Rigorous exploration in a model-centric science via epistemic iteration

Abstract

Urgent attention is needed to address generalizability problems in psychology. However, the current dominant paradigm, which centers dichotomous results and rapid discoveries, may not provide the solution. Instead, we propose a paradigm shift towards a model-centric approach, which can aid in understanding the sources of generalizability and promote systematic exploration. In a model-centric paradigm, scientific activity involves iteratively building and refining theoretical, empirical, and statistical models that communicate with each other. This approach is organic, transparent, and efficient in addressing generalizability issues. We illustrate the nature of scientific activity in a model-centric system and its potential for advancing the field of psychology.

Mapping visual working memory models to a theoretical framework

“…An underscored precursor to the reproducibility crisis in the psychological sciences has been the lack of direct connection from experiments and tested hypotheses to the underlying theory (Guest & Martin, 2021;Klein, 2014;Meehl, 1978;Oberauer & Lewandowsky, 2019;Scheel et al, 2021). That is, an over-reliance on the hypothetico-deductive method and null-hypothesis significance testing without substantial groundwork to construct collective theories has hindered progress in the psychological sciences (Borsboom et al, 2021;Devezer & Buzbas, 2023;Meehl, 1978).…”

“…In my view, by and large, more progress could be achieved by rethinking our theoretical framework and broadly adopting a model-based approach (Devezer & Buzbas, 2023) so as to specify and examine how VWM actually operates and gives rise to its sharp capacity limit and other extant phenomena. A fixation on understanding the exact representative unit of VWM to characterize its capacity limit, a remnant of the slots versus resources comparisons, remains pervasive in shaping current research approaches.…”

“…I hope the map provides a common starting point for discussions across our field, getting researchers who focus on disparate but related phenomena, or use entirely different methods and approaches, on the same page. This may be a catalyst for our field to take a model-oriented approach to empirical research that is grounded in theory, clearly specifying the connection between phenomena and the mechanisms of tested models (Borsboom et al, 2021;Devezer & Buzbas, 2023;Oberauer & Lewandowsky, 2019). From these discussions, one could imagine that the VWM field identifies that it is not yet ready to conduct certain empirical tests -that innovation of measures and methods, or formalization or computational implementation of models are needed -perhaps spurring collaboration across labs on common goals.…”