The operations of deep networks are widely acknowledged to be inscrutable. The growing field of Explainable AI (XAI) has emerged in direct response to this problem. However, owing to the nature of the opacity in question, XAI has been forced to prioritise interpretability at the expense of completeness, and even realism, so that its explanations are frequently interpretable without being underpinned by more comprehensive explanations faithful to the way a network computes its predictions. While this has been taken to be a shortcoming of the field of XAI, I argue that it is broadly the right approach to the problem.

This paper argues that a notion of statistical explanation, based on Salmon’s statistical relevance model, can help us better understand deep neural networks. It is proved that homogeneous partitions, the core notion of Salmon’s model, are equivalent to minimal sufficient statistics, an important notion from statistical inference. This establishes a link to deep neural networks via the so-called Information Bottleneck method, an information-theoretic framework, according to which deep neural networks implicitly solve an optimization problem that generalizes minimal sufficient statistics. The resulting notion of statistical explanation is general, mathematical, and subcausal.

We model scientific theories as Bayesian networks. Nodes carry credences and function as abstract representations of propositions within the structure. Directed links carry conditional probabilities and represent connections between those propositions. Updating is Bayesian across the network as a whole. The impact of evidence at one point within a scientific theory can have a very different impact on the network than does evidence of the same strength at a different point. A Bayesian model allows us to envisage and analyze the differential impact of evidence and credence change at different points within a single network and across different theoretical structures.

Substrate independence and mind-body functionalism claim that thinking does not depend on any particular kind of physical implementation. But real-world information processing depends on energy, and energy depends on material substrates. Biological evidence for these claims comes from ecology and neuroscience, while computational evidence comes from neuromorphic computing and deep learning. Attention to energy requirements undermines the use of substrate independence to support claims about the feasibility of artificial intelligence, the moral standing of robots, the possibility that we may be living in a computer simulation, the plausibility of transferring minds into computers, and the autonomy of psychology from neuroscience.

This paper accounts for broad definitions of memory, which extend to paradigmatic memory phenomena, like episodic memory in humans, and phenomena in worms and sea snails. These definitions may seem too broad, suggesting that they extend to phenomena that don’t count as memory or illustrate that memory is not a natural kind. However, these responses fail to consider a definition as a hypothesis. As opposed to construing definitions as expressing memory’s properties, a definition as a hypothesis is the basis to test inferences about phenomena. A definition as a hypothesis is valuable when the “kinding” of phenomena is ongoing.

It is commonly maintained that neuroplastic mechanisms in the brain provide empirical support for the hypothesis of multiple realizability. We show in various case studies that neuroplasticity stems from preexisting mechanisms and processes inherent in the neural (or biochemical) structure of the brain. We argue that not only does neuroplasticity fail to provide empirical evidence of multiple realization, its inability to do so strengthens the mind-body identity theory. Finally, we argue that a recently proposed identity theory called Flat Physicalism can be enlisted to explain the current state of the mind-body problem more adequately.

Accuracy arguments for the core tenets of Bayesian epistemology differ mainly in the conditions they place on the legitimate ways of measuring the inaccuracy of our credences. The best existing arguments rely on three conditions: Continuity, additivity, and strict propriety. In this paper, I show how to strengthen the arguments based on these conditions by showing that the central mathematical theorem on which each depends goes through without assuming additivity.

While practitioners think highly of randomized studies, some philosophers argue that there is no epistemic reason to randomize. Here I show that their arguments do not entail their conclusion. Moreover, I provide novel reasons for randomizing in the context of interventional studies. The overall discussion provides a unified framework for assessing baseline balance, one that holds for interventional and observational studies alike. The upshot: Practitioners’ strong preference for randomized studies can be defended in some cases, while still offering a nuanced approach to evidence appraisal, one where not all nonrandomized studies are treated equally.

This paper discusses Millstein’s (2020) criticism of the consensus view formed against selected-effects ecological functions. I argue that Millstein’s defense of coevolution-based selected-effects ecological functions applies to a notion of function as an activity, whereas proponents of the consensus view are concerned with a notion of ecological function as the contribution of an organism, population, species, or abiotic item to the maintenance of its community and/or the functioning of its ecosystem. Millstein’s arguments hence do not invalidate the consensus view but draw attention to a notion of function that has been neglected in philosophical discussions and that deserves more attention.

Jonathan Livengood and Justin Sytsma have published a series of studies on “Actual Causation and Compositionality,” in which they investigate causal attributions of laypeople. We use one of their vignettes to follow up on their research. Our findings cast doubt on their conclusion that ordinary causal attributions tend to violate the compositionality constraint if one looks at cases in which someone is responsible for an effect by way of an intermediary that does not share in the responsibility.