Publications

---

Essential Structure for Apt Causal Models   (forthcoming, Australasian Journal of Philosophy)

A paper that introduces a new aptness requirement on causal models, Evident Mediation, that resolves the problem of structural isomorphs without resorting to talk of normality.

Abstract: A definition of actual causation in terms of structural equation models has two components: a recipe for reading claims of actual causation off an apt model, and an articulation of what makes a model apt. The primary focus in the literature has been on the first component. But the recently discovered problem of structural isomorphs has also made the second salient (N. Hall 2007; Hitchcock 2007a; Blanchard and Schaffer 2017; Menzies 2017). This paper presents a new aptness requirement, Evident Mediation, that resolves this problem.

 

---

---

Works in Progress

---

< title redacted for blind review >   (under review)

A paper arguing that a causal model analysis of actual causation is committed to a kind of contrastivism – specifically, a view whereby causation is relative to a specification of background possibilities.

Abstract:  A promising development in the philosophy of causation analyzes actual causation using structural equation models, also called “causal models”. This paper argues that any such analysis is committed to a kind of contrastivism – the view that causation is a three-part relation holding between a cause, an effect, and a set of alternatives (or, “contrasts”). In particular, insofar as a causal model analysis construes causation mind-and-language independently, it must treat causation as relative to a specification of background possibilities. Or, so I argue. Along the way, I flesh out what it means for an interpreted model to be accurate of its target situation.

< title redacted for blind review >   (under review)

A paper demonstrating the analogy between the role similarity plays in a similarity semantics and that played by aptness in an interventionist semantics. Counter to what is claimed in the literature, this means that an interventionist semantics is vulnerable to the same fundamental problem as is more traditional similarity semantics – the importation problem.

Abstract: Structural equation models lend themselves to a semantics of counterfactuals. Call this an interventionist semantics of counterfactuals. It seems like this semantics improves on a traditional similarity semantics by avoiding vague talk of similarity between possible worlds (Pearl, 2000, 2013; Starr, 2019). However, this paper argues that an interventionist semantics remains vulnerable to the same kind of problem. Recently dubbed the importation problem (Priest, 2018), this is the foundational problem for any counterfactual semantics – that of identifying what information is relevant to a counterfactual evaluation. I show that where similarity semantics relies on a vague notion of similarity, an interventionist semantics relies on an equally vague notion of aptness.

< title redacted for blind review >   (under review)

A paper arguing that the nature of the contribution of causal models to a counterfactual reduction of actual causation is purely heuristic.

Abstract:This paper looks at what causal model analyses of causation are up to – specifically, what justifies treating causal model analyses as cutting-edge counterfactual analyses. It focuses, then, on analyses of actual causation that take causal models to represent counterfactual dependencies and, to focus further, on counterfactuals underwritten by a similarity semantics. The task becomes that of identifying what causal models bring to the table such that an analysis invoking them improves on a counterfactual analysis that doesn’t. I argue that no substantive contribution made by the models framework requires models. But this is not to say no contribution is made. The real contribution of a models approach, I argue, is heuristic. It lies in their making plain what has thus far skulked in the shadows of a counterfactual analysis – that conditions constraining the evaluation of causation-relevant counterfactuals cannot be simultaneously determinate, categorical, and mind-and-language independent. I conclude by suggesting a view of causation which, in my view, best responds to this challenge.

Stable Models without Stability

A paper arguing that the Stability principle on apt models is made redundant by another independently motivated principle – Evident Mediation.

Abstract:  The application of causal models to analyzing causation relies on an account of good, or apt, models. This paper focuses on one aptness principle that has been proposed, Stability, which requires of a model’s causal verdicts that they survive adjustments to the model. After clarifying the principle, I consider its utility. Ultimately, I dismiss it as a heuristic that’s no longer needed. The underlying error of unstable models can be diagnosed as a violation of another aptness principle in the literature – Evident Mediation.

Strong Proportionality and Causal Claims 

A paper defending the principle of strong proportionality against several objections in the literature.

Abstract: There are several supposedly lethal objections to the view that causation is essentially proportional. The first targets an account of proportionality in terms of causal models, pointing out that proportionality is too easily satisfied in causal model accounts of causation through manipulation of the range of values that a variable can take (Franklin-Hall, 2016). The second argues that proportionality legitimizes only the most general things as causes, and proportionality thereby contravenes causal intuitions (Bontly, 2005; Franklin-Hall, 2016; McDonnell, 2018, 2017; Weslake, 2013). The final, and perhaps most intractable, objection holds that proportionality counter-intuitively legitimizes disjunctive causes (Shapiro and Sober, 2012; Weslake, 2017; Woodward, 2018). This paper provides a unified response to these objections, which is best formulated in a causal model framework. I first articulate two independently plausible principles of variable selection – exclusivity and exhaustivity. I then show how the adoption of these principles responds to Franklin-Hall’s objection, and dissolves the remaining two.

