*Desender, K., *Teuchies, M.*, Gonzalez Garcia, C., De Baene, W., Demanet, J., & Brass, M., (2021). Metacognitive awareness of difficulty in action selection: the role of the cingulo-opercular network. Journal of Cognitive Neuroscience. [pdf]

Why do some actions feel easy whereas with others are experienced as more difficult?

While there has been quite some behavioral work done on this topic (e.g. Desender et al. 2014; 2017; 2018; Questienne et al. 2018) so far it remains unclear which brain areas support the computation of subjective difficulty. Here, we used a simple masked priming paradigm with additional subjective difficulty ratings to answer this question. As shown in the Figure below, on each trial participants decided whether a target was pointing left or right, and a subliminally presented prime could either cause response conflict or be congruent with the target. After each trial, participants rated how much difficulty they experienced during action selection.

The novelty of the current study is that people were doing this while lying in the fMRI scanner. ROI analyses on two brain regions previously implicated in conflict processing (Teuchies et al. 2016), showed that activity in the rostro-cingulate zone and activity in the anterior insula were both positively related to subjective difficulty. Importantly, these associations remained even after controlling for response congruency and response speed.

A subsequent whole-brain analysis confirmed these findings:

Results of a mediation analyses were consistent with the hypothesis that the RCZ first detects the need for cognitive effort, and subsequently this signal is transferred to the anterior insula. In the discussion, we speculate that subjective difficulty might be inferred based on interoceptive awareness (for which the AI is known to be key). Although this hypothesis remains speculative at this point, we are keen to unravel the role of interoception in subjective difficulty by explicitly asking participants to engage in interoception while performing the same experiment as used here.

Finally, we believe that our findings will be of interest with regard to the question about domain generality of metacognition. Whereas a growing consensus in the field is that the computation of confidence is supported by a set of domain-general mechanisms , here we show that at a specific form of metacognition, namely subjective difficulty, is supported by a different set of brain regions then usually believed to be involved in the computation of confidence.