How Computational Modeling Can Force Theory Building in Psychological Science

Psychology is a broad field that endeavors to develop explanatory theories of human capacities
and behaviors based on a wide variety of methodologies and dependent measures. Here
we argue that whether or not researchers choose to employ modeling (viz., choose to create
computational models of their theories over and above their data during the scientific inference
process) is one of the most important and divisive factors in our field. Modeling is underdiscussed
and underemployed, yet, in our view, holds integrative promise for advancing the
goals of psychological science. The inherent demands of computational modeling oer invaluable
momentum towards a better, and more open, psychological science. These demands force
the scientist to conceptually analyze, specify, and ideally, formalise intuitions and ideas which
would otherwise remain implicit or unexamined — something we propose should be called
“open theory”. Constraining our inference process through specification and modeling is what
will enable us as a field to meaningfully interpret data, and to build theories that explain and
predict. In this piece, we present scientific inference in psychology as a path function, where
each step shapes the next. Computational modeling can constrain the steps in the path, and has
the potential to advance scientific inference over and above the stewardship of the experimental
practice (e.g., preregistration, choosing frequentist or Bayesian statistics, power and sample
size, and other estimation variables). If as a field we continue to eschew, inadvertently avoid,
or remain ignorant of formal and computational modeling, we set ourselves up for a persistent
lack of replicability and, moreover, for failure at coherent theory-building that includes explanatory
force. We explain how the basic steps in the modeling process can be accomplished
and we touch on the cultural and practical issues that need to be faced therein, emphasizing
that the advantages of modeling can be achieved by anyone with benefit to all. The process of
computational modeling promotes transparent theorising; “open science” should include open
theory alongside, e.g., open data and open source code.