This project is a basic fundamental research project, in the field of cognitive sciences. We will conduct a series of experimental studies that may be regrouped into two sets. In the first set, we will address the question of how confidence affects the selection and the realization of a given task, for an agent performing a task in isolation. For instance, we will test whether confidence sets the priorities between different tasks when the agent needs to organize and define which task to do first in a sequence of tasks. In the second set of studies, we will investigate how confidence affects how the agent interacts with the environment and with other agents. For instance, we will quantify the contribution of overconfidence to sub-optimal adjustment to external signals and to sub-optimal interactions with other agents in competitive and collaborative situations.
Our first results show that metacognition positively helps agents to learn in non-supervised learning situation.
In the long run, we aim at developing new knowledge about metacognition and decision-making, and at fostering the emerging dialogue between economics and psychology, two domains that have a shared interest in the basic understanding of human behaviour.
Hainguerlot, M., Vergnaud, J.C., de Gardelle, V. (2018).
Metacognitive ability predicts learning cue-stimulus associations in the absence of external feedback.
Scientific Reports, 8:5602, DOI:10.1038/s41598-018-23936-9
For individuals navigating in a hazardous environment, for companies competing in complex markets, good decisions are essential to survival. Assessing one’s own decisions, an ability called metacognition, might help improving these decisions. As such, metacognition is becoming an important topic in cognitive sciences and neurosciences of decision-making. Our proposal is aimed at better understanding the influence of this metacognitive ability on our behaviour.
Basic studies in cognitive sciences and neurosciences have aimed at quantifying the ability of agents to make good metacognitive judgments, or at determining the neural mechanisms underlying such judgments. Critically however, in most current approaches, metacognitive judgments seem to be commentaries on behaviour, without real impact. Here, we want to clarify the ways in which metacognition might have a direct impact on the behaviour of an individual, using an experimental and computational approach.
We will consider several potential links between metacognition and behaviour, and we will conduct series of behavioural experiments to test these links empirically. Our experimental studies are based on visual psychophysics, a discipline which currently serves as a ground to develop neurobiological and computational models of decision-making. We will complement our empirical investigations with eyetracking recordings, focusing on particular on pupil size which has been linked to confidence in recent work. We will adopt in parallel a modelling approach to better motivate and interpret our empirical investigations.
Our project involves two lines of research. On the one hand, we will investigate the influence of metacognition on the behaviour of an agent performing a task in isolation. For instance, we will test the hypotheses that confidence might act as a value signal during task selection, and as a priority signal during task planning. On the other hand, we will investigate the impact of metacognitive abilities and judgments on the behaviour of an agent who receives information from the environment and from other agents. For instance, we will assess how overconfidence (which is a major issue for instance in behavioural finance) impacts on behavioural adjustment to external information. We will study the influence of metacognition in situations in which agents cooperate or compete with each other.
In sum, the proposed project takes a basic science approach, in the field of behavioural sciences. We wish to study the ability of human observers to assess their own cognitive processes, an ability called metacognition. We will broaden the spectrum of research questions in this domain, in order to provide new insights about metacognition and how it impacts on behaviour. We anticipate that acknowledging the instrumental role of confidence will improve our models of decision-making, and will enrich the emerging dialogue between behavioural psychologists and economists.
Monsieur Vincent DE GARDELLE (Centre d'économie de la Sorbonne (UMR8174 UP1/CNRS))
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
CES-UMR8174 (CNRS DR1) Centre d'économie de la Sorbonne (UMR8174 UP1/CNRS)
Help of the ANR 172,800 euros
Beginning and duration of the scientific project: - 36 Months