Shiffrin et al. (Topics in Cognitive Science. 2025) argue that a systematic gap between predictive success and genuine causal understanding pervades scientific practice. This commentary contends that their framing preserves rather than dissolves the confusion it diagnoses; two inherited confusions do the damage. The first is a commitment to scientific realism: the picture of explanation as a window onto a mind-independent reality, which generates intractable pseudo-problems about measurement and understanding. The second confusion is crypto-Cartesian: the assumption that human action is driven by inner causal forces that psychology must identify and measure. This is not a scientific problem but a grammatical one — reasons and causes belong to different logical categories, and the search for mechanisms of will is incoherent from the start. These two confusions sustain the picture of scientific models as dirty windows onto hidden causal structures, rather than as provisional maps of relations between observable phenomena.
