SOPhiA 2022

Salzburgiense Concilium Omnibus Philosophis Analyticis

SOPhiA ToolsDE-pageEN-page

Programm - Vortrag

Behaviourism vs. Mentalism: Two Approaches to Decision Theory and the Normative Implications of Modelling Practice
(Philosophy of Science, Englisch)

Behaviourism vs. Mentalism:

Two Approaches to Decision Theory and the Normative Implications of Modelling Practice


In her thought-provoking framework of transformative experiences, Laurie Paul (2014, 2015) challenges standard decision theory for its adequacy (1) to model choice settings and (2) to suggest viable rationality norms, when subjective values of the outcomes are undetermined. Paul's critique is based on the epistemic limitations of human agents to know the subjective effects of new experiences on their understanding and assessments. Specifically, she argues that fundamentally new experiences provide new knowledge, which transforms individuals' understanding, sense of self, interests and values in incomprehensible ways. When a decision-maker needs to compare the alternative futures, under which she will have significantly different lived experiences; from her epistemically impoverished first-personal point of view, she cannot assign subjective values. Nor can she decide how she values the alternative selves that she may become, to choose which self she wants to be rationally. Paul's critique aimed at the normative interpretations of standard decision theory and is in two folds. She first questions the theory's object of interest, i.e., subjective values, as a measure to rank and compare the alternatives, in the case of transformative experiences. She also opposes the normative bearing of the rational choice frameworks on real-world decision-makers in the cases of "transformative choices".

Two issues arise regarding in what ways Paul's critique applies to decision theory. First one is about value-formation and its use within the domain of decision theory, i.e., what subjective values represent. Secondly, it is important to establish what the normative implications of decision theory can be. More specifically, what normative truths can be derived from the theoretical frameworks, and whether these truths inform the real-world decision-makers on how they ought to make rational choices. I argue that Paul's critique about subjective values only apply to decision theory if the concept of subjective values concerns the psychological reals and cognitive limitations of decision-makers. I also argue that norms for rational action cannot necessarily be derived from the normative implications of decision models. The formal solutions to represent transformative choices in a decision theoretic framework may not provide action-guidance for rational decision-making to real-world agents.

I use Titelbaum (2013, 2021) to design the inquiry. In his exploratory work on how to practice normative modelling, Titelbaum argues that even the models that are used for normative goals, they are often constructed descriptively: they characterize and explain regularities in choice. That is, when the modeller's goal is to fit real-world phenomena in the model, the outcome is a descriptive framework. Titelbaum, on the other hand, proposes to construct models as distinct objects of study from the targets of the models, and to focus on how to interpret their relations with their targets -before introducing novel phenomena to study with the model. There are two sorts of interpretations necessary for normative modelling: First, to determine how to apply the model, i.e., how to use the formal tools to represent the features of the world. Secondly, to determine what normative truths can be obtained from the outcomes of the model. Titelbaum argues that this approach is useful for interpreting the implications of the models and for acquiring normative truths.

By following Titelbaum's approach, I provide a comparison of the conceptualisation of subjective values and preferences under two approaches to decision theory: mentalism and behaviourism (Okasha, 2016; Dietrich & List, 2016). Behaviourism and mentalism are views of decision theory that differ in their interpretation of preferences and beliefs and in their postulated evidential basis. For behaviourists, the relations and functions describing preferences and beliefs should be interpreted as abstract theoretical concepts, which are useful to describe behaviour patterns and to organize empirical regularities. Here, preferences are the primary source of data and domain of application. Utilities and credences are inferential outcomes of the framework and secondary for the efforts of representation and explanation. Accordingly, subjective values are a derived outcome of the framework, not the domain of study. As such, epistemic limitations do not present an obstacle for the behaviourists to represent rational preferences, so long as preference-formation conforms to the rationality rules.

For the mentalists, the objects of decision-theoretic functions are representations of real psychological or mental phenomena. Utility is a real and independent psychological factor, not derived from other objects such as preferences. Whereas preferences and choices are the outcomes of agents' assessments of their utilities and credences. Accordingly, the problems that Paul puts forward present an obstacle for the mentalist approach, as subjective understanding of decision-makers concerns the domain for representation.

Regarding the normative implications, behaviourist framework does not indicate that decision-makers ought to imagine and evaluate their decisions by applying the axiomatic structure of the framework in their thinking. The choice functions represent rational choice "as if" decision-makers order their preferences in accordance with a utility function. If the "as if" condition holds, decisions can be deemed rational. Accordingly, this approach can inform real-world decision-makers regarding transformative phenomena, should they assess whether their preferences are consistent with the axioms, without the need to assign a cognitively accessible value to their alternatives.

Since in their framework, the value function&_8239;should&_8239;correspond to psychological reals and explain preferences, mentalists will be compelled to reconfigure their representation models to accommodate undetermined subjective values. The normative implications of the mentalist approach then depend on the outcomes of the modified models. We can view (Steele & Stefansson, forthcoming) as an example of such attempt. Their framework provides how awareness growth can account for limits in understanding due to epistemic boundaries. They explain that when awareness growth is taken into consideration, transformative decisions can take place rationally. However, the framework does not provide guidance to decision-makers themselves due to its descriptive nature as understood by Titelbaum. That is, it is unclear how decision-makers should account for awareness growth when they assess the subjective values of their alternatives.


Dietrich, F., & List, C. (2016). Mentalism versus behaviourism in economics: a philosophy-of-science perspective.&_8239;Economics and Philosophy,&_8239;32(2), 249-281.&_8239;

Okasha, S. (2016). On the interpretation of decision theory. Economics and Philosophy, 32(3), 409-433.

Paul, L. A. (2014). Transformative Experience. Oxford: Oxford University Press.

Paul, L. A. (2015). Transformative Choice: Discussions and Replies. Res Philosophica, 92(2): 473-545.

Steele, Katie & Stefansson, H. Orri (forthcoming). Transformative experience, awareness growth and the limits of rational planning. Philosophy of Science.

Titelbaum, M. (2013).&_8239;Quitting certainties: A bayesian framework modeling degrees of belief&_8239;(1st ed.) _1st ed._. Oxford: Oxford University Press.

Titelbaum, Michael G. (2021) Normative Modeling. Preprint URL:

Chair: Ina Jängten
Zeit: 12:00-12:30, 08. September 2022 (Donnerstag)
Ort: SR 1.004

Oyku Ulusoy
(University of Bristol, United Kingdom)

Testability and Meaning deco