In this paper I tackle the question of how mechanisms can be represented within a causal graph framework. I begin with a few words on mechanisms and some of their characteristic properties. I then concentrate on how one of these characteristic properties, viz. the hierarchic order of mechanisms (mechanisms frequently consist of several submechanisms), can be represented within a causal graph framework. I illustrate an answer to this question proposed by Casini, Illari, Russo, & Williamson (2011) and demonstrate on an example that their formalism, though nicely capturing the hierarchic order of mechanisms, does not support two important features of nested mechanisms: (i) a mechanism’s submechanisms are typically causally interacting with other parts of said mechanism, and (ii) intervening in some of a mechanism’s parts should have some influence on the phenomena the mechanism as a whole brings about. Finally, I sketch an alternative approach capable of taking properties (i) and (ii) into account and demonstrate this on the above-mentioned exemplary mechanism. Citation information: Gebharter, A. (2014). A formal framework for representing mechanisms? Philosophy of Science, 81(1), 138–153. doi:10.1086/674206

Mechanisms are usually viewed as hierarchical, with lower levels of a mechanism influencing, and decomposing, its higher-level behaviour. To draw quantitative predictions from a model of a mechanism, the model must capture this hierarchical aspect. Recursive Bayesian networks (RBNs) were put forward as a means to model mechanistic hierarchies (Casini et al., 2011) by decomposing variables into their constituting causal networks. The proposal was criticized by Gebharter (2014). He proposes an alternative formalism, which decomposes arrows. Here, I defend RBNs from the criticism and argue that they off er a better representation of mechanistic hierarchies than the rival account.

We introduce a methodology for the design of parametric mechanisms, which are multiagent systems inhabited by strategic agents, with knobs that can be adjusted to achieve specific goals. We assume agents play approximate equilibria, which we estimate using the probably approximately correct learning framework. Under this assumption, we further learn approximately optimal mechanism parameters. We do this both theoretically, assuming a finite design space, and heuristically, using Bayesian optimization (BO). Our BO algorithm incorporates the noise associated with modern concentration inequalities, such as Hoeffding’s, into the underlying Gaussian process. We show experimentally that our search techniques outperform standard baselines in a stylized but rich model of advertisement exchanges.

Mechanisms play an important role in many sciences when it comes to questions concerning explanation, prediction, and control. Answering such questions in a quantitative way requires a formal represention of mechanisms. Gebharter (2014) suggests to represent mechanisms by means of one or more causal arrows of an acyclic causal net. In this paper we show how this approach can be extended in such a way that it can also be fruitfully applied to mechanisms featuring causal feedback. Citation information: Gebharter, A., & Schurz, G. (2016). A modelling approach for mechanisms featuring causal cycles. Philosophy of Science, 83(5), 934–945. doi:10.1086/687876

Mechanisms are said to consist of two kinds of components, entities and activities. In the first half of this chapter, I examine what entities and activities are, how they relate to well-known ontological categories, such as processes or dispositions, and how entities and activities relate to each other (e.g., can one be reduced to the other or are they mutually dependent?). The second part of this chapter analyzes different criteria for individuating the components of mechanisms and discusses how real the boundaries of mechanisms are.