Testing a computational model of subjective well-being: a preregistered replication of Rutledge et al. (2014)

ABSTRACT

Subjective well-being changes over time. While the causes of these changes have been investigated extensively, few attempts have been made to capture these changes through computational modelling. One notable exception is the study by Rutledge et al. [Rutledge, R. B., Skandali, N., Dayan, P., & Dolan, R. J. (2014). A computational and neural model of momentary subjective well-being. Proceedings of the National Academy of Sciences, 111(33), 12252–12257. https://doi.org/10.1073/pnas.1407535111], in which a model that captures momentary changes in subjective well-being was proposed. The model incorporates how an individual processes rewards and punishments in a decision context. Using this model, the authors were able to successfully explain fluctuations in subjective well-being observed in a gambling paradigm. Although Rutledge et al. reported an in-paper replication, a successful independent replication would further increase the credibility of their results. In this paper, we report a preregistered close replication of the behavioural experiment and analyses by Rutledge et al. The results of Rutledge et al. were mostly confirmed, providing further evidence for the role of rewards and punishments in subjective well-being fluctuations. Additionally, the association between personality traits and the way people process rewards and punishments was examined. No evidence for such associations was found, leaving this an open question for future research.

KEYWORDS:

• Momentary subjective well-being
• computational model
• happiness
• prediction error

Acknowledgements

We would like to thank Robb Rutledge for sharing the code of his experiment (now available on Github: https://www.github.com/RutledgeLab/Rutledge-happiness-task) and analyses, and for providing useful comments on an earlier draft of this paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Ethics

This study was approved by the local ethics committee (Social and Societal Ethics Committee at the KU Leuven). All participants had to complete an informed consent before their participation to the experiment.

Data availability statement

As specified in the article, the preregistration to this article can be found on: https://osf.io/krhyz/. Raw and preprocessed data, and R scripts containing the analyses can be found on OSF using the following link: https://osf.io/9g2zw/.

Authors’ contributions

Levi Devos ran this study for his master’s thesis, under supervision of Peter Kuppens and Francis Tuerlinckx. Sebastiaan Pessers translated the estimation code from Matlab to R and created estimation code for the hierarchical model. Niels Vanhasbroeck confirmed the analyses, rewrote the analysis code, and wrote the first version of this manuscript. All authors, with exception of Sebastiaan Pessers, contributed to all subsequent versions of the manuscript and approved the final version.

Notes

1 Rutledge et al. (2014) also measured task-dependent neural activity in the ventral striatum when participants performed this task. However, this neurological part of the study is not part of our replication study and we will thus leave it undiscussed.

2 In this replication paper, we not only adopt the methods, but also the terminology of the original study.

3 Note that we do not consider this to be a formal analysis procedure. Instead, we evaluate the estimates of the different parameters to reach our conclusions.

4 When comparing the current paper to the preregistration, it may be helpful to know that different parts of the method section from the preregistration were relabelled and reorganised. More specifically, the subsection Sampling plan was changed to Participants (completed with demographic information), Individual difference variables was changed to Materials, and Experimental task was placed under Procedure. Furthermore, the Analysis subsection in this paper consists of the subsection Confirmatory analysis plan of the preregistration, and the parts of the Introduction that contained the mathematical information about the computational model and general information about the analysis plan for the individual difference variables.

5 The actual term that should be used here is “the estimated posterior mean of the population mean”. We opted for a shorter version to increase readability.

Additional information

Funding

This work was supported by the Research Fund of KU Leuven under Grant C14/19/054 and C14/17/047; and Fonds Wetenschappelijk Onderzoek under Grant G073317N and G074219N. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.