Informing cognitive abstractions through neuroimaging: The neural drift diffusion model.

Brandon M. Turner,Leendert van Maanen,Birte U. Forstmann

DOI: https://doi.org/10.1037/a0038894

IF: 8.2473

2015-04-01

Psychological Review

Abstract:Trial-to-trial fluctuations in an observer's state of mind have a direct influence on their behavior. However, characterizing an observer's state of mind is difficult to do with behavioral data alone, particularly on a single-trial basis. In this article, we extend a recently developed hierarchical Bayesian framework for integrating neurophysiological information into cognitive models. In so doing, we develop a novel extension of the well-studied drift diffusion model (DDM) that uses single-trial brain activity patterns to inform the behavioral model parameters. We first show through simulation how the model outperforms the traditional DDM in a prediction task with sparse data. We then fit the model to experimental data consisting of a speed-accuracy manipulation on a random dot motion task. We use our cognitive modeling approach to show how prestimulus brain activity can be used to simultaneously predict response accuracy and response time. We use our model to provide an explanation for how activity in a brain region affects the dynamics of the underlying decision process through mechanisms assumed by the model. Finally, we show that our model performs better than the traditional DDM through a cross-validation test. By combining accuracy, response time, and the blood oxygen level-dependent response into a unified model, the link between cognitive abstraction and neuroimaging can be better understood.

psychology, multidisciplinary