The Evolutionary Psychology FAQ

Edward H. Hagen, Institute for Theoretical Biology, Berlin

What about learning?

Learning is extremely widespread--most organisms, including plants and bacteria, probably have the capability to learn, at least to some degree. Learning is often viewed as an explanation which competes with evolved psychology. If certain behaviors or ideas can be learned, it is claimed, then there is no need for an evolutionary explanation. This is wrong on two counts. First, learning requires specialized structures. Dirt doesn't learn. Rocks don't learn. The ability of any organism to learn requires that it has evolved adaptations which permit learning. Learning adaptations must exhibit a number of sophisticated properties. The quantity of information that any organism could, in principle, learn, is unimaginably vast. Learning capacity is finite, however--if an organism is going to use the information it has learned in any useful way, its learning adaptations require extraordinarily effectively 'filters' so that only 'useful' information is learned. Learning also requires that the organism has tissues which can change state. That is, upon exposure to useful information, these tissues can record it. Finally, the organism must be able to retrieve and use information that has been learned, but do so only when the information is needed. Learning is a very sophisticated ability that would be impossible without specialized adaptations.

Second, imagine that an organism had evolved a perfect learning adaptation. This organism could learn everything. To qualify as an organism, however, it must still reproduce. Recall that reproduction is an enormously complex process, involving countless energy consuming transformations of the world. Learning also requires numerous energy consuming transformations of the world (the world, in this case, is the organism's brain). The energy needed for learning is not trivial, and it is unlikely that organisms would evolve the ability to learn if the energy costs of learning were not compensated by enhanced reproduction. Would a perfect learning mechanism, one that could and did learn everything, enhance reproduction? Very probably not. A perfect learning mechanism would pose the same problem noted above: only a tiny fraction of information is useful for reproduction. Because an organism with a perfect learning mechanism still needs to reproduce--presumably using what it has learned--it would need to identify precisely that information which was needed to reproduce. A difficult problem to say the least.

Perhaps an organism could evolve a learning adaptation which learned exactly the information that it needed to reproduce, and nothing more. That is, it had the ability to learn everything about its own reproductive ecology. Is this possible? Such an organism would obviously need to observe members of its own species reproduce--learning everything it needed to know--or it would need to obtain this information from some other source, perhaps other members of the species who would pass on this critical knowledge. Learning everything by observation is very probably too costly in time and effort. Individuals just wouldn't be able to observe everything they needed to observe. Consider a heavily researched topic in evolutionary psychology: male mate preferences. Evolutionary psychologists have proposed that human males prefer, all else equal, to establish long-term mateships with younger (sexually mature) females. The reason this preference evolved is that males who married younger women had more offspring on average, over evolutionary time. Considerable research has demonstrated that a sexual preference for younger women is a male universal (teenage males tend to prefer women who are slightly older than themselves, and thus fertile, but these women are still young). How could an organism learn this type of information? It would have to very carefully observe lifetime reproductive outcomes for many different mateships, controlling for other variables like health, access to resources, etc. This is obviously impossible. Life is just too short to learn this kind of information, yet men have precisely the preferences predicted by evolutionary theory. These preferences must have evolved.

Obtaining all the information one needs to reproduce from other members of one's species is problematic as well. Evolutionary theory strongly implies that individuals will have conflicts of interest. Obtaining all the information one needs to reproduce from potential competitors is, in itself, a very poor reproductive strategy. There will be strong selection pressures on the providers of such information to manipulate the provided information in ways that benefit themselves reproductively, quite likely at the reproductive expense of the receivers. It is extremely unlikely that learning adaptations could evolve which were completely dependent on other individuals to provide the information necessary for reproduction.

None of this is meant to devalue learning. Learning is a critical strategy for many, if not most, organisms. The environment changes rapidly, and learning is an excellent way to adjust one's reproductive strategies to increase their probability of success. Evolutionary theory indicates, however, that learning is almost certainly structured by natural selection to focus on domains that are critical to reproduction. For example, some animals are dangerous, and others not. Because an organism may find itself in an environment with animals not encountered by its ancestors, it will need to learn which animals are dangerous, and which are not. Adaptations for learning about animals may well be innate, whereas information about which particular animals are dangerous is almost certainly learned.

Copyright 1999-2002 Edward H. Hagen