Bayesian Inference on Visual Word Recognition

In a study investigating cognitive processing, the research demonstrated through data visualization and prior predictive checks that recognition times for non-words are significantly longer than for words. To estimate the underlying parameters, advanced Bayesian methods were employed, including the Metropolis-Hastings algorithm within an MCMC framework. Results were robust, achieving an impressive 99.85% accuracy in posterior estimation using a combination of Grid Estimation and MCMC techniques, validated against analytically-derived true posterior distributions. The precision of parameter estimation was evidenced by narrow 95% credible intervals: α in [418.18, 420.63] and β in [49.79, 53.84]. These findings underscore the methodological rigor and the detailed insights gained into cognitive processes through rigorous statistical approaches.