My question is very much contained in the title. Most human behaviours are the object of many hypotheses based in the theory of evolution. Usually an hypothesis will arise as a model that explains a set of observations, and usually they are very smart and have a satisfying quality to them (for me at least).

But is there actually a way of turning them into scientific theories? Or has there been any study or discussion about how to measure their “scientific worth”? (For instance, say, some systematic way of calculating the ratio between the complicatedness of an hypothesis and its explanatory power.)

More concretely, what set me thinking about this is the recent publication of Troubled sleep: Night waking, breastfeeding and parent–offspring conflict by biologist David Haig (and the paper's wide-reaching vulgarisation, often with garish titles such as Study: Babies cry at night to keep parents from having sex), in which the author offers (roughly) that babies keep their parents awake to prevent them from procreating too early.

After quoting the headline to at least 10 friends in one week, and probably because of a recent column about sociologist Murray S. Davis's paper That's interesting! Towards a phenomenology of sociology and a sociology of phenomenology, I thought that this was perhaps too fun to be true and that I had to read the paper.

What I found is a very smart hypothesis based on existing models of intergenerational tradeoffs (e.g. is it more advantageous to make dozens of babies and not be able to care for them or to make only one and dedicate all one's attention to it) and observations about infant mortality in relation with interbirth interval in third-world countries.

But what I did not find is any suggestion about any way in which this hypothesis could be evaluated. So my question could be interpreted in more concrete terms as: Could anything be done to improve Haig's paper in this respect?

4 Answers 4


The difficulty in testing these things lies in it being an historical hypothesis and with the ethics of human experimentation. First of all, I should respond to alanf's answer: it contains a view of the nature/nurture debate that I consider a red herring...

Genes affect behaviour, and it is testable...

It might be an uncomfortable fact, but behaviour is causally affected by genes. How could it be otherwise? Genes determine our physical make up, and our physical make up is a huge factor in our behaviour (for example, congenital blindness will effect whether someone drives a car). Of course there are social factors and they matter, but just because a behaviour is influenced by culture it does not mean that it isn't influenced by genetics. When alanf says "you can't tell whether the behaviour is a result of genetic of cultural evolution" he is introducing a false dichotomy. Things can have more than one cause, and moreover it is also possible to investigate things with multiple complex causes scientifically.

The fact that we can answer some questions about genetics and behaviour doesn't mean that we can answer all of them. Likewise, not being able to answer some of them does not mean we cannot answer any: "not testable" is a wild generalisation.

Historical difficulties and ethical challenges

With all evolutionary explanations there is always the difficulty that we cannot turn the clock back and do experiments. We cannot actually do experiments to directly determine what would happen if X was different. We can only make inferences about it based on our current situation. We simply don't have experimental access to our ancestors, and we have to work around. This severely restricts what can be answered, as we often have to make inferences about dead ancestors based on their living descendants (this is true of all evolutionary theory).

With the study of non-human species there are fewer ethical barriers, one can intervene at the genetic level, or put animals in situations that would threaten autonomy were they human. With humans, ethics puts huge constraints on what experiments can be performed. One often finds situations where a hypothesis is falsifiable or verifiable in theory, but not in practice.

What this means

The consequence of this is that we must test hypotheses incredibly indirectly, making them rather weak and contingent. Still, one can add weight to them by testing lots of different associated components and use each of them as evidence in the larger argument. It's probably worth pointing out that medical science is quite atypical, here people do directly test: "does this treatment cause my patients to recover?" - elsewhere theories get support from being connected and consistent with lots of other theories and data.

In summary. Human evolutionary disciplines are particularly difficult as the experiments and theories used as evidence are quite "distant" from what is to be shown. There are significant constraints which stem from the historical nature of evolutionary theory and the ethics of human experimentation. These difficulties are probably why no experiment was expected of them.

So, to answer your final question, probably not, at least not if one wishes to behave ethically.


Theories about alleged influences of genes on human behaviour are not testable. The problem is that there are two sources of adaptive complexity in human behaviour: genes and cultural knowledge. Cultural knowledge need not be explicit. It is just any change in behaviour or ideas that one person can copy from another. This copying need not involve having an explanation for the behaviour. So if a person behaves in a particular way without being able to tell you why he is doing so, you can't tell whether the behaviour is a result of genetic of cultural evolution. If the person can tell you why he is doing something the problem is even worse since in that case there is some cultural knowledge involved.

Some people claim to be able to test this by doing experiments with identical twins, fraternal twins and unrelated children. These experiments are irrelevant because how people behave toward one another can be a culturally evolved response to appearance, physical abilities and other traits that may be genetic. For example, in a racist country how a person behaves toward you might depend on what genes you had for skin pigment but that behaviour would not be a result of genetic knowledge but rather of culturally evolved knowledge.

  • "are not testable" - do you include statistical testing here? Apr 23, 2014 at 14:08
  • 1
    Yes. In order to do a statistical test you would have to have some theory about how the probabilities for each cause depend on what else is happening. This is not possible for for behaviour because it can be changed by new ideas, which are unpredictable because the growth of knowledge is unpredictable. If we could predict what knowledge we will have tomorrow we would already have it.
    – alanf
    Apr 23, 2014 at 14:20

David Deutsch builds an evolutionary hypothesis about behaviour in Beginning of Infinity.

Very briefly, the role of genes in evolutionary biology is played by "memes" in his evolutionary explanation of behaviour. "Memes" here are basic patterns of behaviour that are fairly stable between generations. Memes are replicated within a society by social pressure, etc.

David Deutsch makes a few predictions regarding evolution of memes within societies, and, in particular, that societies' tend to evolve into one of two extremes - either a stable (conservative) society or creative one, with hardly any middle ground in the long run. Predictions like these make the "meme evolution" theory falsifiable.

Good read BTW, includes a bit of epistemology and a parody on Plato's dialogues.


It depends in part on what the area in question is, but in some cases you can construct simulations and models that make predictions about how the world should look if such a model was accurate, and then test for that statement instead.

I'm not aware of this being done specifically about behavior, but there was a story somewhat related that came out recently. The jist is scientists created a couple different models for human migration out of Africa, said that based on each models we should observe such-and-such genetic differences in population, and they observed such differences from one model. Thus, even though we couldn't observe the migrations directly, we can observe their consequences.

It should be noted that much of science is the same. We did not confirm general relativity directly, instead we looked for its consequences.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .