Brian Epstein

It seems to be a fact about language that people carry on competent and intelligent conversations, day to day, without being able to explain the words they're using. People don't have definitions, even tacit ones, for their words. This is possible in part because language users live in a social network, relying on one another as well as the structure of the world, which takes the burden of definition off the shoulders of the ordinary language user. ;Still, even simple words in fact stand for extremely complex and not especially natural kinds of things. In many cases, it's unlikely that we can look to the traditional sources---either to the definitions of even the most informed experts, or to the regular structures in nature and human practices---for endowing words with meaning. In this dissertation, I work at the margins of the theory of reference, to develop an account of the factors that make it possible for us to introduce words into language. Social kinds are paradigmatic instances of the phenomenon of establishing reference. ;I argue that establishing reference imposes less of a cognitive burden on language users than classic and contemporary accounts suppose. In particular, we often introduce words into language without even having the ability to identify what factors are involved in settling their reference. The reason is that the world---even the world of candidate social kinds---is a sparse place. A rich set of distributed conceptual factors figure into setting up the candidates for reference and hence for settling the reference of newly introducing words, even when the person fixing reference doesn't invoke these factors. I consider the factors that make the world sparse, in particular focusing on both conceptual and historical factors. Moreover, I suggest that studying reference to social kinds doesn't only shed light on how reference fixing works and on the nature of social kinds, but also can be applied reciprocally to reference, understanding language and reference themselves to be social kinds

Many theorists, including Nietzsche, Adorno, and Foucault, have regarded genealogy as an important technique for social criticism. But it has been unclear how genealogy can go beyond the accomplishments of other, more mundane, critical methods. I propose a new approach to understanding the critical potential of history. I argue that theorists have been misled by the assumption that if a claim is deserving of criticism, it is because the claim is false. Turning to the criticism of concepts rather than criticism of claims, I expand on the distinction between "descriptive semantics" and "foundational semantics" to show that genealogy can be uniquely qualified to explore the foundations of concepts, and hence to criticize concepts that are problematic in nonobvious ways

When can macroscopic data about a system be used to set parameters in a microfoundational simulation? We examine the epistemic viability of tweaking parameter values to generate a better fit between the outcome of a simulation and the available observational data. We restrict our focus to microfoundational simulations—those simulations that attempt to replicate the macrobehavior of a target system by modeling interactions between microentities. We argue that tweaking can be effective but that there are two central risks. First, tweaking risks overfitting the simulation to the data and thus compromising predictive accuracy; and second, it risks compromising the microfoundationality of the simulation. We evaluate standard responses to tweaking and propose strategies to guard against these risks.