Skip to main content

Divided Metacognition: when epistemic norms conflict

Final Report Summary - DIVIDNORM (Divided Metacognition: when epistemic norms conflict)

Do people from different cultures differ in their cognitive decision-making? Our findings allowed us to start addressing this new question as well as clarify underlying conceptual issues concerning the nature and scope of epistemic norms and practices. From a conceptual viewpoint, we offered a naturalistic account of epistemic norms based on the adaptive control of informational reliability - a metacognitive function inherent to many cognitive systems across phylogeny. We argued that dedicated feelings allow agents to monitor their cognitive activity, whether or not they have concepts for describing them, and described how they are used to represent cognitive opportunities and risks (e.g. uncertainty). We also showed that a temporal constraint allowed three trade-offs to emerge concerning the evaluation of an agent's cognitive states, i.e. impulsive, routine and strategic metacognitive control. This suggested to us that there should be both a universally shared affect-based metacognitive competence in humans, and a cross-cultural variability under the influence of local epistemic practices. Using psychological, neuroscientific and anthropological methods, we tested this hypothesis in human children and adults from different cultures.
We indeed found that feeling-based epistemic evaluation is present in very young children. We showed that European 3-4 year-olds are much more reliable in choosing to perform or to reject an epistemic task than in verbally reporting their uncertainty, which indicates that they can evaluate their cognitive output in a merely affective, procedural way. We compared epistemic evaluations in cultures by children and adults in either rural or industrial, European or non-Western backgrounds (Asian, Japanese, Malagasy, Yucatec Mayan, Peruvian Amazon). We found significant similarities between German and Japanese 5 year-old children in their affective sensitivity to error when informing others, and in their ways of solving the conflicts between norms of accuracy and consensus. Japanese children, however, were found to trust consensus on labelling more than accurate labelling at a later age than European children. Mayan children, however, had a comparatively reduced sensitivity to error. We found a high sensitivity to unexpected probabilistic sequences in the Shipibo children from Peruvian Amazon. These children tend to attribute unexpectedly easy performances to a magic intervention.
This might be a domain-general ability used across cultures by children and adults for detecting supernatural events. In French and Japanese adults, we found a biased propensity to feel more confident after an approving-like gaze. Our experimental study in Betsileo Malagasy adults suggests that the sensitive issue of discriminating "clean" (slave-free) from "unclean" ancestry is dealt with through an essentialist conception of lineage. This conception and the traditional rules presented as lineage-preserving tend to guide decision under a norm of accuracy. Little sensitivity to consensus was found except in response to unfamiliar scenarios (e.g. when a clean person has received a blood transfusion from an unclean person).
All our studies suggest that emotions play a universal role in guiding thinkers' own cognitive activity, but that their incidence in specific tasks or contexts is culture-specific. Specific cultural factors modulating such metacognitive diversity have started being identified, from childrearing and learning style (observational vs. instructional) to prevailing communicative practices and social norms such as deference to parents or to elders. Beyond their academic interest, these findings have a number of societal implications concerning the education of children at the age of globalization. Another subject of universal concern is the finding that information about the unreliability of a person agreeing with one's own judgment slightly diminishes, but does not cancel, the resulting gain in confidence. Consensus-based confidence replacing truth defines a "post-truth" attitude.
From a philosophical viewpoint, an important conclusion is that epistemology starts with affective guidance of mental action, rather than requires a conceptual understanding of one's own mind, and that a specific evaluative representational format should be acknowledged. The interplay of such cognitive affordance-sensings and beliefs throws a new light on human rationality.