Metacognition

International audienceIt has often been claimed that metacognition should be defined as 'cognition about one'sown cognition', 'knowledge about one's own knowledge', or 'thinking about one's ownthinking' (Carruthers, 2011, Nelson & Narens, 1992, Perner, 2012). These formulations,however, are now often seen as unduly restricting the scope of metacognition to a form ofreflective judgment. There is evidence that agents with no concept of perception orknowledge, such as monkeys and young children, are nevertheless able to assess when theycan confidently engage in a task (such as finding an object, discriminating visual patterns orrecognizing whether an item was already presented) (Hampton, 2009). On an alternativedefinition, then, metacognition is the ability to evaluate whether one is likely to achieve aspecific cognitive goal or to have successfully achieved it. These evaluations jointlycontribute to the control of one's actions. They allow agents to select contextually efficientcognitive actions, such as trying or not trying to remember the location of an object , and todecide whether or not to rely on a specific cognitive output to act on the world. In both cases,on this broader definition, agents may rely either on the degree of confidence that theyexperience (a noetic feeling), or on what they judge to be the case, given their backgroundbeliefs and concept-based predictions (Arango-Muñoz 2011, Koriat & Levy-Sadot 1999,Proust 2007, 2013, 2015). Metacognition so conceived is distributed in domain-specificabilities, involved in the regulation of mental actions, the distribution of resources betweenrival processes, and the regulation of one's own emotions. Inter-agent metacognitive control isalso exemplified in the pragmatic principles of conversation first explored by Paul Grice, andin collective forms of epistemic investigation and decision-making (such as in scientificresearch)