An individualist viewpoint is arguably justified in the philosophy of metacognition, for classical reasons: it is mainly at the level of the individual organism that it makes sense to analyze mechanisms, feelings, and representational contents constituting epistemic sensitivity. As recognized in the conclusion of The Philosophy of Metacognition, however, metacognition is far from being merely a private affair. Asher Koriat and Chris Frith and his colleagues have started to explore the vast domain of social metacognition. They investigated, in particular, when two brains are better than one. Groups were found to enhance or impair their collective perceptual decisions by merely encouraging or discouraging the public expression by the participants of their own individual confidence level. Philosophically, this result tends to suggest that collective thinking, in contrast with individual thinking, is mainly or even exclusively propositional (this is defended in my “Consensus as an epistemic norm for group acceptance“). Second, it requires exploring what kind of propositional attitude a group generates as a result of its collective agency: believing? Accepting? Accepting under consensus? A strong support of the latter view consists in spelling out the specific conditions of correctness that need to be fulfilled to convert distributed information (a factive condition) into a propositional form. Consensual acceptance, then, can be shown to have a reliable potential for collectively acquiring knowledge, while alternative solutions are not tailored to secure epistemic reliability.
Another facet of social metacognition with a high relevance to philosophy of mind and to epistemology is only starting to be investigated (a collective Oxford University Press book is currently in preparation, Proust & Fortier eds.). It does not have to do with collective epistemic decision-making, but rather with the influence that a culture can respectively exert on individuals’ procedural and analytic metacognition. The evidence collected so far presents us with a puzzle. Procedural metacognition seems essentially similar across cultures. Analytic metacognition (more exactly “core intuitions about knowledge”) seems also to be invariant. How to reconcile a claim of invariant epistemic sensitivity to truth, and a claim of invariant judgments about truth, justification and knowledge, with the anthropological observation of extensive divergences on methods for reducing subjective uncertainty (oracles, astrology, runes, tarot, palm reading, mathematical modeling)? It’s worth drawing the balance sheet of the present evidence.
In favor of the universality in metacognition, evidence from developmental psychology of procedural metacognition is summarized in (A), from the experimental philosophy of analytic metacognition in (B). (C) presents evidence difficult to reconcile with both (A) and (B).
A) The development, across cultures, of procedural metacognition (i.e. feeling-based epistemic evaluation) has been investigated by several groups, including our Europe-based team, including cognitive scientists, anthropologists and philosophers. The children we tested (from Japan, Germany, France, Mayan population in Yucatan) turn out to be able to monitor the (perceptual or memorial) information they have or don’t have, and to form reliable epistemic decisions, several years before being able to provide reliable verbal reports about what they know. In our and in other studies, observed variations, then, seem to concern not the specific sensitivity to epistemic norms, but rather the type of epistemic actions that are valued. For example, in contrast to 6 year-old French children, Japanese children are more likely to endorse an incorrect, but consensual label than a correct label used by a single dissenter. In the US, children tend to raise many ‘why ‘questions. In Kenya, in contrast, children ask ‘where’ and ‘what’ questions rather than why. (Gauvain et al. 2013). To summarize: by and large, children have a similar sensitivity to what is true and consensual, but differ in how this sensitivity is expressed in their behavior. Informativeness and relevance are sensed everywhere, but communication may serve different purposes.
B) The cross-cultural variability of analytic metacognition (i.e., judgment and verbal report about what oneself or others know) has been explored in a sequence of studies in experimental philosophy about “Gettier intuitions”. Edmund Gettier published in 1963 a three-page article where he offered counter-examples to the classical conception of knowledge as justified true belief. You consult the city clock: it presently indicates the correct time, 2 pm. But it is only correct because it is 2pm (the clock is broken). Hence you may be justified to believe that P, where P is a true belief, yet not really know that P. Several philosophers (among them: Nagel et al. 2013, Stich and Machery, 2015, Turri, 2013, Weinberg et al., 2001) have used Gettier-style scenarios to elicit judgments of knowledge in people from different cultures (e.g., Japan, Brazil, US and India, in the Stich and Machery study), socio-economical status or educational background. To the question: “does the character described knows or merely believes (s)he knows that p”?, participants tend to respond in the same way, which is taken to indicate that they share their “intuitions” about the nature of knowledge.
C) These two types of invariance are prima facie difficult to reconcile with the anthropologists’ observation of the diversity of epistemic practices across cultures and social groups: Pythia ravings, astrology, divination, oracles, runes, tarot, etc. Is metacognition absent from irrational practices, or rather silenced? Let’s take the example of the Poison Oracle, a predictive practice in the Azande reported by anthropologist Evans – Pritchard (1937). The questioner proposes a prediction about a future state of affairs, (or about someone being a witch) and commands the oracle to kill the foul. The practitioner then forces a chicken to swallow the poison (an alcanoïd close to strychnine). A second corroborative test is used with the corresponding negative prediction, where the oracle is commanded to spare the foul. The prediction is supposedly correct if the chicken dies or survives as commanded. This method, clearly, instantiates a sensitivity to hypothesis-testing and a preference for observation over testimony, as emphasized by anthropologist Maurice Bloch.
But why do people consider the lethal outcome as caused by the response of the oracle rather than by poison alone? Why does it seem important for the assumed reliability of the poison oracle that the process should respect sexual taboos? These questions, when appropriately framed, point to a solution, which will be explored in my next post, where procedural metacognition plays a prominent role.
The painting on top of this post is “La sortie est à l’intérieur” (2005), by Jean-Michel Alberola