Debates in decision making, such as the debate about the empirical validity of the priority heuristic, a model of risky choice, are sometimes difficult to resolve, because hypotheses about decision processes are either formulated qualitatively or not precisely enough. This lack of precision often leaves empirical tests with response times and other detailed behavioral data inconclusive. One way to increase the precision of decision models is to implement them in broad cognitive frameworks such as the cognitive architecture ACT-R. ACT-R can be used to construct detailed process models of how people make, for example, risky choices, and to derive process predictions about, among others, eye movements, absolute response times, or brain activation. These precise process models make explicit their underlying assumptions, which facilitate direct model comparisons and make the models amenable to strict empirical tests. We demonstrate the level of detail that ACT-R provides with an ACT-R implementation of the inferential heuristic take-the-best. We end by addressing the question of why cognitive architectures are still not widespread in judgment and decision making.