For the academic year 2014—2015, I will be a Bersoff Assistant Professor/Faculty Fellow in the philosophy department at New York University.
My research interests lie at the intersections of the philosophy of science, metaphysics, and epistemology. My work in metaphysics and the philosophy of science focuses on causation, counterfactuals, and chance. In epistemology, I am interested in the rational norms governing partial belief states and their relationship to the kinds of rational norms that govern full belief states.
How to Learn from Theory-Dependent Evidence; or Commutativity and Holism: A Solution for Conditionalizers
2014. The British Journal for the Philosophy of Science. 65 (3): 493-519
Weisberg (2009) provides an argument that neither conditionalization nor Jeffrey conditionalization is capable of accomodating the confirmational holist's claim that beliefs acquired directly from experience can suffer undercutting defeat. I diagnose this failure as stemming from the fact that neither conditionalization nor Jeffrey conditionalization give any advise about how to respond to theory-dependent evidence, and I propose a novel updating procedure which does tell us how to respond to evidence like this. This holistic updating rule yeilds conditionalization as a special case when our evidence is theory-independent.
The Emergence of Causation
Several philosophers have embraced the view that high-level—events events like Zimbabwe's monetary policy and its hyper-inflation—are causally related if their corresponding low-level, fundamental physical events are causally related. I dub the view which denies this without denying that high-level events are ever causally related causal emergentism. Several extant philosophical theories of causality entail causal emergentism, while others are inconsistent with the thesis. I illustrate this with David Lewis's two theories of causation (1973, 2000), one of which entails causal emergentism, the other of which entails its negation. I then argue for causal emergentism on the grounds that it provides the only adequate means of squaring the apparent plenitude of causal relations between low-level events with the apparent scarcity of causal relations between high-level events. This tension between the apparent abundance of low-level causation and the apparent scarcity of high-level causation has been noted before. However, it has been thought that various theses about the semantics or the pragmatics of causal claims could be used to ameliorate the tension without going in for causal emergentism. I argue that none of the suggested semantic or pragmatic strategies meet with success, and recommend emergentist theories of causality in their stead. As Lewis's 1973 account illustrates, causal emergentism is consistent with the thesis that all facts reduce to microphysical facts.
A Theory of Structural Determination
While structural equations modeling is increasingly used in philosophical theorizing about causation, it remains unclear what it takes for a particular structural equations model to be correct. To the extent that this issue has been addressed, the consensus appears to be that it takes a certain family of causal counterfactuals being true. I argue that this account faces difficulties in securing the independent manipulability of the structural determination relations represented in a correct structural equations model. I then offer an alternate understanding of structural determination, and I demonstrate that this theory guarantees that structural determination relations are independently manipulable. The account provides a straightforward way of understanding hypothetical interventions, as well as a criterion for distinguishing hypothetical changes in the values of variables which constitute interventions from those which do not. It additionally affords a reductive semantics for causal counterfactual conditionals which is able to yield a clean solution to a problem case for the standard 'closest possible world' semantics.
Diachronic Dutch Books and Evidential Import
A handful of well-known arguments (the 'diachronic Dutch book arguments') rely upon theorems purporting to establish that you are immune from diachronic Dutch book-ability if and only if you adopt the strategy of conditionalizing (or Jeffrey conditionalizing) on whatever evidence you happen to receive. These theorems require nontrivial assumptions about which evidence you might acquire—in the case of conditionalization, the assumption is that, if you might learn that e, then it's not the case that you might learn something else that is consistent with e. These assumptions may not be relaxed. When they are, not only will non-(Jeffrey) conditionalizers fail to be diachronically Dutch book-able, but (Jeffrey) conditionalizers will themselves be diachronically Dutch book-able. I argue that either there are epistemic situations in which these assumptions break down, and (Jeffrey) conditionalization is diachronically Dutch book-able, or else many (and plausibly most) learning experiences fall outside of the purview of the diachronic Dutch book arguments. Either way, these arguments are invalid.
