"A common charge made against evolutionary hypotheses is that they are inevitably ‘just so’ stories. Behaviour, of course, doesn’t fossilise, nor can we return to the distant past to make the observations that would allow us to falsify such hypotheses. However, even concerning hypotheses that are equally untestable,
some are more plausible than others. Thornhill and Thornhill’s hypothesis (1992) of the existence of a gene for rape really does seem to rely upon the existence of a selective advantage gained by our distant ancestors who engaged in forced mating. The claim that this behaviour did confer that selective advantage is possibly true, but we have reasons to doubt its plausibility. The offspring of such matings may have been abandoned. Those engaging in forced matings may have been expelled from social groups or punished
in other ways. We just don’t know, and it seems we can’t find out.

But contrast the Thornhills’ hypothesis with the claim that predators tend to have worse peripheral vision than prey because it is more calamitous for the latter to be surprised than it is for the former. Of course, this hypothesis also depends upon the facts. There could be some currently unknown factor, which better explains differences in facial structure between carnivores and herbivores. Nonetheless, it seems very likely that in a large number of environments there would be a strong selection pressure on prey species to maximise peripheral vision. In short then these two evolutionary hypotheses differ greatly in plausibility. Indeed, given the limitations on our ability to test evolutionary hypotheses, it is only a high degree of plausibility that saves many evolutionary hypotheses (particularly those concerning behaviour) from being no more than evolutionary ‘just so’ stories.

So in proposing evolutionary explanations concerning the expression of tensed emotions our aim is to provide plausible explanations. Such hypotheses constitute what Dennett approvingly dubs ‘Darwinian reverse engineering’"