Absurdly irrational arguments can be hugely entertaining; but in the wrong context they can also be very dangerous, warns Robyn Dawes, and he uses numerous examples to highlight some worryingly common reasoning errors. Unfortunately the most pernicious philosophical errors, which he doesn’t highlight, are embodied in his own over-zealous scientific analysis.

Parts of this book are worth salvaging. The true force behind his charges of irrationality is the neglect of contrary evidence: you cannot establish a rule without checking evidence that may disconfirm it. This principle is captured in basic tenets of psychological research: include a control group and test the null hypothesis. Psychologists who ignore this advice – and there are many – arrive at very unsafe conclusions. A related principle is familiar to philosophers: avoid the fallacy of affirming the consequent. If a patient shows key signs of schizophrenia, resist the diagnosis until you’ve processed the statistics for the normal population: statistically, it may be more likely that the person is not schizophrenic. This wise advice is reinforced with examples throughout the book.

However, Dawes’ main contribution is to propose a simple, but misconceived, shortcut to resolving philosophical problems. He tries to show, by means of reductio ad absurdum, that unpalatable conclusions are logically absurd, not just factually wrong or implausible. He has a strong motivation for this: it’s easier to defeat an argument decisively on a point of logic than on a point of fact or plausibility.

He begins by defining irrationality very narrowly as self-contradiction. However, none of his examples strictly contradicts itself. To force a contradiction, he augments the arguments with some “obvious” implicit premises, processes the logic (or Bayesian probabilities), and rolls out an absurdity or improbability. In this manner, he discards dubious conclusions for being logically inconsistent, when he has surreptitiously planted the inconsistencies himself. In philosophy class, we had a name for this technique: cheating. The trick lies in augmenting your opponents’ premises to imply a contradiction, then pointing to the contradiction as proof of an error.

This technique is very easy and effective when the target is an obviously nutty argument. Daft conclusions always conflict with an obvious fact, and if your opponent accepts the obvious fact, then he or she is being logically inconsistent. The addition of the fatal implicit premise is uncontroversial.

However, Dawes applies this “implicit premise” technique to areas of deep controversy, neatly avoiding difficult philosophical issues. For instance, the politicians who deferred his advice to subsidize safe drug abuse to combat the spread of HIV, are branded not merely morally misguided but illogical. He is not bothered that his Republican opponents do not actually endorse the implicit premise (a naïve strain of consequentialism) that proves fatal to their argument: he believes it should be obvious to everyone. Consequently, on Dawes’ view, to reject his ethical advice is to flout the principles of logic. Logicians and moral philosophers would say otherwise.

Emotional convictions, according to Dawes, are irrelevant to the logic of an argument: people construct arguments for any number of personal reasons, he says, but the objective logic of the argument itself is untouched by such considerations. This is fine, unless the argument is value-based, as some of his test cases are. He doesn’t resist slipping his own values into the premises.

Next, Dawes introduces a pocket method for invalidating popular explanations in psychoanalysis, history and sociology: use statistics to prove that these retrospective rationalizations are grossly unscientific. First, generalize the causal explanation so it becomes a categorical hypothesis. Then examine other cases to see whether the hypothesis holds statistically. If not, the explanation is spurious. The truth is in the statistics.

For example, it is a mistake to attribute a serial killer’s homicidal fanaticism to bad parental relations, because lots of people have bad parental relations without becoming homicidal fanatics. Statistical reflection undermines the story’s deceptively plausible hypothesis. By the same token, Dawes undermines the forensic analysis of a specific plane crash, historical accounts of the rise of Nazism in Germany, and sociopsychological explanations of the Unabomber’s warped psychology. He concludes that retrospective stories, however intuitively appealing, are poor science and have only tentative speculative value. So much for the humanities, then.

Once again the error is introduced by Dawes. Most historians, psychoanalysts and whoever else, would refuse his first step. Dawes assumes that story-based explanations support simple generalization. Yet proper science demands a ceteris paribus (‘all else being equal’) clause; the examples Dawes condemns are too complex to support ceteris paribus generalization. Stories, being contextual, can handle this kind of complexity; general laws cannot. The truth is in the specifics.

By Dawes’ logic, we can prove (statistically) that no-one dies of a heart attack, and contributing factors don’t cause cot death or chronic fatigue. The weakness of his approach becomes frankly disturbing when he dismisses allegations of ritual abuse as “sexual abuse hysteria”, citing various statistics from studies of isolated abuse. If, on average, 60% of abused children fail to report sexual abuse, and all 10 children in one particular case have been silent up to the time of the investigation, then the likelihood that they have all been abused, but silent, is 0.006. Although the abuse is not impossible, it’s extraordinarily unlikely. So we can close the case without further investigation. Simple! Of course, for the statistics to hold, we have to ignore any systematic differences between ritual abuse and isolated abuse, research into cultural mind-control, and any other plausible story-based rationalization. Dawes is happy to do so. The truth, for him, lies in whichever numbers can prove his point. The misuse of statistics here is as dangerous as the faulty arguments he criticizes. In fact his own explanation for the phenomenon — teleological manipulation by psychologists — rests, as with most of his other examples, on precisely the kind of intuitive plausibility appeal that he disallows. Reductio ad absurdum, anyone?

Dawes’ writing suffers from an unsettling tone of opinionated pedagogy, and his statistical discussions are laborious. His critique of dubious conclusions is well-intentioned, and there are some morsels of genuinely useful advice here, but his formal analysis is fundamentally misconceived and — when he misapplies it — even more pernicious than the faulty reasoning he denounces.

Further reading: Once Upon A Number (Penguin, 1998), by the mathematician John Allen Paulos, carefully compares the merits and demerits of statistical logic and intuitively plausible story-based explanations, and serves as a balanced contrast for Dawes’ misguided radical formalism.

© Sam Brown, 2001

Sam Brown is writing up a PhD on the neuroscience of emotional intelligence. He has an MA in Philosophy and an MPhil in Cognitive Science from the University of Dundee. He has particular interest in neuroscience and naturalist philosophy, focusing on the pathology of emotional disorders and the epistemology of intuition. He works as an IT tutor and database consultant.

Categories: Philosophical, General