Thesis Perrine PORTE

Metacognition reflects the ability to evaluate and control one's own mental states, and to form accurate confidence judgements about what one knows or perceives. In recent years, sophisticated theoretical and statistical tools have been developed to better understand how confidence arises from a decision, whether it involves a memory or perceptual task (Mazancieux et al., in press). We thus know why we are confident after a correct decision, and how we are able to detect our own errors. The vast majority of studies that have characterised confidence judgements are based on simple tasks involving the presentation of a visual stimulus about which participants have to make a two-alternative forced choice. For instance, a cloud of moving dots is presented and participants have to report whether the cloud moves to the right or left (Rahnev et al., 2020). Although adequate to outline the cognitive bases giving rise to the feeling of confidence, these oversimplified tasks remain far removed from the decisions one is led to make in ecological conditions. Notably, the vast majority of studies in meta-perception and meta-memory focus on the visual modality, even though we know that our percepts and memories are in essence multisensory (Faivre et al., 2017). A few recent studies have begun to characterize the links between multisensory integration and metacognition, including comparing metacognitive performance across the senses (Faivre et al., 2018), in unimodal and bimodal conditions (Arbuzova et al., 2020; Charles et al., 2020), and assessing our ability to form confidence judgments about McGurk-like audiovisual illusions (Kimmet et al., 2023; Meijer et al., 2023). It seems to emerge from this literature that confidence obeys supramodal rules, involving Bayesian integration processes (Deroy & Noppeney, 2016). However, the literature overlooks a critical aspect of decision making, which is our ability to form accurate confidence judgments about the presence or absence of a percept or memory (Mazor & Fleming, 2020). Thus, we do not yet know the rules that govern the processes used to assess the probability that a multisensory mental object is present in the sensory environment (e.g., a mosquito in a room) or in our memory (e.g., the memory of a concert). While it seems obvious that such a confidence judgment can be based on several sensory channels simultaneously (e.g. the image and the sound of the mosquito; the view of the stage and the music played), the precise rules involved and in particular their agreement with the Bayesian inference framework remain unknown to date. A better characterisation of these rules will allow us to better understand how agents evaluate their confidence in high ecological value situations. Furthermore, collecting confidence in the detection of multisensory mental objects will allow us to better characterize different types of false percepts (feeling like I hear a mosquito if I see one) or false memories (hearing a loved one's voice if I remember their face).
The thesis will therefore involve developing an experimental paradigm in which healthy volunteers will be asked to detect the presence of audiovisual stimuli in which each modality is presented at the detection threshold, and then report the confidence with which they believe they have detected each modality correctly. This paradigm will involve developing new two-dimensional confidence scales (one dimension for each sensory modality). A later version of the task will be developed in the memory condition, so that participants first encode a list of bimodal stimuli, and then report the confidence with which they think they can correctly recall each encoded dimension (Tatz et al., 2021 ; Duarte et al., 2022). In both cases, detection and confidence judgements will be analysed using generalized linear mixed-effects models. Stereo-electroencephalography will be recorded in a second phase via a collaboration with the epilepsy exploration unit at the University Hospital of Grenoble to evaluate the cortical correlates of multisensory confidence. We will thus evaluate the existence of neural correlates of consciousness and memory that generalise across the senses (Sanchez et al., 2020).

Supervisors

Direction de thèse : Nathan FAIVRE - nathan.faivreuniv-grenoble-alpes.fr (nathan[dot]faivre[at]univ-grenoble-alpes[dot]fr)

Co-encadrant : Louise GOUPIL - louise.goupiluniv-grenoble-alpes.fr (louise[dot]goupil[at]univ-grenoble-alpes[dot]fr)

Keywords : multisensory integration,metacognition,consciousness,