Models of Mechanisms: The Case of the Replicator Dynamics - Mechanism and Causality in Biology and Economics - page 87

details. BRD abstracts from these details. For example, it deals with strategies,
abstracting from any concrete content of behavioural plans. Furthermore, it deals
with generic organisms, not specific species or individuals. Finally, it abstracts from
any differences between organisms, as when it assumes that organisms have the
same fitness base rate.
Besides omitting and hence abstracting from many features, BRD also makes
many specific assumptions about the processes it purportedly represents, even
though these assumptions are likely to be false of many of these processes. Typical
idealisations of this sort include the assumption that organisms match and interact
with others randomly, hence idealising possible local interactions and network
structure. It also idealises inheritance, assuming away epigenetic effects and sexual
reproduction. BRD thus is an abstract and idealised representation of a process that
supposedly can be found in many different concrete instantiations.
That it is seen as a representation of one abstract mechanism lies in the success
of its application: many phenomena - in particular those involving frequency-
dependent selection, like sex ratios, fighting behaviour or cooperation - have
been successfully explained by reference to this abstract mechanism.

5 The Social RD

From the 1980s onwards, social scientists have increasingly adopted EGT for their
own explanatory purposes. In particular, EGT has been used in order to explain the
evolution of social institutions, in particular of conventions, norms and fairness
preferences.
Sometimes, social scientists not only employed the general RD model but also
resorted to its biological interpretation. For various reasons, this is today not
considered adequate for most social science purposes. 7 Instead, specifically social
interpretations of the RD have been proposed. These social interpretations represent
mechanisms that account for the social interaction between individuals and for the
social replication of these individuals' traits. A particularly important class of such
mechanisms has been described as learning . Learning is an extremely open con-
cept, and in the following I will only concentrate on those kinds of learning that are

7

These difficulties spring from many sources; I just want to sketch three reasons here. First, while
animals largely exist on the subsistence level, humans mainly do not. It is consequently much less
clear what the causal effect of, say, adherence to norms is for survival and reproduction in humans,
than what the causal effect of daily competition for food, shelter and mating opportunities is for
survival and reproduction in nonhuman animals. Secondly, while it may be plausible that some
basic animal behaviour is encoded in ways that can be inherited through reproduction, it is much
less clear that complex human behavioural characteristics, like compliance with norms, can.
Thirdly, the speed of cultural evolution is often much higher than human reproduction.
Conventions in small groups, for example, can emerge or change within days, thus making
reference to player reproduction inadequate. For these as well as other reasons, strategy replication
often has to be thought of in ways independent of player reproduction.