We like to thank Dries Trippas for providing his materials. We also thank André Aßfalg for technical support with the online data collection.Materials, participant data, and analysis scripts with complete output for all reported experiments can be found in the Open Science Framework archive https://osf.io/9avjc/.

Signal Detection Theory (SDT) plays a central role in the characterization of human judgments in a wide range of domains, most prominently in recognition memory. But despite its success, many of its fundamental assumptions are often misunderstood, especially when its comes to its testability. The present work examines five main assumptions that are characteristic of existing SDT models -- the existence of a random scale representation, independent sampling, monotonic likelihood, Receiver Operating Characteristic (ROC) asymmetry, and a continuous (non-threshold) representation. In each case, we derive a testable consequence and test it against data collected in the appropriately designed recognition memory experiment. We also discuss the connection between yes-no, forced-choice, and ranking judgments. This connection introduces additional behavioral constraints and yields an alternative method of reconstructing yes-no ROC functions. Overall, the reported results provide a strong empirical foundation for SDT modeling in recognition memory.

Ongoing discussions on the nature of storage in visual working memory have mostlyfocused on two theoretical accounts: On one hand we have a discrete-state accountpostulating that information in working memory is supported with high fidelity for alimited number of discrete items by a given number of “slots”, with no informationbeing retained beyond these. In contrast with this all-or-nothing view, we have acontinuous account arguing that information can be degraded in a continuous manner, reflecting the amount of resources dedicated to each item. It turns out that the core tenets of this discrete-state account constrain the way individuals can express confidence in their judgments, excluding the possibility of biased confidence judgments. Importantly, these biased judgments are expected when assuming a continuous degradation of information. We report two studies showing that biased confidence judgments can be reliably observed, a finding that rejects a large number of discrete-state models, dismissing the idea that change-detection judgments consist of a mixture of guesses and high-fidelity memory representations.