Neurofeedback is a promising technique for brain rehabilitation and cognitive enhancement; however, it suffers from an inefficacy problem, as more than a third of participants do not learn to regulate their own brain activity. Lack of task engagement, probably related to an inadequate difficulty level, has been proposed among the possible factors that can affect the learning process. Here, we explored the possibility of monitoring cognitive engagement using physiological signals through eye tracking and electrodermal activity. We recorded these signals while participants completed different tasks designed to stimulate cognitive load and attentional focus. From the assumption that high performance in any task requires an optimal level of engagement, we finally trained a linear model to predict participants' performance during different cognitive tasks by using their physiological signals. Results showed that pupil diameter strongly discriminates internal focus of attention from mind-wandering states, and that pupil diameter and skin conductance response are sensitive to differentiate between task and rest conditions. In addition, both features are sensitive to habituation and cognitive load effects. The model was able to predict performance when trained on a specific task, as well as when combining different tasks together. Our results provide new insights regarding the physiological bases of cognitive engagement and propose a model with the potential to monitor it, while predicting performance. The capacity of our model to be generalized across different tasks encourage further research to test its potential in other domains, for example, in the context of individualized neurofeedback protocols.
Fragueiro et al. (Fri,) studied this question.