Periodic Reporting for period 1 - MetaChange (A decision within a decision: Exploring the psychological, electrophysiological and computational mechanisms underlying changes of mind and metacognitive judgements.)
Reporting period: 2023-06-15 to 2025-06-14
Making and evaluating decisions is one of the most fundamental cognitive processes of humans and other animals. Indeed, we almost constantly make decisions in our everyday lives, with potentially dramatic consequences if we are not correctly able to quickly evaluate or alter a given decision. In this context, MetaChange's aim is to understand how humans are able to integrate sensory information to inform a decision, even after the decision has been made. This question arises from an increasing interest in our ability to evaluate a decision even in the absence of explicit feedback, as well as our ability to alter an ongoing decision based on additional internal or external new information.
Thus, the general objective of MetaChange is to understand how these metacognitive abilities are impacted by the availability of post-decisional information, that is information relevant to the task but available only after an initial decision has been made.
The specific objectives of MetaChange are first to understand the impact of post-decisional information at the behavioural level, second to harness invasive electroencephalography to unravel the underlying neuronal processes, and third to build a computational model of these decisional and metacognitive processes to provide a unifying framework to understand our behavioural and electrophysiological results. As such, it is meant to be the first full characterization of decisional and post-decisional processes combining these three approaches and will help us to better understand how the human brain explores the world.
Thus, the general objective of MetaChange is to understand how these metacognitive abilities are impacted by the availability of post-decisional information, that is information relevant to the task but available only after an initial decision has been made.
The specific objectives of MetaChange are first to understand the impact of post-decisional information at the behavioural level, second to harness invasive electroencephalography to unravel the underlying neuronal processes, and third to build a computational model of these decisional and metacognitive processes to provide a unifying framework to understand our behavioural and electrophysiological results. As such, it is meant to be the first full characterization of decisional and post-decisional processes combining these three approaches and will help us to better understand how the human brain explores the world.
To achieve Metachange’s objective several theoretical and technical developments were required. First, we needed to develop an experimental decision-making task which would allow us not only to precisely detect when participants made their choice but also allow us to manipulate the information available to them immediately after their decision as well as to measure how confident they were in the outcome. This was achieved by combining conventional psychophysical methods with real-time mouse tracking to detect decision onset with high precision. This novel behavioural paradigm was calibrated on 15 healthy participants and validated on another group of 15 healthy participants.
Second, we needed to adapt this novel task to the context of invasive electroencephalography (sEEG). Indeed, sEEG is performed in participants with pharmacologically intractable epilepsy, who are implanted with invasive electrodes throughout their cortex during the course of their medical treatment, which required further testing and development, as experiences were performed at their bedside in the hospital, an environment much less stable and controlled than the usual laboratory setting. Seven epileptic participants were recorded across two recording centers, and the data collection process is still ongoing. In order to analyse the ensuing data a novel preprocessing tool was developed, allowing us to convert the raw electrophysiological data recorded at the bedside of patients into a usable dataset following the gold standards of data curation within our field. This tool was designed to be versatile and has been made freely available to the scientific community.
Finally, several analytic tools were developed to begin to make sense of the neural activity recorded during our experiments, including tools based on previous research and novel tools based on machine learning techniques. Again, these tools were freely shared with the scientific community, in the spirit of transparency and open science.
Second, we needed to adapt this novel task to the context of invasive electroencephalography (sEEG). Indeed, sEEG is performed in participants with pharmacologically intractable epilepsy, who are implanted with invasive electrodes throughout their cortex during the course of their medical treatment, which required further testing and development, as experiences were performed at their bedside in the hospital, an environment much less stable and controlled than the usual laboratory setting. Seven epileptic participants were recorded across two recording centers, and the data collection process is still ongoing. In order to analyse the ensuing data a novel preprocessing tool was developed, allowing us to convert the raw electrophysiological data recorded at the bedside of patients into a usable dataset following the gold standards of data curation within our field. This tool was designed to be versatile and has been made freely available to the scientific community.
Finally, several analytic tools were developed to begin to make sense of the neural activity recorded during our experiments, including tools based on previous research and novel tools based on machine learning techniques. Again, these tools were freely shared with the scientific community, in the spirit of transparency and open science.
Considering that data collection and analysis is still ongoing, the results from MetaChange are still in their infancy. However, the analysis of the behavioural data has already started to reveal the underlying process behind the integration of post-decisional sensory information. We have for instance shown that this type of information is indeed integrated towards forming a feeling of confidence and the possibility of triggering a change of mind, an observation which is still a matter of debate in the decision-making community. We furthermore showed that the nature of the post-decisional information impacts our feeling of confidence, as well as our ability to change our minds, in a differential manner. For instance, we showed that providing post-decisional information supporting the initial decision tended to increase the feeling of confidence without triggering more changes of mind (even though changing their mind would have sometimes increased the participants' performance), whereas providing information contradicting the initial decision gave rise to a decreased feeling of confidence and an increased proportion of changes of mind. Furthermore, we showed that the directionality of the abovementioned effect of post-decisional information on confidence was reversed during change-of-mind trials, indicating an interconnection between the two processes, rooted in post-decisional integration of sensory information. Further analyses will be required before disclosing the results regarding neural data, but overall, these preliminary results strongly support the hypotheses on which MetaChange was built and provide novel insight into the temporality of the decision-making process.