Individual differences in information demand have a low dimensional structure predicted by some curiosity traits

Hayley Jach, Roshan Cools, Alec Frisvold, Michael A. Grubb, Catherine A. Hartley, Jochen Hartmann, Laura Hunter, Ruonan Jia, Floris P. de Lange, Ruby Larisch, Rosa Lavelle-Hill, Ifat Levy, Yutong Li, Lieke van Lieshout, Kate Nussenbaum, Silvio Ravaioli, Siyu Wang, Robert C. Wilson, Michael Woodford, Kou Murayama, Jacqueline GottliebPsyArXiv 2023 · DOI

Abstract

To understand human learning and progress, it is crucial to understand curiosity. But how consistent is curiosity’s conception and assessment across scientific research disciplines? We present the results of a large collaborative project assessing the correspondence between curiosity measures in personality psychology and cognitive science. 820 participants completed 15 personality trait measures and 9 cognitive tasks that tested multiple aspects of information demand. We show that shared variance across the cognitive tasks was captured by a dimension reflecting directed (uncertainty-driven) versus random (stochasticity-driven) exploration and individual differences along this axis were significantly and consistently predicted by personality traits. However, the personality metrics that best predicted information demand were not the central curiosity traits of openness to experience, deprivation sensitivity, and joyous exploration, but instead included more peripheral curiosity traits (need for cognition, thrill seeking, and stress tolerance) and measures not traditionally associated with curiosity (extraversion and behavioral inhibition). The results suggest that the umbrella term “curiosity” reflects a constellation of cognitive and emotional processes, only some of which are shared between personality measures and cognitive tasks. The results reflect the distinct methods that are used in these fields, indicating a need for caution in comparing results across fields and for future interdisciplinary collaborations to strengthen our emerging understanding of curiosity.

Introduction: The "Jingle Fallacy" in Curiosity Research

Curiosity is widely recognized as the engine of human learning, yet its scientific definition remains fragmented. Personality psychologists typically measure curiosity through self-report scales (traits), while cognitive scientists quantify it through behavior in controlled experimental tasks (information demand). This divergence leads to the "jingle fallacy": the assumption that because both fields use the term "curiosity," they are measuring the same underlying psychological construct.

If personality scales do not reliably predict how individuals actually seek information in a laboratory setting, our interdisciplinary understanding of learning rests on a precarious foundation. This study aims to bridge this gap by examining the correspondence between these two distinct methodological worlds in a large-scale, collaborative effort.

Methodology

The researchers recruited 820 participants via Amazon Mechanical Turk to complete a comprehensive battery of 9 cognitive tasks and 15 personality trait measures. The analytical approach followed a two-step pipeline:

  1. Principal Component Analysis (PCA): To identify a common structure across diverse behavioral measures, the team extracted 35 behavioral parameters from the 9 tasks (e.g., the Chips task for causal learning and the Horizon task for explore-exploit trade-offs). PCA was then used to reduce these parameters into a primary behavioral dimension.
  2. Machine Learning Prediction: Using the behavioral dimension as a Dependent Variable (DV), the researchers employed Elastic Net regression and Random Forest models to determine which of the 15 personality traits (Independent Variables, IVs) could best predict a participant's information-seeking style.
Overview of the analytical pipeline: (A) PCA on behavioral parameters to derive a primary Dependent Variable (DV); (B) Machine learning workflow to predict the DV from personality traits.

Key Insights and Results

The analysis revealed a robust, low-dimensional structure in human information-seeking behavior.

The Directed-Random Axis

The primary behavioral dimension (PC1) captured a continuum between Directed Exploration—where individuals calculate which information most effectively reduces uncertainty—and Random Exploration—where individuals introduce stochasticity to discover new options. This axis remained consistent regardless of whether the tasks involved monetary incentives or were purely non-instrumental.

Predictive Power and Unexpected Winners

Personality traits successfully predicted behavioral scores with a correlation of $r = .44$, explaining approximately 18% of the variance. However, the specific traits that drove this prediction were surprising.

The study's dataset: 1 behavioral DV (PCA scores) and 15 personality IVs (Big 5, Curiosity dimensions, and BIS/BAS) across 820 participants.
  • The "Core" Curiosity Gap: Traditional "core" curiosity traits, such as Openness to Experience, Joyous Exploration, and Deprivation Sensitivity, were notably poor predictors of task behavior.
  • Directed Exploration Predictors: The strongest positive predictors of directed (uncertainty-driven) search were Need for Cognition (the preference for effortful mental activity), Stress Tolerance (the ability to manage the anxiety of the unknown), and Behavioral Inhibition (sensitivity to errors).
  • Random Exploration Predictors: Conversely, higher Extraversion and Thrill Seeking were the most robust predictors of random exploration. This suggests that for some individuals, "curiosity" is less about systematic puzzle-solving and more about the excitement of unpredictable outcomes.

Discussion and Future Directions

The findings highlight a significant disconnect between how we describe ourselves on personality scales and how we behave in abstract cognitive tasks. One major limitation is the abstractness of laboratory stimuli (e.g., colored marbles or abstract probabilities), which may fail to engage the "rich personal knowledge" (like a passion for history) that self-report scales often tap into.

Future research should focus on developing "naturalistic" cognitive tasks that involve real-world knowledge or trivia. Furthermore, the current battery did not fully capture "valence-driven" curiosity—seeking information to manage emotional states like dread or pleasure—marking a clear path for future experimental design.

Summary updated

Abstract

To understand human learning and progress, it is crucial to understand curiosity. But how consistent is curiosity’s conception and assessment across scientific research disciplines? We present the results of a large collaborative project assessing the correspondence between curiosity measures in personality psychology and cognitive science. 820 participants completed 15 personality trait measures and 9 cognitive tasks that tested multiple aspects of information demand. We show that shared variance across the cognitive tasks was captured by a dimension reflecting directed (uncertainty-driven) versus random (stochasticity-driven) exploration and individual differences along this axis were significantly and consistently predicted by personality traits. However, the personality metrics that best predicted information demand were not the central curiosity traits of openness to experience, deprivation sensitivity, and joyous exploration, but instead included more peripheral curiosity traits (need for cognition, thrill seeking, and stress tolerance) and measures not traditionally associated with curiosity (extraversion and behavioral inhibition). The results suggest that the umbrella term “curiosity” reflects a constellation of cognitive and emotional processes, only some of which are shared between personality measures and cognitive tasks. The results reflect the distinct methods that are used in these fields, indicating a need for caution in comparing results across fields and for future interdisciplinary collaborations to strengthen our emerging understanding of curiosity.
BibTeX
@misc{jach2023individual,
  author = {Hayley Jach and Roshan Cools and Alec Frisvold and Michael A. Grubb and Catherine A. Hartley and Jochen Hartmann and Laura Hunter and Ruonan Jia and Floris P. de Lange and Ruby Larisch and Rosa Lavelle-Hill and Ifat Levy and Yutong Li and Lieke van Lieshout and Kate Nussenbaum and Silvio Ravaioli and Siyu Wang and Robert C. Wilson and Michael Woodford and Kou Murayama and Jacqueline Gottlieb},
  title = {{Individual differences in information demand have a low dimensional structure predicted by some curiosity traits}},
  year = {2023},
  doi = {10.31234/osf.io/aj3rp},
  howpublished = {PsyArXiv},
}
Figure