The neurocognitive basis of confabulatory introspection : Choice blindness and the brain

Abstract: The goal of this thesis is to advance our understanding of introspection by studying when it fails, without us being aware of it. To do so, I have used the choice blindness paradigm. Choice blindness is a surprising phenomenon in which people fail to detect mismatches between their intention and outcome in a decision task, and then spontaneously confabulate reasons why they preferred an alternative they did not choose. Very little is known about the mechanisms of choice blindness, both when people detect or not, and how this leads to confabulation. My contribution consists in making this phenomenon less puzzling, by dissecting its neurocognitive basis. This involves building a first framework of false feedback detection in choice blindness, and in so doing investigating the monitoring mechanisms and reasoning processes that allows us to keep track of our intentions and their consequences in the world. In addition, by studying how our brain uses confabulation and post hoc rationalization to integrate false information about one's' choices, I want to highlight the deeply interpretative nature of our self-knowledge, a facet that has often escaped our intuitive understanding of ourselves. In the introduction of this thesis, I also review the state of the art on interpretative models of introspection, highlight gaps in the literature, and formulate new research tracks in light of my findings.In paper 1, I show that CB can arise without deception, as failures to detect false feedback persist even when participants are instructed to detect them. The study also shows the limits of our monitoring mechanism as well as how prior beliefs modulate false feedback acceptance. Building on these results, I outline a framework to understand choice blindness as the result of an interplay between automatic monitoring and reasoning systems. In paper 2, I show that the neural correlates of false feedback detection are consistent with the monitoring and inference framework described in paper 1. In the study, I find that detection is associated with reward monitoring (midbrain, basal ganglia, insula, ACC), sensory predictions (superior temporal sulcus, angular gyrus), as well as dorsal frontoparietal networks associated with executive control and reasoning. Paper 3 goes one step further, studying how outcome and motor levels of monitoring interact with each other. In the study, I show that motor monitoring can override outcome monitoring, leading people to reject outcomes they want when these are obtained through an action they perceive to be wrong. In paper 4, I investigate what happens at a neural level once monitoring failes, and people start constructing reasons for choices they never made. Here the finding is that confabulation involves the same theory of mind network that we use to make sense of others (mPFC, TPJ, STS) as well as areas related to reality monitoring (rPFC BA10) and executive function (dlPFC). This strengthens the interpretative perspective on introspection, suggesting that we use the same cognitive mechanisms to understand ourselves as those we use to understand others.