Abstract
Drawing inferences from data relies on satisfying the assumptions of analytical methods. Yet, the study of many psychological processes that involve change over time (e.g., learning) instead uses methods that assume a lack of change over time (e.g., within- “block” averages). Recent research has demonstrated a variety of theoretical and empirical benefits in aligning the assumptions about the generative process (e.g., changing perceptual sensitivity due to learning) with assumptions in analyses (e.g., changing estimate of sensitivity as a continuous function of time). In this review, we examine methods for estimating performance on a trial-to-trial basis as it changes due to learning. We then explore applications of these methods, including increased efficiency and statistical power, as well as the ability to more effectively investigate questions of learning generalization and individual differences. We highlight the applicability and utility of continuous-time models in perceptual learning, cognitive training, and beyond.