Evidence Accumulation Models

The model accounts for the relationship between task difficulty, choice, response time, and confidence judgments in perceptual decision tasks. It assumes a DDM based decision process with post-decisional accumulation time. In addition, it includes a second accumulation process (the visibility process) that accrues evidence about the task difficulty, which evolves in parallel and independent of the decision process. Confidence is based on the final amount of evidence from the decision process, the visibility process, and the total accumulation time.

Psychological research often relies on mathematical models to explain and predict human behavior. Such models aim to formalize cognitive processes by mapping latent psychological constructs to model parameters and specifying how these generate manifest data. In this tutorial, we go through the steps of a principled Bayesian workflow that is imperative when developing and applying cognitive models. This workflow includes the following steps: (I) Prior pushforward and prior predictive checks to assess whether the model is consistent with our domain expertise; (II) Computational faithfulness checks to ensure that our estimation method can accurately approximate the posterior distributions; (III) Model sensitivity to examine if our inferences provide sufficient information for answering our research question; (IV) Posterior retrodictive checks to assess whether our model can capture the relevant structure of the true data-generating process.\r\nTo demonstrate how such a workflow is performed in an amortized manner using BayesFlow, we will take a complex model from the evidence accumulaton model (EAM) family which is intractable for standard Bayesian methods.

An extension on the Drift-diffusion model (Ratcliff, 1978) which addresses some of the more implausible assumption in more ecological contexts. Through the IRT-framework certain person- and item-parameters are added and allowed to vary to enable rigid assumptions of the original model to loosen up.

The drift-diffusion model (DDM) is a model of sequential sampling with diffusion signals, where the decision maker accumulates evidence until the process hits either an upper or lower stopping boundary and then stops and chooses the alternative that corresponds to that boundary.

An extension of the Drift-diffusion model (Ratcliff, 1978) in which eye-tracking is used to track fixation on either one of the presented options and used as an additional parameter in the decision-making process.