Transformative Experiences, Cognitive Modelling and Affective Forecasting

Marvin Mathony, Michael MesserliErkenntnis 2022 · DOI

TL;DR

This paper investigates whether transformative experiences pose unique challenges for cognitive modeling, a key aspect of decision-making. By operationalizing cognitive modeling as affective forecasting, the authors' meta-analysis reveals that people are equally adept (or inaccurate) at predicting their future happiness regardless of whether the experience is transformative. This finding directly challenges L.A. Paul's influential argument that transformative experiences fundamentally break cognitive modeling.

What Problem Does This Paper Address?

Philosophers, notably L.A. Paul, have argued that "transformative experiences" – those that are both epistemically (teaching something new) and personally (changing core preferences) transformative – present a unique hurdle for decision-making. Specifically, Paul claims that our ability to "cognitively model" these experiences, essentially simulating future outcomes and their subjective value, fails. This failure, she argues, makes rational decision-making about such experiences impossible. However, there's been a lack of empirical research directly testing this claim. This paper aims to bridge that gap by empirically investigating whether the transformative nature of an experience actually hinders cognitive modeling.

Methods

The core of this research is a meta-analysis of existing affective forecasting studies. The authors operationalize "cognitive modeling" as "affective forecasting," which is the ability to predict how future experiences will impact one's happiness. They then compare the accuracy of these predictions for transformative versus non-transformative experiences. The meta-analysis involved:

  1. Categorizing Studies: Studies were classified as investigating transformative experiences if they met both epistemic and personal transformation criteria. Non-transformative experiences were those that did not meet these criteria.
  2. Data Extraction: Effect sizes (Hedges's g) representing the accuracy of affective forecasts were extracted from studies.
  3. Weighted Averaging: A weighted average of effect sizes was calculated for both transformative and non-transformative experience groups, giving more weight to larger studies.
  4. Statistical Comparison: A Welch's t-test was used to compare the average effect sizes, followed by an equivalence test (TOST) to determine if the differences were statistically insignificant.

Datasets

The paper does not introduce new datasets. Instead, it relies on a meta-analysis of 84 studies on affective forecasting previously compiled by Levine et al. (2012). The authors then filtered these studies based on their specific criteria for transformative versus non-transformative experiences.

Key Insights and Ideas

  • Cognitive Modeling as Affective Forecasting: The paper's central move is to equate cognitive modeling with affective forecasting. This allows for empirical testing, as affective forecasting studies directly measure people's ability to predict their future emotional states (happiness).
  • The "Impact Bias" is Universal: The research highlights that people are generally not very good at predicting their future happiness, a phenomenon known as the "impact bias." This inaccuracy appears to be a general feature of human prediction, not specific to transformative experiences.
  • Transformative Nature Doesn't Add Difficulty: The most significant insight is that the transformative nature of an experience does not, in itself, make affective forecasting (and thus cognitive modeling) any more difficult or inaccurate than for non-transformative experiences.

Key Results

The meta-analysis found:

  • Weighted Average Effect Sizes:
    • Non-transformative experiences: g = 0.6969
    • Transformative experiences: g = 0.7057
  • Statistical Equivalence: A Welch's t-test showed no significant difference between the effect sizes (p=0.4860). An equivalence test confirmed that the effect sizes were statistically equivalent within a small effect size margin (d=-0.24 to 0.24).
  • Conclusion: The results indicate that people's ability to predict their future happiness is equally inaccurate for both transformative and non-transformative experiences. This directly contradicts L.A. Paul's claim that transformative experiences pose unique problems for cognitive modeling.

Limitations and Future Work

The authors acknowledge several limitations and suggest avenues for future research:

  • Operationalization of Happiness: While they argue for a broad interpretation of happiness compatible with Paul's view, they admit that the exact relationship between happiness, cognitive modeling, and subjective value in Paul's framework is not fully known. Future research could explore different conceptualizations of happiness.
  • First-Person vs. Third-Person Perspective: The meta-analysis cannot definitively distinguish whether affective forecasting is done from a first-person (experiential) or third-person (observational) perspective. While they argue that Paul's emphasis on "what it is like" locutions suggests a first-person perspective is relevant, they lack the data to fully rule out third-person forecasting. Future studies could explicitly investigate this.
  • Revelatory Value: The paper does not address L.A. Paul's "revelatory approach," which suggests basing decisions on the value of discovering an experience rather than predicting its hedonic outcome. The authors note that revelatory value might be determined independently of cognitive modeling, and future research could explore revelatory forecasting.
  • Impact Bias Debate: The authors acknowledge that the status of the "impact bias" itself is debated. However, they argue their meta-analysis is robust because it focuses on the difference in forecasting accuracy between transformative and non-transformative experiences, regardless of whether the impact bias is a true phenomenon or a procedural artifact.
  • Future Research: The authors suggest that future research could investigate revelatory forecasting and explore the role of first-person versus third-person perspectives in affective forecasting.

Summary updated

Abstract

In the last seven years, philosophers have discussed the topic of transformative experiences. In this paper, we contribute to a crucial issue that is currently under-researched: transformative experiences' influence on cognitive modelling. We argue that cognitive modelling can be operationalized as affective forecasting, and we compare transformative and non-transformative experiences with respect to the ability of affective forecasting. Our finding is that decision-makers' performance in cognitively modelling transformative experiences does not systematically differ from decision-makers' performance in cognitively modelling non-transformative experiences. This claim stands in strict opposition to L.A. Paul's main argument.
BibTeX
@inproceedings{mathony2022transformative,
  author = {Marvin Mathony and Michael Messerli},
  title = {{Transformative Experiences, Cognitive Modelling and Affective Forecasting}},
  booktitle = {Erkenntnis},
  year = {2022},
  doi = {10.1007/s10670-022-00523-z},
}

Loading related content...

Figure