Choice Group

The von Neumann and Morgenstern framework is frequently adopted to determine a numerical measure of individual wellbeing or goodness, i.e. of how good the realisation of different possible states of affairs would be for an individual. It is also sometimes used to determine a measure of social wellbeing or of the moral (overall) goodness of different states of affairs. The panel will debate the appropriateness of these uses of the von Neumann and Morgenstern framework and in particular whether the measure it determines best represents individual and moral goodness.

Abstract: Many blame today's economic inequality, stagnation, and political instability on the free market. The solution is to rein in the market, right? We challenge this thinking— as well as most arguments in favor of and against markets. We offer new, but historically rooted, ways to imagine markets for the common good. We argue that the emancipatory force of genuinely open,…

Abstract: TBA

The principle that rational agents should maximize expectations is intuitively plausible with respect to many ordinary cases of decision-making under uncertainty. But it becomes increasingly implausible as we consider cases of more extreme, low-probability risk (like Pascal’s Mugging), and intolerably paradoxical in cases like the St. Petersburg Lottery and the Pasadena Game. In this paper I show that, under certain assumptions, stochastic dominance reasoning can capture many of the plausible implications of expectational reasoning while avoiding its implausible implications. More specifically, when an agent starts from a condition of background uncertainty about the choiceworthiness of her options representable by a probability distribution over possible degrees of choiceworthiness with exponential or heavier tails and a sufficiently large scale parameter, many expectation-maximizing gambles that would not stochastically dominate their alternatives “in a vacuum” turn out to do so in virtue of this background uncertainty. Nonetheless, even under these conditions, stochastic dominance will generally not require agents to accept extreme gambles like Pascal’s Mugging or the St. Petersburg Lottery. I argue that the sort of background uncertainty on which these results depend is appropriate for any agent who assigns normative weight to aggregative consequentialist considerations, i.e., who measures the choiceworthiness of an option in part by the total amount of value in the resulting world. At least for such agents, then, stochastic dominance offers a plausible general principle of choice under uncertainty that can explain more of the apparent rational constraints on such choices than has previously been recognized.

The "Belief Models" framework is a very general formal theory of rational belief. It encompasses propositional logic belief sets, but also ranking functions and lower previsions (i.e. lower probabilities). de Cooman (2005) shows that AGM-style expansion and revision operators can be defined in this framework. In this paper I show that "merging operators" -- a kind of general theory of aggregation -- in the style of Konieczny and Pino-Perez (2002) can also be generalised from propositional logic to Belief Models. I will focus on the consequences for aggregating probabilities and lower probabilities.

For roughly two decades evidentialism, of some nearby variant thereof, has been the orthodoxy about normative reasons for belief, or as I would prefer to put it, the determinants of what one ought to believe. While some form of evidentialism or alethicism remains the dominate view, a small but growing number of philosophers are defending the view that there are pragmatic reasons for belief. Amongst them the most popular view is strict pragmatism: the view that all reasons for belief are pragmatic reasons for belief. While some authors suggest a pluralist alternative, many seem to imply that evidential/alethic reasons for belief and pragmatic reasons for belief nonetheless cannot be compared. In this talk I offer some considerations about why we should expect to be able to compare alethic and pragmatic reasons for belief and offer a new suggestion about how to do so. Or to put the point in a preferable way, I offer a suggestion about how alethic and pragmatic considerations contribute to determining what we ought to believe.

Ambivalence can typically be defined as the holding by a single person of two opposing attitudes towards the same thing. Ambivalence is denied and marginalized by a variety of approaches, and yet is pervasive and central both in the lives of individuals and in public life. My book argues that ambivalence is a rational and unitary (yes!) engagement, and defends revised accounts of subjectivity, rationality, belief, and desire-action relations. Analysing ambivalence of value judgement, I argue that we often aptly judge that two irreducibly conflicting things hold together, for example we ambivalently judge that some action both ought and ought not to be done. This goes hand in hand with a study of the relations of ambivalence with deliberation on the one hand, and with action on the other. Here I argue that it is inherent to action that people can quite successfully act on both of their opposing attitudes together. In the talk we’ll retrace some of the problems and consider some aspects of my positive investigation.

Nicolas Cote: "Liberalism and Social Choice" Abstract: One of the core commitments of liberalism is to the protection of individuals against intolerable intrusions upon their liberty. It enshrines these protections through rights, which safeguard us from coercion. Sen’s celebrated “Impossibility of a Paretian Liberal” result proved that welfarism, as expressed by the Pareto principle, is inconsistent with even the most minimal concern for liberal rights. However,…

Abstract: In this talk I will defend two mutually supporting theses about the relation between chances and credences. The first is that chances should be identified with the judgements or credences of a (hypothetical) unbounded and fully informed perfect inductive reasoner and not with any of the possible grounds for her judgements, such as frequencies or propensities. The second is…

This session will be a discussion with James Joyce of his recent paper on Newcomb problems. Those wishing to attend should read the paper in advance since there will be no presentation, just a question and answer session.

