Choice Group

Abstract: A decision maker (DM) needs to come up with a probability judgement over a set of events based on the judgments of a set of information sources such as experts. How? There are two basic approaches. These are often referred to as (Supra-)Bayesian vs. "mechanical" or "axiomatic". More appropriate terms and conceptualizations for these distinctions may be "belief revision"…

1000 time travellers travel back in time, each with the intention of killing his or her own infant self. Do they succeed? We start with the assumption that there is no branching time. If the possibility of backwards time travel is not to lead to logical contradiction, every time traveller must fail for some reason or another: perhaps one slips on a banana peel, another kills the wrong child, etc. Although a logically consistent story can be told in which each time traveller fails, it is seemingly inexplicable that something will go wrong for each one...

The final Choice Group session for the year will be devoted to a debate on Christian List's recently published book "Why Free Will is Real" in which he presents a conception of free will that meets the various scientific challenges to its reality. The session will begin with a brief presentation by Christian of the main claims of the book, after which Anna Mahtani and Laurenz Hudetz will comment critically on the two of its chapters, before the debate is opened to the floor.

The ultimatum game was developed to help identify the fundamental motivators of human behaviour, typically by asking participants to share a windfall endowment with an anonymous responder. Common observation is that proposers offer, and responders refuse to accept, a much larger share of the endowment than is predicted by rational choice theory. However, in the real world, windfall money is rare, and thus external validity would instead require the use of earned income. This article reports tests of the ultimatum game over earned endowments and finds that the shares that proposers offer and responders accept are significantly lower than that observed with windfall money, moving the overall outcome somewhat closer to the subgame-perfect equilibrium. Moreover, the shares offered by proposers and accepted by responders are significantly lower when the task undertaken to earn the endowment is relatively mundane rather than creative. Finally, the results suggest that offers in the ultimatum game over earned endowments are driven by strategic self-interest rather than a concern with the final distribution of outcomes (assuming that the responder has done nothing), and that an important driver of the participants’ responses appears to be the notion of desert.

Two of our current PhD students, Joe Roussos and Bastian Steuwer, present their research to the Choice Group.

Economics has traditionally been opposed to paternalism. However, the findings of behavioural economics have made popular one kind of paternalism that appears to be more innocuous: The kind of paternalism that respects an agent's ends, or her non-instrumental, intrinsic valuations, and merely helps her purse them effectively. This paper discusses one prominent and initially promising way to inform means paternalist policies addressed at agents who violate expected utility theory (EUT), namely what I call ‘CPT debiasing’. It assumes the descriptive adequacy of cumulative prospect theory (CPT), which allows us to identify a utility function for the target agent. This is often thought of as providing us with a measure of her ends. We then plug this utility function into an expected utility calculation in order to determine a rational way for her to pursue those ends, which the means paternalist then imposes on her. This paper argues that CPT debiasing should be opposed on general anti-paternalist grounds, even if we grant the normative authority of EUT, the descriptive adequacy of CPT, and the idea that means paternalism is at least sometimes immune to general anti-paternalist concerns. First, this is because there are reasons to doubt that the utility function measured within a CPT framework provides us with a measure that isolates the agent's non-instrumental values, or her ends. Second, even if it does, the resulting means paternalism is a problematic type of means paternalism that should be ruled out by the same considerations that motivate economists' opposition to ordinary paternalism.

Conditionalization is one of the central norms of Bayesian epistemology. But there are a number of competing formulations, and a number of arguments that purport to establish it. In this paper, I explore which formulations of the norm are supported by which arguments. In their standard formulations, each of the arguments I consider here depends on the same assumption, which I call Deterministic Updating. I will investigate whether it is possible to amend these arguments so that they no longer depend on it. As I show, whether this is possible depends on the formulation of the norm under consideration.

Abstract: I investigate two questions about irreversible loss. First, is there anything especially bad about an irreversible loss, as opposed to a loss that can be reversed in due course? Second, do we have a duty to avoid imposing (some) irreversible losses on future generations, and if yes, what is this duty based on? In this paper, I aim to…

How do issues related to replication translate into the context of qualitative research? As Freese and Peterson (forthcoming) forewarn, discussions of replication in quantitative social science cannot be directly transposed into this realm. However, we can identify analogs for the various combinations of same data vs. new data, same procedures vs. different procedures scrutiny that have been discussed in the quantitative context. While some of our analogs share the same overarching definitions and import as their quantitative relatives, others diverge significantly. The differences in these instances arise from distinctions between frequentism, which underpins orthodox statistics, and Bayesianism, which a growing body of research identifies as the most promising methodological foundation for inference in qualitative research (Bennett 2015, Humphreys and Jacobs 2015, Fairfield and Charman 2017).

