Definitions and first paradoxes

The vocabulary of metacognition

If cognition can be defined schematically as all the mental processes that enable us to process information (internal or external), then metacognition could be defined as all the knowledge and beliefs we hold about our own cognitive processes (past, present or future), as well as the processes that enable us to manipulate them. Meta-memory, for example, concerns all our knowledge and beliefs about our own memorization and retrieval processes. When a student, doubting his knowledge, decides to revise before an exam, he makes a metacognitive judgment about the weaknesses of his memory. Calculate 23 + 18. You can undoubtedly report the order and nature of the operations you perform: they are available at a metacognitive level. On the other hand, you have no introspection on how you carry out elementary calculations such as 2 + 1.

For each mental operation at level n, metacognition presupposes the existence of a mental representation at level n + 1 or " meta-representation " of the mental operations performed at the lower level. According to this hierarchical scheme, introspection (or cognitive monitoring in the English-speaking literature) is exercised from the bottom up: it consists in uncovering the meta-representation on the basis of access to information at the lower level (access which may be partial or illusory). In this way, we exercise introspection when we detect an error in our calculation. Conversely, metacognitive control is top-down: it consists in exploiting meta-representation to modify the strategy and mental operations performed at the lower level. This is the case, for example, when we become aware of our mistakes and decide to slow down and pay more attention on the next try.

A brief history

Until the end of the 19th century, introspection was considered the central method for studying the human mind. Many psychologists, scarcely aware of the extent of non-conscious operations, considered direct observation of mental facts possible. For Wilhelm Wundt (1832-1920), the very object of psychology is the study of subjective mental experience, and introspection is the only method. Franz Brentano (1838-1917) promoted a "descriptive psychology" or "phenomenology" (before Husserl), which consists in studying the phenomena of inner perception from a subjective, "first-person" point of view. Oswald Külpe (1862-1915), a pupil of Wundt and leader of the Würzburg school, developed methods for the verbal description of introspection, although he discovered a first limitation of introspection: during "thought without images", the subject cannot always report his percepts.

Edward Titchener (1827-1927) at Cornell, Edwin Boring (1886-1968), Théodule Ribot (1839-1916) and even Alfred Binet (1857-1911) in France held similar views: " introspection, it may be said, is the basis of psychology, it characterizes psychology in such a precise way that any study that is made by introspection deserves to be called psychological, and that any study that is made by any other method belongs to another science " (A. Binet, Introduction à la psychologie expérimentale, 1894).

However, this alleged specificity of psychology was immediately the subject of debate. Auguste Comte countered with an argument known today as Comte's paradox. According to Comte, " the human mind can directly observe all phenomena except its own. For, by whom would the observation be made ? [...] The thinking individual could not be divided into two, one of whom would reason, while the other would watch reasoning. The observed organ and the observing organ being, in this case, identical, how could the observation take place ? " (Auguste Comte, Cours de philosophie positive [1830-1842], Vol. 1, pp. 31-32).

A vigorous response to this criticism was provided by John Stuart Mill : " It might have occurred to M. Comte that it is possible to study a fact by means of memory, not at the very instant when we perceive it, but in the moment afterwards : and this is, in fact, the mode in which we generally acquire the best of our science concerning our intellectual acts. We reflect on what we have done when the act has passed, but when the impression of it is still fresh in our memory. [...] This simple fact destroys M. Comte's whole argument. " (John Stuart Mill, Auguste Comte and Positivism [1865], pp. 68-69). Contemporary cognitive neuroscience might add that there is nothing impossible about a brain circuit, located for example in the prefrontal cortex, receiving and regulating information from other hierarchically lower circuits.

Comte's paradox was therefore not a paradox at all, but suspicion arose that introspection was an inadequate scientific method. In a violent critique that has remained famous, behaviorist leader John Watson put it forcefully: "Psychology as the behaviourist sees it is a purely objective branch of the natural sciences. Its theoretical aim is the prediction and control of behavior. Introspection is not one of its essential methods, and the scientific value of its data does not depend on how well they lend themselves to interpretation in terms of consciousness." Out with introspection: the subjectivity of his observations would make any scientific construction impossible.

In retrospect, however, it seems to us that Watson's critique confuses introspection as a method, and introspection as an object of study. Introspection is certainly not an infallible method of gaining access to mental architecture: even a perfectly trained psychologist cannot faithfully report mental processes, otherwise experimental psychology would be a very easy task! However, the performance and limits of introspection are a major and perfectly legitimate subject of research, which we shall see leads to empirical results that can be reproduced from one individual to another.

As early as 1971, at the dawn of the cognitive revolution, John Flavell introduced the study of meta-memory. In 1979, he proposed an initial theorization of metacognition, distinguishing between knowledge (conscious or unconscious, right or wrong), experiences, goals, tasks, strategies and metacognitive actions. Influenced by Piaget, he already emphasized the importance of metacognition in children's education. Indeed, on the basis of their self-understanding, children are led to devise, rightly or wrongly, learning and memory search strategies that influence their performance.

The 1960s-1990s saw lively debate about the fidelity of introspection. For Nisbett and Wilson (1977), introspective judgments are very often fictitious and therefore useless. For Ericsson and Simon (1980), on the other hand, verbal reports are often adequate as long as the reported information is present in short-term memory. Ericsson and Simon introduce a classification of introspective tasks that distinguishes between the timing of the verbal report (immediate or delayed), and the type of report (direct, with recoding, or unrelated to initial experience). Their review of experimental data suggests that verbal report can be extremely faithful when it is direct and describes the present content of short-term memory. Under these conditions, there is a close correspondence between what subjects say and what they do: introspection is credible and useful.

Metacognition methods

Since the 1970s, the validity of introspection studies has been further strengthened by the advent of rigorous experimental paradigms for measuring metacognition. In the field of meta-memory, the judgment of learning asks participants, after a learning phase, to estimate how they will perform in a subsequent memory test. The "feeling of knowing" measure, on the other hand, requires the participant to estimate, immediately after failing to remember an item, whether he or she would be able to recognize it from among several others. In both cases, the introspective prediction can be compared with the objective reality measured a few minutes later.

More generally, second-order judgments involve estimating one's degree of confidence in a previous response (known as first-order), betting on the veracity of one's response (wagering), or detecting one's own errors. As Jérôme Sackur notes, psychophysics itself regularly calls upon introspection in the form of a subjective report, verbal or non-verbal, which can be carefully quantified, replicated and quantitatively compared with objective reality.