The Explanatory Potential of Artificial Societies, Synthese 169:539-555.
It is often claimed that artificial society simulations contribute to the explanation of social phenomena. At the hand of a particular example, this paper argues that artificial societies often cannot provide full explanations, because their models are not or cannot be validated. Instead, artificial societies may sometimes provide potential explanations. It is shown that these potential explanations, if they contribute to our understanding, considerably differ from the standard kind of potential causal explanations. Instead of possible causal histories, simulations offer possible functional analyses of the explanandum. The paper discusses how these two kinds of potential explanations differ, and how possible functional analyses can be appraised.
Learning from Minimal Economic Models, Erkenntnis 70(1): 81-99.
It is argued that one can learn from minimal economic models. Minimal models are models that are not similar to the real world, do not resemble some of its features, and do not adhere to accepted regularities. One learns from a model if constructing and analysing the model affects one’s confidence in hypotheses about the world. Economic models, I argue, are often assessed for their credibility. If a model is judged credible, it is considered to be a relevant possibility. Considering such relevant possibilities may affect one’s confidence in necessity or impossibility hypotheses. Thus, one can learn from minimal economic models.
The Roles of Stories in Applying Game Theory (with P. Schweinzer), Journal of Economic Methodology 15(2): 131–46.
Game theoretic models consist of a formal game form and an informal model narrative or story. When game theory is employed to model economic situations, the stories play a central role in interpreting, constructing and solving game forms. We analyze the architecture of game theory and distinguish between game models and the theory proper. We present the different functions of the model narrative in the application of game models to economic situations. In particular, we show how model narratives support the choice of solution concepts defined and provided by the theory proper. We further argue that the narrative’s role in interpretation, construction and solution makes it a necessary part of a game model that is intended to be a model of an economic situation. We conclude that game theory is not a universal theory of rationality, but only offers tools to model specific situations at varying degrees and kinds of rationality.
Bounded Rationality, Philosophy Compass 2 (3): 534-563.
The notion of bounded rationality has recently gained considerable popularity in the behavioural and social sciences. This article surveys the different usages of the term, in particular the way ‘anomalous’ behavioural phenomena are elicited, how these phenomena are incorporated in model building, and what sort of new theories of behaviour have been developed to account for bounded rationality in choice and in deliberation. It also discusses the normative relevance of bounded rationality, in particular as a justifier of non-standard reasoning and deliberation heuristics. For each of these usages, the overview discusses the central methodological problems.
Game Theory, Internet Encyclopaedia of Philosophy
This article presents the basic tenets of game theory in a non-formal way. It then discusses two broad philosophical issues arising from the theory. First, whether the rationality concept employed by the theory is justifiable – whether it is intuitively rational to choose as the theory prescribes. Second, whether the theory can in principle be a good predictive theory of human behavior – whether it has empirical content, whether it is testable and whether there are good reasons to believe that it is true or false.
'Philosophy of game theory (with Aki Lehtinen), forthcoming in Handbook of the Philosophy of Economics, ed. Uskali Mäki, Elsevier.
Philosophy and game theory are connected in multiple ways. Game theory has been used as a tool in philosophical discussions, and some crucial game theoretical concepts have been developed by philosophers.2 Game theory also has been the object of philosophical inquiry itself. Our discussion will concentrate on the latter. Since game theory relies heavily on mathematical models, the standard epistemic issues concerning modelling and unrealistic assumptions in philosophy of economics are also relevant for game theory. But since game theory transcends economics, a number of other philosophical issues also arise. Perhaps the most important of these is the interpretation of the theory: is game theory to be understood mainly as a tool for recommending rational choices, for predicting agents’ behaviour, or for merely providing an abstract framework for understanding complex interactions (e.g., Blackburn 1998; Aydinonat 2008)? If we settle for the first interpretation, the issue of whether the rationality concept employed by the theory is justifiable becomes pressing. Is it intuitively rational to choose as the theory prescribes? If the second interpretation is adopted, one must ask whether the theory can in principle be a good predictive theory of human behaviour: whether it has empirical content, whether it is testable and whether there are good reasons to believe that it is true or false. If the third interpretation is adopted, the question arises concerning which qualities of the theory contribute to this understanding, and to what extent these qualities are different from the prescriptive or predictive function discussed in the first two interpretations.
