Prediction maps as a gateway to better understand and test scientific theories

Scientific theories aim to explain and predict phenomena, yet assessing how experimental evidence impacts a theory remains a significant challenge. This is particularly evident in the psychological and cognitive sciences, where a lack of formal models and precise quantitative predictions often obscures the "centrality" of a given hypothesis for the theory that generates it. Because auxiliary hypotheses are often omitted or vaguely related to core ideas of a theory, researchers can easily safeguard core theoretical tenets against falsified predictions, rendering theory testing largely ineffective. To address this problem, we introduce prediction maps: a novel methodological approach that represents scientific theories as networks of beliefs structured in a core-periphery way. They visualize theoretical claims and empirical predictions as nodes, with inferential relationships as edges. By applying graph-theory and network analysis metrics, we can quantify the centrality of specific predictions. This allows for a systematic evaluation of "evidential weight," identifying which experimental results would be the most meaningful tests for a theory. We demonstrate the utility of this approach within the science of consciousness – a field characterized by high theoretical fragmentation and underspecified models. We show how prediction maps can be used to design more informative experiments, compare competing frameworks, and assess empirical success. Finally, we present an open-access online platform designed to help researchers across disciplines structure their theories. Prediction maps offer a transparent, structural analysis of the link between theory and evidence, providing a rigorous tool for navigating the complex landscape of theory testing.