The Index of Cognitive Activity Predicts Cognitive Processing Load in Linguistic Task

The Index of Cognitive Activity (ICA) has been shown to index cognitive processing load in language processing tasks in previous work.

However, the effect has so far only been shown as an aggregate effect across many subjects and trials. The feasibility and reliability of using ICA as a predictor of language-related cognitive processing load in a single-trial setting have not yet been assessed. Therefore, in this study, we compare the single-trial performance of various classification models, including Support Vector Machine (SVM), k-Nearest Neighbors (k-NN), Random Forest, and Gradient Boosting on two sentence-reading experiments. These algorithms were trained on ICA values to predict semantic and syntactic violations of the sentences read by the participants. The results showed that all trained classifiers performed above the 50% chance level. Of these classifiers, Gradient Boosting showed the best performance with an accuracy of 74.48% for semantic violation detection and 71.61% for syntactic violation detection, respectively. Our results indicate that the ICA is a viable measure for detecting cognitive processing load caused by language violation on a single-trial bas