Katie Steele

The Bayesian maxim for rational learning could be described as conservative change from one probabilistic belief or credence function to another in response to newinformation. Roughly: ‘Hold fixed any credences that are not directly affected by the learning experience.’ This is precisely articulated for the case when we learn that some proposition that we had previously entertained is indeed true (the rule of conditionalisation). But can this conservative-change maxim be extended to revising one’s credences in response to entertaining propositions or concepts of which one was previously unaware? The economists Karni and Vierø (2013, 2015) make a proposal in this spirit. Philosophers have adopted effectively the same rule: revision in response to growing awareness should not affect the relative probabilities of propositions in one’s ‘old’ epistemic state. The rule is compelling, but only under the assumptions that its advocates introduce. It is not a general requirement of rationality, or so we argue. We provide informal counterexamples. And we show that, when awareness grows, the boundary between one’s ‘old’ and ‘new’ epistemic commitments is blurred. Accordingly, there is no general notion of conservative change in this setting.

On the face of it, ethics and decision theory give quite different advice about what the best course of action is in a given situation. In this paper we examine this alleged conflict in the realm of environmental decision-making. We focus on a couple of places where ethics and decision theory might be thought to be offering conflicting advice: environmental triage and carbon trading. We argue that the conflict can be seen as conflicts about other things (like appropriate temporal scales for value assignments, idealisations of the decision situation, whether the conservation budget really is fixed and the like). The good news is that there is no conflict between decision theory and environmental ethics. The bad news is that a great deal of environmental decision modelling may be rather simple minded, in that it does not fully incorporate some of these broader issues about temporal scales and the dynamics of many of the decision situations.

We argue that concerns about double-counting—using the same evidence both to calibrate or tune climate models and also to confirm or verify that the models are adequate—deserve more careful scrutiny in climate modelling circles. It is widely held that double-counting is bad and that separate data must be used for calibration and confirmation. We show that this is far from obviously true, and that climate scientists may be confusing their targets. Our analysis turns on a Bayesian/relative-likelihood approach to incremental confirmation. According to this approach, double-counting is entirely proper. We go on to discuss plausible difficulties with calibrating climate models, and we distinguish more and less ambitious notions of confirmation. Strong claims of confirmation may not, in many cases, be warranted, but it would be a mistake to regard double-counting as the culprit. 1 Introduction2 Remarks about Models and Adequacy-for-Purpose3 Evidence for Calibration Can Also Yield Comparative Confirmation3.1 Double-counting I3.2 Double-counting II4 Climate Science Examples: Comparative Confirmation in Practice4.1 Confirmation due to better and worse best fits4.2 Confirmation due to more and less plausible forcings values5 Old Evidence6 Doubts about the Relevance of Past Data7 Non-comparative Confirmation and Catch-Alls8 Climate Science Example: Non-comparative Confirmation and Catch-Alls in Practice9 Concluding Remarks

In the biomedical context, policy makers face a large amount of potentially discordant evidence from different sources. This prompts the question of how this evidence should be aggregated in the interests of best-informed policy recommendations. The starting point of our discussion is Hunter and Williams’ recent work on an automated aggregation method for medical evidence. Our negative claim is that it is far from clear what the relevant criteria for evaluating an evidence aggregator of this sort are. What is the appropriate balance between explicitly coded algorithms and implicit reasoning involved, for instance, in the packaging of input evidence? In short: What is the optimal degree of ‘automation’? On the positive side: We propose the ability to perform an adequate robustness analysis as the focal criterion, primarily because it directs efforts to what is most important, namely, the structure of the algorithm and the appropriate extent of automation. Moreover, where there are resource constraints on the aggregation process, one must also consider what balance between volume of evidence and accuracy in the treatment of individual evidence best facilitates inference. There is no prerogative to aggregate the total evidence available if this would in fact reduce overall accuracy.

In this paper, we develop a novel response to counterfactual scepticism, the thesis that most ordinary counterfactual claims are false. In the process we aim to shed light on the relationship between debates in the philosophy of science and debates concerning the semantics and pragmatics of counterfactuals. We argue that science is concerned with many domains of inquiry, each with its own characteristic entities and regularities; moreover, statements of scientific law often include an implicit ceteris paribus clause that restricts the scope of the associated regularity to circumstances that are ‘fitting’ to the domain in question. This observation reveals a way of responding to scepticism while, at the same time, doing justice both to the role of counterfactuals in science and to the complexities inherent in ordinary counterfactual discourse and reasoning.