Artificial Worlds and Simulation, forthcoming in Sage Handbook of Philosophy of Social Science ed. J. Zamora Bonilla and I. Jarvie
Agent-based simulations (ABS) are often hailed as a new way of doing social science. By generating ‘artificial societies’ on the computer, social scientists may generate a new object of study, or a new tool for scientific study. This article investigates these claims by examining in detail the methods of ABS and its uses in the social sciences. It starts out by illustrating ABS different uses, and giving a short historical overview of its emergence. ABS is then contrasted with related practices in the social sciences and other disciplines. A brief philosophical discussion of some of its basic concepts follows. After this overview, the article addresses the novelty claim head on, and discusses it in relation to existing scientific practices. The general position here will be cautious: the author’s view is that ABS does not so much constitute a novel scientific practice, but is closely linked to the existing modelling practices in the social sciences. Thus the view of ABS as experiments or as theory will be treated critically, while the interactions between these practices will be shown to be fruitful. The view of ABS as models, in contrast, will be presented in a more favourable light, but important differences that concern the way of construction, the validation and link to the target system are pointed out. Finally, potential reasons for the notable difference with which different disciplines have adopted ABS will be discussed. Based on these results, the uses of ABS in the social sciences are investigated and appraised. This discussion will clarify the differences between the various ABS models. In particular, it is shown that uses determine many characteristics of ABS models, in terms of their level of abstraction, their necessary validation, and their need for replication. Again, the position presented here will be cautious: against the many enthusiastic claims made about ABS, this article points out the various deficits that many contemporary simulation studies exhibit, particularly when in the business of explanation or forecasting. Instead, the articles points out uses with somewhat lesser profiles, like deliberation support and heuristic application, as important areas where ABS can be fruitfully employed.
Preface to Economic Models as Credible Worlds or as Isolating Tools?, Special issue of Erkenntnis 70(1): 1-2.
Review of Cognitive Economics by P. Bourgine and J.P. Nadal, Economics and Philosophy 22 (2006): 1–8.
Review of The Elgar Companion to Economics and Philosophy by J. B. Davis, A. Marciano and J. Runde, Theoria 2006(3): 254-259.
From Belief Revision to Preference Change (with S. O. Hansson), forthcoming in Preference Change: Approaches from Philosophy, Economics and Psychology, ed. Grüne-Yanoff and Hansson, Theory and Decision Library A, Springer: 159-184.
We propose to model the consistency-preserving aspect of preference change after the fashion of belief revision. First, we discuss the formal properties of the preference notion. Second, we discuss the various consistency requirements imposed on preference sets. Third, we discuss representations of consistency-driven preference change and compare them to models of belief change. Last, we discuss the specific needs of introducing a priority index in models of preference change. We conclude that while the general input-assimilating framework from belief change can be transferred to preference change, several modifications are necessary. In particular, the input model has to be complicated with the introduction of a distinction between primary (non-linguistic) and secondary (linguistic) inputs. Sentential representation has to be used with somewhat more caution for preferences than for beliefs. The prioritysetting mechanism has to be adjusted, and priority-related information must be included in the inputs.
Hume’s Framework for a Natural History of the Passions (with Edward McClennen) in: David Hume's Political Economy, edited by Carl Wennerlind and Margaret Schabas (London: Routledge): 86-104.
We discuss Hume’s dynamic framework of passion change. We argue for three claims.First, passions in Hume's framework are the basic explanans of human action, but change is a result of historical developments. Second, the historical development of commerce according to Hume is driven by the rise of interest. However, contrary to Hirschman’s thesis, Hume evaluates this passion ambivalently. Third, the version of interest that Hirschman neglects, but which is salient in Hume’s work, is not identical to love of gain; instead, it includes interests of a much more general sort, and also includes rational self-control through both “internal” and “external” restraints.
Preference Change: An Introduction(with S. O. Hansson), forthcoming in Preference Change: Approaches from Philosophy, Economics and Psychology ed. Grüne-Yanoff and Hansson, Theory and Decision Library A, Springer: 1-26.
In this introduction, we discuss a number of reasons why preference change has been neglected in the social sciences, in particular in economics. We argue that recent developments make this neglect less acceptable than it may have been in the past. We then propose a modelling approach to preference change that starts out with the standard preference notion and pays careful attention to its formal properties, in particular the connections between preference relata, the logical constraints on preferences, and their temporal specification. Based on this proposal, we categorise preference change models into four groups: those that derive changed preference from more basic structures, those that refer to the temporal dimension, those that focus on consistency preservation, and finally those that offer an evolutionary account. Using this categorization, we also introduce the other papers of this anthology.
Review of The Structure of Values and Norms, by S.O. Hansson, Economics and Philosophy, 20(2): 396-403.
