The Evolutionary Psychology FAQ

Edward H. Hagen, Institute for Theoretical Biology, Berlin

What about spandrels?

Evolutionary adaptation is a special and onerous concept that should not be used unnecessarily, and an effect should not be called a function unless it is clearly produced by design and not by chance. When recognized, adaptation should be attributed to no higher a level of organization than is demanded by the evidence. George C. Williams, opening words of Adaptation and Natural Selection, 1966.

Repeating an argument made earlier by George Williams (1966), a couple of Harvard guys (Gould and Lewontin 1979) got a lot of mileage out of the observation that many organism 'traits' are not adaptations, but simply incidental byproducts. This is obviously true because the vague term 'trait' can refer to any conceivable aspect of an organism, like the grumbling of its stomach or snoring. Unlike the Harvard guys, Williams offered a way to separate the wheat from the chaff: adaptations must exhibit evidence of design.

Williams' criterion is critical. Without it, it is possible to assign every molecule, cell, and tissue in the body to a spandrel. Consider this thought experiment. A CAT scan produces a detailed 2D image of a cross-section of the body, like slicing open an orange and photographing the freshly revealed surface. By taking a large number of 2D scans perpendicularly along the length of the body and inputting the stack of images into a computer, one can build up an amazing 3D view of the body's internal anatomy, just as one could build up a 3D view of the internal structure of an entire orange by slicing it into a large number of thin sections, photographing each one, and scanning the stack of 2D photographs into 3D software.

Imagine that a team of scientists who know nothing of anatomy gets hold of a large stack of CAT scans of an entire human body, revealing all its tissues in detailed cross-sectional images. The scientists begin analyzing the body using the 2D images, not realizing that the individual scans can be composited into a single, 3D model. Instead, each scientist gets her own 2D image to analyze independently from the others. Each scientist develops sophisticated statistical models of the patterns of light and dark on her image, scribbling down elegant equations describing the image's shapes and curves. The statistics and equations developed by the team are a rigorous, factual description of the entire body, but it is a description that is empty. The patterns of tissues revealed by the CAT scans are, if considered alone, spandrels of the true, underlying functional organization that the team has failed to recognize. Ask the wrong questions, and virtually all normal body tissues will be part of a spandrel. Ask the right questions, and most normal body tissues will be recognized as playing a vital, functional role in the survival and reproduction of the organism.

But wait! Isn't it somehow scientifically dangerous (e.g., Gould 1997), or at least embarrassing, to over hypothesize adaptive functions for traits that might not be adaptations? Nope. Such mistakes are no more scientifically dangerous than the opposite—under attributing function. Consider this example. Lumps of tissue at the back of the throat often become infected and therefore are (or were) frequently removed by surgery. Which scientific response do you prefer?: (1) Mock any suggestion that the lumps (tonsils) might serve an important function by loudly insisting that not all traits have adaptive functions; or (2) generate and test as many functional hypotheses as you can think of to make sure that by removing the tonsils no lasting harm is done to the patient?

Just as anatomists have made mistakes, EPs will sometimes over attribute function to psychological phenomena that aren't really adaptations (my work, Hagen 1999, 2003, could be a prime example) and sometimes they will fail to recognize genuine functions. On one level I find it bizarre that Gould, Lewontin, or anyone else could possibly fear the "dangers and fallacies" (Gould 1997) of what is in fact routine science with an outstanding record—proposing and testing functional hypotheses for organism structure. On another level, however, I understand Gould and Lewontin's distress. EP has rudely broken into the cathedral of the mind, spray-painting 'sex', 'violence', and 'competition' across their beloved spandrels.

Make no mistake, many spandrels, which EP terms byproducts, are enormously important in their own right. Symons (1979), a founder of EP, argued, for example, that the capacity of women to orgasm is a byproduct of a male adaptation for orgasm. It is an unstated premise of EP, however, that, by failing to recognize the evolved functional organization of the brain, psychology and the rest of the human behavioral sciences, like our team of misguided scientists, are condemned to study nothing but spandrels.


Gould SJ (1997) The exaptive excellence of spandrels as a term and prototype. Proceedings of the National Academy of Sciences 94:10750-10755.

Copyright 1999-2002 Edward H. Hagen