Skip to main content

Mindware - Richard Nisbett *****

There's no doubt that Richard Nisbett's book, subtitled 'tools for smart thinking' is great, despite two issues. I want to get those issues out of the way first before we get onto the good stuff, with which it is packed. One issue is the writing style. This is a touch clumsy and could do with a little professional help. Nisbett has a tendency to overuse unnecessary jargon in sentences like this:
Our construal of objects and events is influenced not just by the schemas that are activated in particular contexts, but by the framing of judgments we have to make.
Nor ideally worded. The second issue I suspect comes more from the publisher, which is the attempt to frame this book (sorry, couldn't resist the italics) as a self-help title as much as popular science. It doesn't work particularly well as a practical self-help toolkit - it's not structured in a way to make this a good use, particularly because a large part of the book is focused on how we get things wrong, rather than how to do things better.

But what makes this book a pure delight is the way that it analyses our human take on the world and shows the flaws in the typical ways that we think which, if overcome, would enable us to make better decisions. Some of this is fairly theoretical, starting with the nature of inference and exploring the holes in the Popperian disdain for inference, but there's also an exploration of the results of a plethora of experiments which take everyday decisions and situations and try to understand what is happening.

One great example is over the analytic input of the subconscious. Nisbett shows us how the old saw about leaving a problem overnight to reach a better decision really works. He describes an experiment where different people are shown a range of apartments and asked to decide which is 'best'. (As always with psychology experiments, there is room for questioning here as 'best' is so subjective with accommodation, but the experimenters try their best using a series of criteria against which each apartment is scored.) The subjects are divided into three groups. Some have to make an almost instant decision, others are allowed to weigh up the pros and cons, while a third group doesn't actually think about the problem but makes their decision after sleeping on it. By far the best results come from the third group, while the 'instant' decision maker are as effective as those weighing up the options. This neatly takes the wind out of the usual moan that people make a decision about house buying far too quickly.  The subconscious can do a surprising amount of the heavy lifting for us (though, as Nisbett points out, it's hopeless at doing sums).

A very interesting section for those who are fans of Freakonomics and its successor books is where Nisbett tears apart a technique often used in the 'Freak' decision process - multiple regression analysis, where the idea is to analyse data by correcting for various unwanted variables, leaving the one being studied dominant. Nisbett, tongue in cheek, refers to it as 'eekonomics'. It's an infamously poor approach (the reason why cohort studies on diet etc. are so difficult to use), because it's almost impossible to be sure you've allowed for all the variables, and the impact of some can be little more than guesswork. Instead, Nisbett suggests, it would be much better to do far more experimental work in these fields, with proper double blind controls, even though he admits that's not always possible.

With other sections on correlation versus causality, sample sizes, the nature of logic and dialectic, and more, there's plenty of meat here in a truly fascinating read about the nature of human decision making, where it goes wrong and how (at least in principle) we could do it better. It's not always big stuff. He points out how we often fail to deal with the way that money we've spent is already written off. There's no point carrying on with something you are now getting no value out of due to changing circumstances, or a bad initial decision, just because you've already spent a lot on it. Yet we all tend to do it (as do governments).

The part contrasting 'Western' logic with 'Eastern' dialectic towards the end of the book is probably the least satisfactory as it doesn't really explain how the dialectic approach can produce specific useful results rather than fuzzy statements, but it's still interesting. Overall, an ideal book for anyone who raises an eyebrow at statistics in the press. If, for example, you are a fan of Radio 4's More or Less, this book takes the whole look at the way we make informed decisions using numbers to a new level.


Review by Brian Clegg


Popular posts from this blog

Astrophysics for People in a Hurry – Neil deGrasse Tyson *****

When I reviewed James Binney’s Astrophysics: A Very Short Introduction earlier this year, I observed that the very word ‘astrophysics’ in a book’s title is liable to deter many readers from buying it. As a former astrophysicist myself, I’ve never really understood why it’s considered such a scary word, but that’s the way it is. So I was pleasantly surprised to learn, from Wikipedia, that this new book by Neil deGrasse Tyson ‘topped The New York Times non-fiction bestseller list for four weeks in the middle of 2017’.

Like James Binney, Tyson is a professional astrophysicist with a string of research papers to his name – but he’s also one of America’s top science popularisers, and that’s the hat he’s wearing in this book. While Binney addresses an already-physics-literate audience, Tyson sets his sights on a much wider readership. It’s actually very brave – and honest – of him to give physics such prominent billing; the book could easily have been given a more reader-friendly title such …

Once upon and Algorithm - Martin Erwig ***

I've been itching to start reading this book for some time, as the premise was so intriguing - to inform the reader about computer science and algorithms using stories as analogies to understand the process.

This is exactly what Martin Erwig does, starting (as the cover suggests) with Hansel and Gretel, and then bringing in Sherlock Holmes (and particularly The Hound of the Baskervilles), Indiana Jones, the song 'Over the Rainbow' (more on that in a moment), Groundhog Day, Back to the Future and Harry Potter.

The idea is to show how some aspect of the story - in the case of Hansel and Gretel, laying a trail of stones/breadcrumbs, then attempting to follow them home - can be seen as a kind of algorithm or computation and gradually adding in computing standards, such as searching, queues and lists, loops, recursion and more.

This really would have been a brilliant book if Erwig had got himself a co-author who knew how to write for the public, but sadly the style is mostly heavy…

A turnround from Tyson

I am delighted that one of our reviewers has been able to give a five star review to Neil deGrasse Tyson's latest book. The astrophysicist has taken over Carl Sagan's old post as the number one science populariser in the US, but his written output in the past has been patchy, to say the least.

There have been at least two significant problems. One is dubious history of science. For example, in the cases of both Galileo and Bruno he has passed on undiluted the comic book version of history where Galileo is persecuted for mentioned heliocentricity (rather than his disastrous political handling of the  pope) and mutters 'Eppur si muove!' at his trial, and Bruno is burned at the stake for his advanced scientific ideas (both misrepresentations). Some argue that it getting history of science accurate doesn't matter if we get the right message about science across - but if we are prepared to distort historical data, why should anyone take scientific data seriously?

The o…