How Proportionality Resolves Preemption 

A paper arguing that the view of causation that I develop can resolve the problem of redundant causation.

Abstract: Redundant causation is when the presence of an independently sufficient cause means that a causal relation holds despite there being no counterfactual dependence. For example, late preemption occurs whenever the completion of one causal process to achieve a certain result prevents, or preempts, a second process that would have been sufficient for that same result were it allowed to run its course. It’s normally enough to simply imagine a causal process as never having happened in order to identify whether it made a difference to the effect. In late preemption, though, removing the first process allows the second to complete, which produces the same effect. In this paper, I first show how standard treatments of preemption in causal model accounts of causation fail due to an inconsistency in adherence to standard principles of model representation. I then argue that a non-objectivist theory of actual causation along the lines that I’ve defended, in tandem with the principle of strong proportionality, can resolve the problem of redundant causation. In particular, counterfactual dependence between an effect and its intuitive cause is revealed once we recognize that what we pick out as the intuitive cause is always either the proportional cause or else the actual realizer of the proportional cause. I conclude by offering a plausible explanation of this practice. 

 

---

---

Dissertation

Short Abstract

The philosophy of causation has been recently galvanized by what might be done with structural equation models – highly effective formal models that link correlational information up with causal information. Since these models have been so successful in causal discovery, perhaps they can help us define what causation is. But attempts to define actual causation in terms of these models largely bracket the question of what makes a model appropriate or apt for this purpose. My dissertation provides this needed account of aptness, uncovering something surprising in the process: these models represent situations only relative to some set of background possibilities or other. As they stand, definitions of actual causation in the literature tend to existentially quantify over all such sets. But this delivers previously unrecognized counterintuitive causal verdicts. As an alternative, I introduce and defend the theory that takes this relativity at face value – a context-sensitive view of actual causation whereby a causal relation holds only relative to a set of background possibilities.

Long Abstract

My dissertation invokes the apparatus of causal models to motivate and develop a novel variant of a counterfactual account of causation. Causal models – also called structural equation models (SEMs) or Bayesian nets – have been developed and refined in recent decades by special scientists for the purposes of identifying the causal structure of a target system. Philosophers have also begun to use these models to shed light on the nature of causation itself, developing accounts of causation in terms of them.

To illustrate, say we want to know whether smoking causes cancer. We first construct an apt causal. Consider a model with two variables: X, representing whether an individual smokes or not, and Y, representing whether that individual gets cancer or not. Then, smoking is a cause of cancer just in case intervening on the model to change the value of X leads to a change in Y.

Such an account provides conditions for a causal relation holding between two things in terms of features of a model that represents those things. But not just any model will be relevant. Some models will be defective for one reason or another. Any account therefore needs to restrict the focus to just those models that represent the world in the right way – the apt models. But precisely how to articulate the notion of aptness has been largely set aside – designated an art rather than a science. Work progresses by a reliance on “natural” models of target situations. Arguably, though, we find these models natural because they capture what we already understand about the causal structure of the situation in question. Thus, if the models are to independently illuminate the nature of causation, an articulation of aptness is crucial.

My dissertation begins by addressing the question of when a causal model is apt. I first define what it is for a model on a given interpretation to be accurate of a target situation. This involves a systematization and explication of various representational principles gleaned from throughout the literature. I then explain and address two reasons for which accuracy is insufficient for aptness. The first reason – already discussed in the literature – is the problem of structural isomorphs. In response, I propose the aptness condition of Evident Mediation. The second reason – which has yet to be noticed – is the problem of the indeterminacy of accuracy. As I demonstrate, a given model on a given interpretation is accurate of a target situation only relative to a set of background possibilities – what I call a modal profile.

However, this observation raises a problem for theories of actual causation in terms of these models – namely, that models representing certain modal profiles deliver counterintuitive verdicts. I argue that the best response to this problem is to take the relativity at face value, building it into the metaphysics of causation. I propose that relativity of actual causation to modal profile is to be understood on analogy with the relativity of simultaneity to reference frame. The resulting view is a kind of causal pluralism – which I call Causal Relativism – whereby actual causation is a three-part relation that holds between a cause, and effect, and a modal profile. An implicit reference to modal profile is taken to be a hidden parameter in any causal claim. Finally, I explore one benefit of this proposal: that the resulting account can defend the principle of proportionality against several standing objections. Proportionality is the view that the causation is proportional, where a cause is proportional to its effect whenever it is fine-grained enough to ensure the effect but no more than necessary. I argue that the three principal objections to this view dissolve upon recognition that causal relations are relative to modal profiles.

The view of actual causation that I ultimately defend, then, holds that causation is proportional counterfactual dependence relative to a collection of background possibilities, aka. a modal profile.

---