Scientists and philosophers frequently speak about levels of description, levels of explanation, and ontological levels. In this paper, I propose a unified framework for modelling levels. I give a general definition of a system of levels and show that it can accommodate descriptive, explanatory, and ontological notions of levels. I further illustrate the usefulness of this framework by applying it to some salient philosophical questions: (1) Is there a linear hierarchy of levels, with a fundamental level at the bottom? And what does the answer to this question imply for physicalism, the thesis that everything supervenes on the physical? (2) Are there emergent properties? (3) Are higher‐level descriptions reducible to lower‐level ones? (4) Can the relationship between normative and non‐normative domains be viewed as one involving levels? Although I use the terminology of “levels”, the proposed framework can also represent “scales”, “domains”, or “subject matters”, where these are not linearly but only partially ordered by relations of supervenience or inclusion. A preprint is available at: http://personal.lse.ac.uk/list/pdf-files/levelsfinal.pdf

Algorithmic decision-making (ADM) is increasingly being used to support probabilistic, predictive decision-making procedures of political and social significance. This includes, for instance, decisions in the criminal justice process, law enforcement, benefits allocation, hiring, and credit lending. A growing body of research in computer science and statistics reveals that algorithmic ex ante judgments are often subject to algorithmic bias: a disproportionately high risk of error concerning predictive assessments about racial and religious minorities, women, and other socially disadvantaged groups. Furthermore, due to certain features of algorithmic decision-making in its contemporary form—such as low explainability, low contestability, and automation bias—there is reason to worry that ADM may not only reproduce large-scale social injustices, but that it may also exacerbate them...

One kind of value that a person’s life has is subjective – the value of life to the person whose life it is. This value plays an important role in certain moral philosophies, in health economics and health policy, and in a wide range of normative judgments about death and saving lives. It is surprisingly difficult, however, to say just what it is we are referring to with this concept...

Suppose a building is collapsing, and six children are trapped inside. If you do nothing, they will die. If you support the wall to your left, one child will escape unharmed. If you instead support the wall to your right, the other five will escape unharmed. Supporting either wall would result in both your arms being crushed. It seems plausible that (1) it is permissible not to save any of the children, and (2) it is wrong to save the one rather than the five. Joe Horton has argued that (1) and (2) together support the seemingly implausible claim that, if you are not going to save all five children, you ought to save none. I argue that there is a plausible way of avoiding this claim while nonetheless accepting (1) and (2). I then turn to the implications for charitable giving. I present three types of argument against suboptimal charitable giving, based on claims like (2). I show that, while the first two appeal to controversial general principles, the third does not, and constitutes a flexible tool for arguing against a variety of forms of suboptimal charitable giving.

In this talk I examine the role of parameter calibration in the confirmation and validation of complex computer simulation models. Focusing in particular on Bayesian approaches to confirmation. I distinguish several different Bayesian approaches to confirmation and propose a novel Bayesian argument for the thesis that complex simulation models exhibit a predictivist effect: Complex computer simulation models constitute a case in which predictive success, as opposed to the mere accommodation of evidence, provides a more stringent test of a model.

Two of our current PhD students present their research to the LSE Choice Group.

We interact with Recommender Systems on a regular basis when we use online services and apps. They shape individual experience of online environments and social interactions. They collect, curate, and act upon vast amounts of data. Given their significant impact, they deserve urgent ethical scrutiny. However, research into this area is still in its infancy. The debate is fragmented across different scientific communities, who tend to focus on specific aspects and applications of RS in a variety of context, and is hampered by proprietary issues, making it difficult for independent researchers to access information about RS operations.

Building on a systematic review of the existing literature, I identify six areas of concern posed by RS, and map them onto a proposed taxonomy of different kinds of ethical impacts. The analysis uncovers a gap in the literature: currently user-centred approaches to do not consider the interests of a variety of other stakeholders—as opposed to just the receivers of a recommendation—in assessing the ethical impacts of a recommender system. Going forward, we need to develop a better understanding of what, exactly, constitutes a good recommendation, and how to capture it using quantifiable metrics.

Gilles Campagnolo, Research Professor (Economic Philosophy) at CNRS Aix-Marseille, will be giving a short presentation on a recently funded network on 'Economic Philosophy' for which he is the PI and which includes the CPNSS. The grant includes provision for research exchanges between network members and postdoctoral positions at Aix-Marseille. Gilles' talk will be an opportunity to find out more about the research themes of…