We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealised explanation game in which players collaborate to find the best explanation for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping causal (...) patterns of variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. (shrink)

We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealisedexplanation gamein which players collaborate to find the best explanation(s) for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping causal patterns of (...) variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a (conditionally) optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. (shrink)

Volume 51, Issue 13, December 2019, Page 1293-1303.

Group agents can act, and they can have knowledge. How should we understand the species of collective action which aims at knowledge? In this paper, I present an account of group inquiry. This account faces two challenges: to make sense of how large-scale distributed activities might be a kind of group action, and to make sense of the kind of division of labour involved in collective inquiry. In the first part of the paper, I argue that existing accounts of group (...) action face problems dealing with large-scale group actions, and propose a minimal alternative account. In the second part of the paper, I draw on an analogy between inquiry and conversation, arguing that work by Robert Stalnaker and Craige Roberts helps us to think about the division of labour. In the final part of the paper I put the accounts of group action and inquiry together, and consider how to think about group knowledge, deep ignorance, and the different kinds of division of labour. (shrink)

. Given the importance of the issue of scientific objectivity in our democratic societies and the significant development of citizen science, it is crucial to investigate how citizen science may either undermine or foster scientific objectivity. This paper identifies a variety of epistemic risks and benefits that participation of lay citizens in scientific inquiries may bring. It also discusses concrete actions and pending issues that should be addressed in order to foster objectivity in citizen science programs.