Kluge: The Haphazard Construction of the Human Mind (Houghton Mifflin; 2008)

In the same vein, Steven Pinker has argued that "the parts of the mind that allow us to see are indeed well engineered, and there is no reason to think that the quality of engineering progressively deteriorates as the information flows upstream to the faculties that interpret and act on what we see."

This book will present a rather different view. Although no reasonable scholar would doubt the fact that natural selection can produce superlatively well engineered functional designs, it is also clear that superlative engineering is by no means guaranteed. What I will argue, in contrast to most economists, Bayesians, and evolutionary psychologists, is that the human mind is no less of a kluge than the body. And if that's true, our very understanding of ourselves — of human nature — must be reconsidered.

In the extensive literature on evolutionary psychology, I know of only a few aspects of the human mind that have been attributed to genuine quirks. Although most evolutionary psychologists recognize the possibility of suboptimal evolution in principle, in practice, when human errors are discussed, it's almost always to explain why something apparently nonadaptive actually turns out to be well engineered.

Take, for example, infanticide. Nobody would argue that infanticide is morally justifiable, but why does it happen at all? From the perspective of evolution, infanticide is not just immoral, but puzzling. If we exist essentially as gene-propagating vessels (as Richard Dawkins has argued), why would any parent murder his or her own child? Martin Daly and Margo Wilson have argued that from the gene's-eye view, infanticide makes sense only in a very limited set of circumstances: when the parent is not actually related by blood to the child (stepparents, for example), when a male parent is in doubt about paternity, or when a mother is not currently in a position to take good care of the child, yet has prospects for taking better care of some future child (say, because the current infant was born hopelessly unhealthy). As Daly and Wilson have shown, patterns of murder and child abuse fit well with these hypotheses.

Or consider the somewhat unsurprising fact that men (but not women) systematically tend to overinterpret the sexual intentions of potential mates.* Is this simply a matter of wishful thinking? Not at all, argue the evolutionary psychologists Martie Haselton and David Buss. Instead, it's a highly efficient strategy shaped by natural selection, a cognitive error reinforced by nature. Strategies that lead to greater reproductive success spread (by definition) widely throughout the population, and ancestral males who tended to read too much into the signals given by possible partners would have more opportunities to reproduce than would their more cautious counterparts, who likely failed to identify bona fide opportunities. From the gene's-eye view, it was well worth it for our male ancestors to take the risk of overinterpretation because gaining an extra reproductive opportunity far outweighs the downside, such as damage to self-esteem or reputation, of perceiving opportunity where there is none. What looks like a bug, a systematic bias in interpreting the motives of other human beings, might in this case actually be a positive feature.

When reading clever, carefully argued examples like this one, it's

*Except, tellingly, those of their sisters.

easy to get caught up in the excitement, to think that behind every human quirk or malfunction is a truly adaptive strategy. Underpinning such examples is a bold premise: that optimization is the inevitable outcome of evolution. But optimization is not an inevitable outcome of evolution, just a possible one. Some apparent bugs may turn out to be advantages, but — as the spine and inverted retina attest — some bugs may be genuinely suboptimal and remain in place because evolution just didn't find a better way.

Natural selection, the key mechanism of evolution, is only as good as the random mutations that arise. If a given mutation is beneficial, it may propagate, but the most beneficial mutations imaginable sometimes, alas, never appear. As an old saying puts it, "Chance proposes and nature disposes"; a mutation that does not arise cannot be selected for. If the right set of genes falls into place, natural selection will likely promote the spread of those genes, but if they don't happen to occur, all evolution can do is select the next best thing that's available.

To think about this, it helps to start with the idea of evolution as mountain climbing. Richard Dawkins, for example, has noted that there is little chance that evolution would assemble any complex creature or organ (say, the eye) overnight — too many lucky chance mutations would need to occur simultaneously. But it is possible to achieve perfection incrementally. In the vivid words of Dawkins,