Laurie Paul argues that, when it comes to many of your most significant life-changing decisions, the principles of rational choice are silent. That is because, in these cases, you anticipate that one of your choice options would yield a transformative experience. We argue that such decisions are best seen as ones in which you anticipate awareness growth. You do not merely lack knowledge about which possible outcome will arise from a transformative option; you lack knowledge about what are the possible outcomes. We show how principles of rational choice can be extended to cases of anticipated awareness growth.

This article argues that common intuitions regarding (a) the specialness of ‘use-novel’ data for confirmation and (b) that this specialness implies the ‘no-double-counting rule’, which says that data used in ‘constructing’ (calibrating) a model cannot also play a role in confirming the model’s predictions, are too crude. The intuitions in question are pertinent in all the sciences, but we appeal to a climate science case study to illustrate what is at stake. Our strategy is to analyse the intuitive claims in light of prominent accounts of confirmation of model predictions. We show that on the Bayesian account of confirmation, and also on the standard classical hypothesis-testing account, claims (a) and (b) are not generally true; but for some select cases, it is possible to distinguish data used for calibration from use-novel data, where only the latter confirm. The more specialized classical model-selection methods, on the other hand, uphold a nuanced version of claim (a), but this comes apart from (b), which must be rejected in favour of a more refined account of the relationship between calibration and confirmation. Thus, depending on the framework of confirmation, either the scope or the simplicity of the intuitive position must be revised.

There has been much recent interest in imprecise probabilities, models of belief that allow unsharp or fuzzy credence. There have also been some influential criticisms of this position. Here we argue, chiefly against Elga (2010), that subjective probabilities need not be sharp. The key question is whether the imprecise probabilist can make reasonable sequences of decisions. We argue that she can. We outline Elga's argument and clarify the assumptions he makes and the principles of rationality he is implicitly committed to. We argue that these assumptions are too strong and that rational imprecise choice is possible in the absence of these overly strong conditions.

In this paper we explore the connections between ethics and decision theory. In particular, we consider the question of whether decision theory carries with it a bias towards consequentialist ethical theories. We argue that there are plausible versions of the other ethical theories that can be accommodated by “standard” decision theory, but there are also variations of these ethical theories that are less easily accommodated. So while “standard” decision theory is not exclusively consequentialist, it is not necessarily ethically neutral. Moreover, even if our decision-theoretic models get the right answers vis-`a-vis morally correct action, the question remains as to whether the motivation for the non-consequentialist theories and the psychological processes of the agents who subscribe to those ethical theories are lost or poorly represented in the resulting models.

This paper considers a puzzling conflict between two positions that are each compelling: it is irrational for an agent to pay to avoid `free' evidence before making a decision, and rational agents may have imprecise beliefs and/or desires. Indeed, we show that Good's theorem concerning the invariable choice-worthiness of free evidence does not generalise to the imprecise realm, given the plausible existing decision theories for handling imprecision. A key ingredient in the analysis, and a potential source of controversy, is the general approach taken for resolving sequential decision problems { we make explicit what the key alternatives are and defend our own approach. Furthermore, we endorse a resolution of the aforementioned puzzle { we privilege decision theories that merely permit avoiding free evidence over decision theories for which avoiding free evidence is uniquely admissible. Finally, we situate this particular result about free evidence within the broader `dynamic-coherence' literature.

Many fine-grained decisions concerning climate change involve significant, even severe, uncertainty. Here, we focus on modelling the decisions of single agents, whether individual persons or groups perceived as corporate entities. We offer a taxonomy of the sources and kinds of uncertainty that arise in framing these decision problems, as well as strategies for making a choice in spite of uncertainty. The aim is to facilitate a more transparent and structured treatment of uncertainty in climate decision making

In engineering, as in other scientific fields, researchers seek to confirm their models with real-world data. It is common practice to assess models in terms of the distance between the model outputs and the corresponding experimental observations. An important question that arises is whether the model should then be ‘tuned’, in the sense of estimating the values of free parameters to get a better fit with the data, and furthermore whether the tuned model can be confirmed with the same data used to tune it. This dual use of data is often disparagingly referred to as ‘double-counting’. Here, we analyse these issues, with reference to selected research articles in engineering. Our example studies illustrate more and less controversial practices of model tuning and double-counting, both of which, we argue, can be shown to be legitimate within a Bayesian framework. The question nonetheless remains as to whether the implied scientific assumptions in each case are apt from the engineering point of view