In this chapter, we advance two positions that we believe could help build common ground among quantitative and qualitative scholars. First, we advocate restricting the use of the term replication to a narrowly-defined set of new-data, same-procedures scrutiny that applies to orthodox statistical analysis and experimental research. Second, we argue that the overarching concern in all scientific inquiry is reliability of inference: how much confidence we can justifiably hold in our conclusions. Reliability encompasses but extends beyond the notion of replication. Our discussion therefore focuses on practices that could help improve how we assess evidence, build consensus among scholars, and promote knowledge accumulation within a Bayesian framework, which provides a natural language for evaluating uncertainty.

A choice function C is rational iff: if it allows a path through a sequence of decisions with a particular outcome, then that outcome is amongst the ones that C would have chosen from amongst all the possible outcomes of the sequence. This implies, and it is the strongest definition that implies, that anyone who is irrational could be talked out of their own preferences. It also implies weak but non-vacuous constraints on choices over ends. These do not include alpha or beta.

Two of our PhD students, Chloé de Canson & Nicolas Cote, present their research to the Choice Group.

A stranger in the pub bumps into you spilling your drink everywhere, and then doesn’t apologise, or someone pushes you out of the way to grab a seat on the train. A colleague makes a dismissive remark about your work in front of your boss. A man cat-calls a woman on the street, or wears a T-shirt declaring, ‘keep calm, watch lesbians’. One reaction to slights like these is to take offence. Philosophers have said a great deal about causing offence, especially whether we should punish or prevent it — but less about what is to take offence, let alone whether we should take offence, or within what limits we should do so. Hitherto the focus of moral and legal philosophy has tended to be the offender, not the offended.

In this talk, I offer an analysis of what it is to take offence and what doing so is like. On this analysis, a more nuanced, even positive appraisal of this emotion becomes possible, as compared to its popular reputation. First, then, I survey the shortfalls of the limited amount that philosophers have said about taking offence and offer my alternative account. Second, I distinguish offence from nearby other-condemning emotions, of anger, disgust, and contempt. Third, I turn to the implications for not only for how we conceptualise offence but what we think about those who take it. On my account, offence tends to be a smaller-scale and more everyday emotion than those who make claims about its threats to society suppose. While offence may appear excessive that is most likely, only in limited cases: those requiring symbolic withdrawal or proxy forms of estrangement.

Abstract: Continuity conditions appear to be innocuous and have not received much scrutiny. Completeness, by contrast, is a substantive and contentious condition that has been questioned on numerous occasions. Surprisingly, however, continuity implies completeness. There are two types of continuity conditions that jointly rule out incompleteness. This paper puts forward a proof that clearly exhibits how the different continuity conditions…

In Rational Powers in Action (Oxford, forthcoming), I defend what might be called an "action-first" conception of instrumental rationality; that is, a conception of instrumental rationality that takes intentional action as the fundamental category of the theory of instrumental rationality. In this talk, I will first explain more precisely the commitments of an action-first model as
well as the central tenets of the specific version of the model I defend. I then outline the main attractions and advantages of such a view, especially with regards of how it can deal with the extended nature of agency. However, the model seems to face a major obstacle. The central principles of the model seem to have no application for risky contexts: its two central principles (a means-end principle of derivation and a coherence principle that requires the agent not to pursue incompatible ends) presuppose a context of knowledge for their proper application. I try to show, however, that there are promising ways to extend this model to contexts of risk and uncertainty, and that the model can both incorporate the insights of decision theory in some of these contexts, as well as providing plausible accounts of ordinary choice dispositions that are puzzling from the point of view of orthodox decision theory.

Two of our PhD students, Ze’ev Goldschmidt & Nick Makins, present their research to the Choice Group.

Do people tend to be overconfident in their opinions? Many psychologists think so. They have run calibration studies in which they ask people a variety of questions, and then test whether their confidence in their answers matches the proportions of those answers that are true. Under certain conditions, an “overconfidence effect” is robust—for example, of the answers people are 80% confident in, only 60% are true. Psychologists have inferred that people tend to be irrationally overconfident. My question is when and why this inference is warranted. Although it may at first appear hopelessly flawed, I show that under controlled conditions it is a warranted inference. However, I argue that real-world studies standardly fail to meet these conditions—and, as a result, that rational people can often be expected to display the “overconfidence effect.” Thus in order to test whether people are overconfident, we must first predict whether and to what extent a rational person will display this effect, and then compare that prediction to real people’s performance. I show how in principle this can be done—and that doing so may overturn the standard interpretation of robust empirical effects. Upshot: have much less evidence for overconfidence than many have thought.

In this talk, I discuss the role of consent in permissible sexual interaction. I’ll suggest that the focus on consent as an individual act may be a symptom of a narrow focus on individual rights and permissions as framing the moral domain; but that joint, co-operative activity is not well fitted into this normative vocabulary. This raises a further puzzle. It is a manifest sociological fact that consent, and competence at tracking its presence or absence, has come to the forefront of the attempted regulation of the activities of teenagers and young adults. What are we to make of this focus, if it is not grounded in our understanding of the wrongs of bad sex? I offer a speculative hypothesis about this sociology, and draw out a couple of normative consequences.