theories are symptoms of what some are calling an ‘epistemological crisis’ and a ‘post-truth era.’ Another symptom is fake news … Also rampant are beliefs in ghouls, black magic, and other superstitutions. … “ (pp. 284–5) How can we explain, and fight, “this pandemic of poppycock”? (p. 286)

Towards an answer, Pinker dives into Motivated Reasoning and the Myside Bias which are somewhat immune against Bayesian updating because, “It is difficult to get a man to understand something, when his salary depends upon his not understanding it.” (p. 290, Pinker quoting Upton Sinclair) Pinker argues that “there is a different and more perverse rationality to the myside bias, coming not from Bayes’s rule but from game theory.” (p. 297) In essence he argues that outlandish claims allow one to identify like-minded folks: “Unfortunately, what’s rational for each of us seeking acceptance in a clique is not so rational for all of us in a democracy seeking the best understanding of the world. Our problem is that we are trapped in a Tragedy of the Commons.” (p. 298) Which strikes me as a pretty lazy way of looking at the world, since game theorists and experimentalists for that matter do understand under what circumstances tragedies of the commons can be overcome. Elinor Ostrom got a Nobel for her work on the issue.

To appeal for “a valorization of the norm of rationality itself” (p. 311) is not a constructive proposal. Nor is the exhortation to break up “universities’ suffocating left-wing monoculture, with its punishment of students and professors who question dogmas on gender, race, culture, genetics, colonialism, and sexual identity and orientation. Universities have turned themselves into laughingstocks for their assaults on common sense … ” (p. 313) Wtaf.

Ultimately, Pinker — the public intellectual — has really no answer to the question that he asks in the title of the chapter. Not that it is an easy question to answer but I wish he would have tried harder. And apparently I am not the only one.

In Chapter 11, titled with the sub-title of the book, Pinker proceeds to tell us why, in his view, rationality matters and whether indeed “the conscious application of reason actually improves our lives and makes the world a better place.” (p. 319) Not surprisingly Pinker’s answer is affirmative.

He enumerates the advantages, for one’s personal wealth and health, of understanding exponential growth, Bayesian thinking, and probabilistic concepts more generally. He mentions Tim Farley’s What’s The Harm website and twitter feed and the anecdotal evidence that he has amassed of “the economic consequences from blunders in critical thinking” (p. 322) He also discusses the work of Bruine de Bruin, Parker, & Fischhoff who have demonstrated that “people’s reasoning skills did indeed predict their life outcomes: the fewer fallacies in reasoning, the fewer debacles in life.” (p. 323)

Pinker then goes on to document material progress that was facilitated on a societal level. His argument mirrors the arguments in Factfulness: Ten Reasons We’re Wrong About The World — And Why Things Are Better Than You Think by Hans Rosling, Ola Rosling, and Anna Rosling Ronnlund (e.g., dramatically increased life expectancy, reduction of poverty, the “great material enrichment of humanity (that) began with the Industrial Revolution of the nineteenth century” (p. 326), the reduction in number and lethality of wars, and the moral progress that we made since the dark ages (e.g., regarding the acceptability of slavery.)

I am with Pinker on this assessment although the fact that people like Trump succeed (temporarily) and that thugs like Putin, Xie, and Kim Jong Un make it occasionally seem as if spaceship Earth is going to hell in a handbasket, provides fertile ground for uncertainty about where we are headed. (To what extent this perception of uncertainty is a function of social media is an interesting question.) It is surely re-assuring that Trump’s attempted coup got reined in then, and continues to be reined in now, and that tyrants like Putin get shown their limits.

Consider making your opinion known by applauding, or commenting, or following me here, or on Facebook.