Richard Rudner famously argues that the communication of scientific advice to policy makers involves ethical value judgments. His argument has, however, been rightly criticized. This article revives Rudner’s conclusion, by strengthening both his lines of argument: we generalize his initial assumption regarding the form in which scientists must communicate their results and complete his ‘backup’ argument by appealing to the difference between private and public decisions. Our conclusion that science advisors must, for deep-seated pragmatic reasons, make value judgments is further bolstered by reflections on how the scientific contribution to policy is far less straightforward than the Rudner-style model suggests

There are at least two plausible generalisations of subjective expected utility (SEU) theory: cumulative prospect theory (which relaxes the independence axiom) and Levi’s decision theory (which relaxes at least ordering). These theories call for a re-assessment of the minimal requirements of rational choice. Here, I consider how an analysis of sequential decision making contributes to this assessment. I criticise Hammond’s (Economica 44(176):337–350, 1977; Econ Philos 4:292–297, 1988a; Risk, decision and rationality, 1988b; Theory Decis 25:25–78, 1988c) ‘consequentialist’ argument for the SEU preference axioms, but go on to formulate a related diachronic-Dutch-book-style’ argument that better achieves Hammond’s aims. Some deny the importance of Dutch-book sure losses, however, in which case, Seidenfeld’s (Econ Philos 4:267–290, 1988a) argument that distinguishes between theories that relax independence and those that relax ordering is relevant. I unravel Seidenfeld’s argument in light of the various criticisms of it and show that the crux of the argument is somewhat different and much more persuasive than what others have taken it to be; the critical issue is the modelling of future choices between ‘indifferent’ decision-tree branches in the sequential setting. Finally, I consider how Seidenfeld’s conclusions might nonetheless be resisted.

I focus my discussion on the well-known Ellsberg paradox. I find good normative reasons for incorporating non-precise belief, as represented by sets of probabilities, in an Ellsberg decision model. This amounts to forgoing the completeness axiom of expected utility theory. Provided that probability sets are interpreted as genuinely indeterminate belief, such a model can moreover make the “Ellsberg choices” rationally permissible. Without some further element to the story, however, the model does not explain how an agent may come to have unique preferences for each of the Ellsberg options. Levi holds that the extra element amounts to innocuous secondary “risk” or security considerations that are used to break ties when more than one option is rationally permissible. While I think a lexical choice rule of this kind is very plausible, I argue that it involves a greater break with xpected utility theory than mere violation of the ordering axiom

We argue that concerns about double-counting -- using the same evidence both to calibrate or tune climate models and also to confirm or verify that the models are adequate --deserve more careful scrutiny in climate modelling circles. It is widely held that double-counting is bad and that separate data must be used for calibration and confirmation. We show that this is far from obviously true, and that climate scientists may be confusing their targets. Our analysis turns on a Bayesian/relative-likelihood approach to incremental confirmation. According to this approach, double-counting is entirely proper. We go on to discuss plausible difficulties with calibrating climate models, and we distinguish more and less ambitious notions of confirmation. Strong claims of confirmation may not, in many cases, be warranted, but it would be a mistake to regard double-counting as the culprit.

This paper considers a special case of belief updating—when an agent learns testimonial data, or in other words, the beliefs of others on some issue. The interest in this case is twofold: (1) the linear averaging method for updating on testimony is somewhat popular in epistemology circles, and it is important to assess its normative acceptability, and (2) this facilitates a more general investigation of what it means/requires for an updating method to have a suitable Bayesian representation (taken here as the normative standard). The paper initially defends linear averaging against Bayesian-compatibility concerns raised by Bradley (Soc Choice Welf 29:609-632, 2007), as well as problems associated with multiple testimony updates. The resolution of these issues, however, requires an extremely nuanced interpretation of the parameters of the linear averaging model—the so-called weights of respect. We go on to propose a role that the parameters of any 'shortcut' updating function should play, by way of minimal interpretation of these parameters. The class of updating functions that is consistent with this role, however, excludes linear averaging, at least in its standard form

This paper considers a key point of contention between classical and Bayesian statistics that is brought to the fore when examining so-called ‘persistent experimenters’—the issue of stopping rules, or more accurately, outcome spaces, and their influence on statistical analysis. First, a working definition of classical and Bayesian statistical tests is given, which makes clear that (1) once an experimental outcome is recorded, other possible outcomes matter only for classical inference, and (2) full outcome spaces are nevertheless relevant to both the classical and Bayesian approaches, when it comes to planning/choosing a test. The latter point is shown to have important repercussions. Here we argue that it undermines what Bayesians may admit to be a compelling argument against their approach—the Bayesian indifference to persistent experimenters and their optional stopping rules. We acknowledge the prima facie appeal of the pro-classical ‘optional stopping intuition’, even for those who ordinarily have Bayesian sympathies. The final section of the paper, however, provides three error theories that may assist a Bayesian in explaining away the apparent anomaly in their reasoning