Preferences (with S.O. Hansson), The Stanford Encyclopedia of Philosophy, Edward N. Zalta (ed.)
The notion of preference has a central role in many disciplines, including moral philosophy and decision theory. Preferences and their logical properties also have a central role in rational choice theory, a subject that in its turn permeates modern economics, as well as other branches of formalized social science. The notion of preference and the way it is analysed vary between these disciplines. A treatment is still lacking that takes into account the needs of all usages and tries to combine them in a unified approach. This entry surveys the most important philosophical uses of the preference concept and investigates their compatibilities and conflicts.
Intentional Action Explanations are Not Inherently Normative Theoria 74(1): 60-78.
Inherent normativity is the claim that intentional action explanations necessarily have to comply with normatively understood rationality constraints on the ascribed propositional attitudes. This paper argues against inherent normativity in three steps. First, it presents three examples of actions successfully explained with propositional attitudes, where the ascribed attitudes violate relevant rationality constraints. Second, it argues that the inference rules that systematise propositional attitudes are qualitatively different from rationality constraints both in the justification and their recipients. Third, it rejects additional conditions on propositional attitudes, which purport to necessitate a normative commitment. Thus, inherent normativity is rejected; and with it the claim that intentional action explanations differ substantially from other explanations because they are inherently normative.
‘Why don’t you want to be rich? Preference explanations on the basis of causal structure’ in: Campbell, J. K., O'Rourke, M. (eds.): Topics in Contemporary Philosophy: Explanation and Causation , MIT Press: 217-240.
When social scientists explain an agent’s actions, they often employ preferences of different degrees of individuation: preferences either between different concrete options the agent faces, worlds, or preferences between highly abstract aspects, prospects. Preferences over worlds, I argue, are the only ones that can be derived from observed behaviour; but preferences over prospects are necessary for the explanation and prediction of behaviour. If the outcomes over which preferences are defined are too specific, then the theory is empty – one cannot explain any situations with it that are not exact repetitions of past choices. Thus for successful action explanation a principle of equivalence is needed that connects preferences over worlds with preferences over prospects. I provide such a principle that is based on a model of causal beliefs, and that is acceptable on the basis of very weak plausibility considerations.
The problems of testing preference axioms with revealed preference theory, Analyse & Kritik , 26/2 (2004): 382-397.
In economics, it has often been claimed that testing choice data for violation of certain axioms – particularly if the choice data is observed under laboratory conditions – allows conclusions about the validity of certain preference axioms and the neoclassical maximization hypothesis. In this paper I argue that these conclusions are unfounded. In particular, it is unclear what exactly is tested, and the interpretation of the test results are ambiguous. Further, there are plausible reasons why the postulated choice axioms should not hold. Last, these tests make implicit assumptions about beliefs that further blur the interpretations of the results. The tests therefore say little if anything about the validity of certain preference axioms or the maximization hypothesis.
Mismeasuring the Value of Statistical Life, Journal of Economic Methodology 16(2); 109-123.
The value of a statistical life (VSL) is an important tool for cost-benefit analysis of regulatory policies that concern fatality risks. Its proponents claim that it measures people’s risk preferences, and that VSL therefore is a tool of vicarious governance. This paper criticizes the revealed preference method for measuring VSL. It specifies three minimal conditions for vicarious governance, sensitivity, fairness and hypothetical compromise, and shows that the VSL measure, in its common application in policy formation and analysis, violates these conditions. It therefore concludes that the revealed preference VSL measure, in its current form, is not a tool of vicarious governance.
Old Wine In New Casks: Libertarian Paternalism Still Violates Liberal Principles under review.
Champions of Libertarian Paternalism (LP) claim that it is ‘liberty-preserving’, and that under its policies, ‘people should be free to do what they like’ (Thaler and Sunstein 2008, 5). Based on this claim, they suggest that Libertarian Paternalism is compatible with liberal positions. In this paper, I argue against this compatibility claim. Libertarian Paternalism, I show, violates core liberal principles, first because it limits freedom, and second because it fails to justify these limitations in ways acceptable to liberal positions. In particular, section 2 argues that (i) some LP policies constitute non-transparent manipulations, and hence reduce people’s degree of republican liberty, and that (ii) some LP policies interfere in choice processes, and hence reduce people’s degree of negative liberty. Section 3 clarifies LP’s justification of these limitations of liberty as a trade-off between liberty and people’s welfare. Section 4 shows that the welfare notions LP uses respect neither the subjectivity nor the plurality of values, and hence are not compatible with liberal positions.