Linking Memory Models and Decision Models: Insights on Frequency and Speed/Accuracy Trade-off

Many models used to explain accuracy and response time in recognition memory have separated retrieval from decision: Retrieval produces a value of memory strength that drives a decision process characterized by evidence that does not change until a decision is reached. Cox and Shiffrin (2012) have used an alternative dynamic approach that assumes these stages interact: As time passes after presentation of a test item, more information joins a probe of memory, changing the response from memory over time; the evidence that drives a decision thus changes from moment to moment. This model is consistent with many findings in recognition memory and has been used to explain puzzling “fluency” effects (Cox, Lewis, & Shiffrin, 2013). Here, we apply the model to two large-scale studies of accuracy and response time in recognition memory, showing that the model performs comparably with existing separate-stage models while affording richer conceptual interpretations of findings concerning word frequency effects and speed/accuracy trade-offs.