Choosing prediction over explanation in psychology: Lessons from machine learning

By Tal Yarkoni, J. Westfall. 2017

1,587 citations recorded by the source corpus.

Psychology has historically been concerned, first and foremost, with explaining the causal mechanisms that give rise to behavior. Randomized, tightly controlled experiments are enshrined as the gold standard of psychological research, and there are endless investigations of the various mediating and moderating variables that govern various behaviors. We argue that psychology’s near-total focus on explaining the causes of behavior has led much of the field to be populated by research programs that provide intricate theories of psychological mechanism but that have little (or unknown) ability to predict future behaviors with any appreciable accuracy. We propose that principles and techniques from the field of machine learning can help psychology become a more predictive science. We review some of the fundamental concepts and tools of machine learning and point out examples where these concepts have been used to conduct interesting and important psychological research that focuses on predictive research questions. We suggest that an increased focus on prediction, rather than explanation, can ultimately lead us to greater understanding of behavior.

Subfields: Cognitive Neuroscience, Experimental and Cognitive Psychology, General Decision Sciences

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

In Experimental and Cognitive Psychology

In General Decision Sciences

Citations in the corpus

Challenges and Future Directions of Big Data and Artificial Intelligence in Education (2020, background): “Hence, research on big data and AI is gaining increasing signiﬁcance in education (Johnson et al., 2011; Becker et al., 2017; Hwang et al., 2018) and psychology (Harlow and Oswald, 2016; Yarkoni and Westfall, 2017; Adjerid and Kelley, 2018; Cheung and Jak, 2018).”
Artificial intelligence and illusions of understanding in scientific research (2024, neutral): “…datasets yielded by big data approaches pose challenges to human capabilities that are not easily addressed by established statistical approaches, either in terms of the volume of data that need to be curated or the potential complexity of the models needed to explain and predict the data 48,49 .”
Moving Forward: Challenges and Directions for Psychopathological Network Theory and Methodology (2017, background): “A general problem with any exploratory model is that it is not necessarily the best model that will generalize to other datasets (Yarkoni & Westfall, 2017).”
Deep Neural Networks as Scientific Models. (2019, neutral): “Third, predictive power ultimately is a criterion for successful explanation: we cannot be satis ﬁ ed with a model that offers an explanation but also does not predict well [41,42].”
Cross‐validation failure: Small sample sizes lead to large error bars (2017, background): “However, too strong a focus on well-controlled explanation may limit the generality of the results (Yarkoni and Westfall, 2016).”
Community-level Research on Suicidality Prediction in a Secure Environment: Overview of the CLPsych 2021 Shared Task (2021, background): “In regard to the goals of prediction versus scientific explanation and understanding, it is worth noting the argument by Yarkoni and Westfall (2017) that psychology research as a whole, including research with explanatory goals, would benefit by taking a predictive approach.”
Ten simple rules for predictive modeling of individual differences in neuroimaging (2019, background): “Overfitting, or “the tendency for statistical models to mistakenly fit sample-specific noise as if it were signal” (Yarkoni and Westfall, 2017), yields models that fail to generalize and that have high performance variance.”
Successful Explanations Start with Accurate Descriptions: Questionnaire Items as Personality Markers for More Accurate Predictions (2018, methodology): “32: 186–201 (2018) DOI: 10.1002/per with commensurately low practical value for describing what is going on in the real world (Yarkoni & Westfall, 2017).”