Perceptual decision making: drift-diffusion model is equivalent to a Bayesian model | JoVE Visualize

Area of Science:

Cognitive Neuroscience
Computational Neuroscience
Decision Science

Background:

Perceptual decision-making experiments commonly use the drift-diffusion model to analyze behavioral data, explaining accuracy and reaction times by accumulating evidence.
Bayesian models have emerged as a framework for understanding how the brain processes noisy information in perceptual tasks.
A known but unexplicated link exists between drift-diffusion and Bayesian models in explaining information extraction.

Purpose of the Study:

To explicitly derive the mathematical relationship between the drift-diffusion model and functional Bayesian models.
To demonstrate the practical utility of this derived equivalence for analyzing multi-subject behavioral data.
To explore novel decision variables suggested by the Bayesian model and their potential neural correlates.

Main Methods:

Utilized a Bayesian model framework to derive equations connecting parameter values between drift-diffusion and Bayesian models.
Applied the derived equivalence to fitting behavioral data from multi-subject perceptual decision-making experiments.
Investigated alternative decision variables within the Bayesian model that yield equivalent behavioral predictions.

Main Results:

Successfully derived explicit equations that precisely relate the parameters of the drift-diffusion model and Bayesian models.
Demonstrated that this derived equivalence offers practical advantages for fitting and analyzing data from multiple subjects.
Identified distinct decision variables within the Bayesian framework that predict identical behavioral responses, offering avenues for neural discrimination.

Conclusions:

The explicit mathematical link between drift-diffusion and Bayesian models provides a unified framework for analyzing perceptual decision-making.
The Bayesian model offers extensions and insights, such as novel decision variables, that are challenging to derive within the drift-diffusion model alone.
These findings pave the way for designing new experiments to test hypotheses about evidence accumulation and neural mechanisms underlying decision-making.