Probabilistic Causation and the Markov Condition
On one understanding of probabilistic causation, the probability attaches to the effect—c probabilistically caused e iff c deterministically caused the probability of e. On the other understanding, the probability attaches to the causal relation—c probabilistically caused e iff c caused e, though there was some probability that it wouldn't. Hausman and Woodward (1999, 2004) appeal to the former conception of probabilistic causation to show that the Causal Markov Condition will be satisfied when causation is probabilistic. I demonstrate that, pace Hausman and Woodward, this conception of probabilistic causation is incapable of establishing the Causal Markov Condition except by fiat. I then demonstrate that, on the latter conception of probabilistic causation, there are violations of the Markov Condition; however, there is a general locality constraint which is sufficient to restore the Markov Condition.
An Inaccuracy Minimization Argument for Holistic Conditionalization
It is shown that the update rule defended in Gallow (2014), holistic conditionalization, follows from a natural strategy for minimizing the expected inaccuracy of your posterior credences when you acquire theory-dependent evidence.
Deterministic Chance and the Principal Principle
I present an account of deterministic chance which takes, as its jumping-off point, the physico-mathematical approach to theorizing about deterministic chance known as the method of arbitrary functions. This approach promisingly yields deterministic probabilities which align with what we take the chances to be—it tells us that there is a 1/2 probability of a spun roulette wheel stopping on black, a 1/2 probability of a flipped coin landing heads up—but it requires some probabilistic materials to work with. Some have claimed that these probabilistic materials are supplied by certain a priori symmetries; others find the required probablistic materials in actual or hypothetical frequencies; still others postulate them as brute laws of nature. I contend that the right probabilistic materials are found in reasonable initial credence distributions. I note that, with some rather weak normative assumptions, the resulting account entails that deterministic chances obey a variant of Lewis's 'principal principle'. I additionally argue that deterministic chances, so understood, are capable of explaining certain facts about long-run frequencies, such as the fact that, in a long series of coin flips, about half of the time, the coin lands heads.
The Emergence of Causation
The understanding of the world's causal structure provided by fundamental physics stands in stark contrast to the understanding provided by the special sciences. At the level of fundamental physics, causal influence appears to be relatively undiscriminating—it appears that most past fundamental physical events causally influence most future fundamental physical events. In contrast, the special sciences appear tell us that many past high-level events are causally irrelevant to many future high-level events—e.g., China's one child policy did not cause Zimbabwe's hyperinflation.
In part, the dissertation focuses on what this disparity can teach us about the nature of causality. There is a popular thesis according to which high-level events are causally related whenever their low-level, fundamental physical realizers are causally related. I contrast this with a thesis I call causal emergentism, according to which, though there is high-level causation, what it is for high-level events to be causally related isn't just for their low-level realizers to be causally related. I argue for causal emergentism on the grounds that it provides the only adequate means of squaring the apparent plenitude of low-level causation with the apparent scarcity of high-level causation.
I additionally provide a theory of causation which is capable of explaining how and why the world's causal structure varies across different levels of description. According to this theory, causation requires the presence of what I call structural determination relations between different properties of the world. Following authors like Halpern, Hitchcock, Pearl, Woodward, Spirtes, Glymour, and Scheines, I represent these properties of the world with variables, and represent structural determination relations as equations relating the values of those variables. Most of these authors have interpreted structural determination relations as encoding counterfactual information. I show that, if structural determination relations are understood in this way, then they will not be guaranteed to be independently manipulable—a property these relations must possess if they are to be used to provide an account of causation along the lines many of these authors propose. In contrast, I propose an account according to which one variable is structurally determined by upon others if and only if, roughly, the values of the latter variables are sufficient for the value of the former throughout a certain nearby region of modal space. Given either definition, however, there need not be a relations of structural determination between two high-level properties of the world simply because there is a relations of structural determination between the low-level properties which realize them.
After articulating this account of structural determination, I formulate an account of causal counterfactuals and singular causation in terms of networks of structural determination. On this account, structural determination relations provide the pathways along which causal influence propagates; and causation is, roughly, counterfactual dependence within a network of structural determination relations. A few extra bells-and-whistles are necessary. However, once causal counterfactual conditionals have been properly grounded in networks of structural determination, the standard objections to counterfactual accounts of causation dissolve.