Cognitive models that assume that judgments are based on processes of sampling from memory have a long history in psychology and take a variety of forms, but the exact cognitive interpretations of them differ, are unclear, or remain elusive. Using the Precise/Not Precise (PNP) model (Sundh et al., 2021) we have revived an old approach to intuition and analyses, originally proposed by Egon Brunswik (1956). The model is based on the distinction between analytic algorithms that usually yield the same exact output and approximate intuitive algorithms that are rarely far off the mark but are inevitably perturbed by a random noise. The PNP model distinguishes intuitive and analytic processes depending on the error distributions around the model estimates. By combining the PNP model with specific cognitive algorithms, one can determine if analytic or intuitive cognitive processes implement the cognitive algorithms. In this chapter, we argue that also the memory sampling processes observed in multiple-cue judgments, characterized by good fit of the Generalized Context Model (Nosofsky, 2015), come in two different forms: one that involves analytic application of root-memorized individual exemplars and one that involves a noisy similarity-based inference about the likely criterion. We demonstrate that different parameterizations of the Generalized Context Model naturally imply response distributions that realize the distinction implied by the PNP model. With data from multiple-cue judgment, we show how the PNP model identifies, not only intuitive and analytic rule-based processes, but also processes of memory sampling with the empirical hallmarks of intuition and analysis.