Gary Marcus (1970–)

I really like the engineering bent here as it makes sense from my life experience: human brainpower has necessarily evolved as a kluge and can be analyzed as an imperfect system. Beware the author's opinions; if you are pro-Bush or creationist, you may not like his asides and assumptions upon the reader. Having analyzed the system, the author gives us a handy set of steps to avoid our brain's inefficiencies:

1. Whenever possible, consider alternative hypotheses.
2. Reframe the question.
3. show more Always remember that correlation does not entail causation.
4. Never forget the size of your sample.
5. Anticipate your own impulsivity and pre-commit.
6. Don't just set goals. Make contingency plans.
7. Whenever possible, don't make important decisions when you are tired or have other things on your mind.
8. Always weigh benefits against costs. (opportunity costs)
9. Imagine that your decisions may be spot-checked.
10. Distance yourself.
11. Beware the vivid, the personal, and the anecdotal.
12. Pick your spots. (Reserve your most careful decision
making for the choices that matter most.)
13. Try to be rational. show less

Exercise for the reader: Define "rational thought."

That was the one problem that I had in reading this well-researched, thought-provoking book. How do we know that certain conclusions are "rational," others "irrational"? We know because, ultimately, our emotions tell us what is valuable and what is not. From there, we use "rational" thought to achieve those ends. But if we changed the emotions, would not what is "rational" change? Therefore any dichotomy between "reason" and "emotion" is show more largely false, and there are places where author Gary Marcus fails to realize that he has made an unmotivated argument.

Yet the general thesis is patently true: Evolution does not produce the best solution to a problem; it merely follows a series of branches that lead in the direction of some sort of "improvement" (properly, something more "fit" than what it replaces). So our whole bodies and minds are full of kludges (a spelling Marcus avoids because that isn't how he pronounces it, but it's how every engineer I know pronounces the word). If you avoid discussions of what is "rational" and simply ask, "Is this the best way to achieve what we want?" then there can be no question: our brains are kludgey.

I do have one other gripe, mostly with the material at the end. Marcus assumes that a lot of things that file under "abnormal psychology" (e.g. depression) are the result of kludges, and maladaptive. This really, truly does not follow. A trait can be adaptive or maladaptive depending on context -- or a gene can be adaptive if unreinforced and maladaptive if reinforced. (The classic example of this is sickle cell anemia: One copy of the relevant gene and you're semi-immune to malaria; two copies and you're probably dead.) Major depression, or extreme autism, or out-of-control anxiety, are maladaptive. But mild autism, or minor depression, are associated with great originality and creativity. So our mental "problems" may also be our strengths.

Of course, that largely proves the point: If our brains weren't kludged up, we could have the benefits of depression and autism without the social failures and the sadness.

Something for the next round of evolution to work on, no doubt. show less

Tries to refer to too many disparate things, and ends up not really covering any topic in sufficient depth, or with sufficient rigor. The basic thesis is that the human mind, especially in the ways it doesn't work very well (cognitive biases, mental illness, even things like the haphazard pursuit of pleasure) show the imperfections of evolution: the ultimate kluge.

Trouble is, he pretty well says everything he needs to say in the first few chapters, and thereafter the book gets more and more show more scatty. Hard not to imagine him sitting there writing it, going "Ooh, ooh, I know, I can talk about X!" and five minutes later "Ooh, ooh I can say Y! And Z. And maybe I can winkle A in there too." Whether they advance the argument or not doesn't seem to matter. Neither does questions of how well he can discuss them -- how, for example, do you take a discussion of ambiguity in language seriously, when he doesn't even mention metaphor? (His lack of awareness of the fundamentals of the philosophy of language doesn't stop him trying to rope the subject in. Or taking a swing at Noam Chomsky along the way.)

Ultimately the problem is that he just hasn't really planned the book well enough. It's entertaining enough, deals (poorly, but accessibly) with some interesting and important concepts. It certainly shouldn't frighten anyone off popular science books. But others have done it much, much better -- try Daniel Kahnemann (for cognitive biases), Guy Deutscher (language), Steven Pinker (ditto), Robert Sapolsky (evolutionary physiology) and Vilayanur Ramachandran (general weirdness with the way the brain is wired). You'll learn more, and be much more entertained. Or if you like your science light and congenial, try Robert Winston and Oliver Sacks. Better writers, and more knowledgeable.

This book, however enthusiastically intentioned, is itself quite a significant kluge. It won't scare the horses, but you can easily find better. show less

The book reflects a belief I've been having for some time now, mainly that we are currently barking up the wrong tree when it comes to AI.
We are making amazing tools, thats certain, but most people see the latest improvements in solving a very specific problem and extrapolate from that to believe we are close to having a general purpose AI, or True AI.

It compares classical AI with modern ML driven AI and talks about the strengths and weaknesses of both. Modern AI is amazing, but the flawed show more classical AI with all of it's impracticality is the only way we can ever achieve AGI, not by improving neural networks or making ML more efficient.

My belief is that when it comes to developing true AI we need to go back to the drawing board and classical AI is better at helping achieve it than the modern approaches that are being overhyped by Media with hyperboles, clickbait and in some cases fraud. If you share this belief than you'll enjoy this book. show less