How to interpret attentional blink findings? A practical MCMC tool to assess the attentional blink with the eSTST model
Previous research has shown that Attentional blink (AB) data can differ between tasks, or subjects and it can be challenging to interpret these differences. In this paper, we provided a ready-to-use tool that allows researchers to map their data onto the episodic Simultaneous Type, Serial Token (eSTST) model. This tool uses the Markov Chain Monte Carlo algorithm to find the best set of 3 model parameters to simulate a given AB pattern. These 3 parameters have cognitive interpretations, such that differences in these parameters between different paradigms can be used for inferences about the timing of attentional deployment or the encoding of memory. Additionally, our tool allows for a combination of quantitative fitting against the overall pattern of data points, and qualitative fitting for theoretically important features. We demonstrate the algorithm using several data sets, showing that it can find cognitively interpretable parameter sets for some of them, but fails to find a good fit for one data set. This indicates an explanatory boundary of the eSTST model. Finally, we provide a feature to avoid overfitting of individual data points with high uncertainty, such as in the case of individual participant data.