We sometimes seem to face fundamental moral uncertainty, i.e., uncertainty about what is morally good or morally right that cannot be reduced to ordinary descriptive uncertainty. This phenomenon raises a puzzle for noncognitivism, according to which moral judgments are desire-like attitudes as opposed to belief-like attitudes. Can a state of moral uncertainty really be a noncognitive state? So far, noncognitivists have not been able to offer a completely satisfactory account. Here, we argue that noncognitivists should exploit the formal analogy between moral uncertainty and moral multi-objective decision problems. Our ‘multi-objective story’ enables noncognitivists to save our moral uncertainty thought and talk by explaining how the underlying phenomenon could be noncognitive.

Group decisions raise a number of substantial philosophical and methodological issues. We focus on the goal of the group decision exercise itself. We ask: What should be counted as a good group decision-making result? The right decision might not be accessible to, or please, any of the group members. Conversely, a popular decision can fail to be the correct decision. In this paper we discuss what it means for a decision to be "right" and what components are required in a decision process to produce happy decision-makers. Importantly, we discuss how "right" decisions can produce happy decision-makers, or rather, the conditions under which happy decision-makers and right decisions coincide. In a large range of contexts, we argue for the adoption of formal consensus models to assist in the group decision-making process. In particular, we advocate the formal consensus convergence model of Lehrer and Wagner (1981), because a strong case can be made as to why the underlying algorithm produces a result that should make each of the experts in a group happy. Arguably, this model facilitates true consensus, where the group choice is effectively each person's individual choice. We analyse Lehrer and Wagner's algorithm for reaching consensus on group probabilities/utilities in the context of complex decision-making for conservation biology. While many conservation decisions are driven by a search for objective utility/probability distributions (regarding extinction risks of species and the like), other components of conservation management primarily concern the interests of stakeholders. We conclude with cautionary notes on mandating consensus in decision scenarios for which no fact of the matter exists. For such decision settings alternative types of social choice methods are more appropriate.

Many examples of calibration in climate science raise no alarms regarding model reliability. We examine one example and show that, in employing Classical Hypothesis-testing, it involves calibrating a base model against data that is also used to confirm the model. This is counter to the "intuitive position". We argue, however, that aspects of the intuitive position are upheld by some methods, in particular, the general Cross-validation method. How Cross-validation relates to other prominent Classical methods such as the Akaike Information Criterion and Bayesian Information Criterion is also discussed.

Summary 1. Ecologists and conservation biologists consider many issues when designing a field study, such as the expected value of the data, the interests of the study species, the welfare of individual organisms and the cost of the project. These different issues or values often conflict; however, neither animal ethics nor environmental ethics provides practical guidance on how to assess trade-offs between them. 2. We developed a decision framework for considering trade-offs between values in ecological research, drawing on the field of ecological ethics. We used a case study of the population genetics of three frog species, in which a researcher must choose between four methods of sampling DNA from the study animals. We measured species welfare as the reduction in population growth rate following sampling, and assessed individual welfare using two different definitions: (i) the level of suffering experienced by an animal, and (ii) the level of suffering combined with loss of future life. 3. Tipping the tails of tadpoles ranked as the best sampling method for species welfare, while collecting whole tadpoles and buccal swabbing of adult frogs ranked best for the first and second definitions of individual welfare, respectively. Toe clipping of adult frogs ranked as the worst sampling method for species welfare and the first definition of individual welfare, and equal worst for the second definition of individual welfare. 4. When considering species and individual welfare simultaneously, toe clipping was clearly inferior to the other sampling methods, but no single sampling method was clearly superior to the other three. Buccal swabbing, collecting tadpoles and tail tipping were all preferred options, depending on the definition of individual welfare and the level of precision with which we assessed species welfare. 5.Synthesis and applications. The decision framework we present can be used by ecologists to assess ethical and other trade-offs when planning field studies. A formal decision analysis makes transparent how a researcher might negotiate competing ethical, financial and practical objectives. Defining the components of the decision in this way can help avoid errors associated with human judgement and linguistic uncertainty.