Metacognition: Central Philosophical issues

Why is metacognition central for contemporary philosophy? A first reason is that it is an essential ingredient in cognitive actions, i.e. actions that one performs in order to decide whether it’s worth trying to remember, solve a problem, and whether the result you obtained is valid or not. Metacognition offers epistemic guidance each time your cognitive system is slow in providing you with the information you need to complete an action, or when there is a risk of being wrong. So-called “noetic feelings”, such as the feeling of cognitive effort, the feeling of familiarity, the feeling of knowing, the feeling of being right (or wrong), are  keys to prompting an usually reliable epistemic decision-making.

Second, metacognition is a key to understanding the evolution of cognition. Currently, there is  evidence that the dispositions to evaluate one’s own perception and one’s own memory are present in rodents and in rhesus monkeys as well as in humans. Metacognition, then, should include not only the function of reading one’s own mind, “thinking about one’s own thinking”, but also the function of controlling and monitoring one’s own cognitive actions.

Third, the existence and occasional interplay between two forms of metacognition, one based on noetic feelings, the other based on concepts related to one’s own mental activity (such as “perception”, “memory”, “reasoning”), offer new angles on the relations between linguistic and experience-based representations, and between conceptual and non-conceptual content. It also suggests ways of revisiting such central issues as self-knowledge and its relation to knowledge of others.

A final motivation for studying metacognition is to clarify the relations between world-directed and mind-directed rationality, and between instrumental and epistemic agency. Do you monitor and control your own mind as you monitor and control the world? Although trying to remember a proper name depends on the perceived utility of having this name available, sensitivity to an epistemic requirement of correctness is essential for the action of remembering to be performed at all. How does sensitivity to instrumental norms combines with sensitivity to epistemic norms? Experimental evidence on animal metacognition suggests that our brain has two different systems for evaluating confidence and evaluating reward and risk. It can be speculated that a feeling-based representational system is able to integrate these two forms of evaluation in order to flexibly guide decision. This in turn suggests new ways of addressing the difficult issue of the nature and function of consciousness.



The images featured on the various posts of this week are works by Jean-Michel Alberola.  La conscience claire (2016), Celui qui stratège (2001-2002),  Le roi de rien IV (2002-2003), La sortie est à l’intérieur (2005). (images on the next posts will be by Ulysses Belz).


  1. I’m very glad to see that this symposium about metacognition is taking place in this blog. Metacognition is a rather neglected mental capacity in the philosophical literature. Probably, this is the case because many philosophers either think that metacognition is the same capacity as mindreading, or they don’t see the importance of noetic or metacognitive feelings (some philosophers may think that metacognitive feelings are just weird experiences without any cognitive function).

    Now, I think that the main contribution of Joëlle Proust to the contemporary philosophy is to attract the philosophers’ attention towards this capacity. I hope that this post and the followings will show to philosophers the importance of metacognition. Some difficult questions that need to be addressed are the following: How does metacognition work, if not by means of metarepresentation of mental states? How could metacognition take into account instrumental as well as epistemic norms? What’s the exact relationship between metacognition and consciousness? Among others.

    P.D. I was very glad to see that Andy Clark acknowledged the importance of metacognition in a recent paper (“what the extended me knows”, 2015). This shows that philosophers are starting to see the importance of this mental capacity.

  2. Joelle Proust


    Thanks for your post, Santiago. I will address the first question you raised, which is indeed very important. For, in this particular case, the question of the mechanisms should be of major interest not only to psychologists and neuroscientists, but also to philosophers. Not only those philosophers who are interested in how to relate mechanisms with contents, but more specifically, those who want to understand what are the informational sources of our critical thinking. I will develop this point in a new post.

  3. Tobias Störzinger

    Thanks very much to the Brains Blog for the opportunity to discuss the important topic of metacognition in general and to talk to Joëlle Proust about her book! I would like to ask two questions, one related to the metacognition debate more generally and one specifically about Proust’s book:

    1) I am very interested in the interrelation between analytical-concept-based and procedural-feeling based forms of metacognition. Especially with regard to the question of which role procedural metacognition could play in analytical evaluation. As far as I know these two ‘kinds’ of metacognition are mostly seen as separate cognitive processes, which interrelate only in so far as system-1 procedural metacognition can be an input for system-2 analytical metacognitive self-ascription. But:
    If analytical metacognition is not restricted to self-ascription of mental states but should instead also be applied to evaluation of mental states (Not just self-mind-reading: “I believe, that I believe that p” but self-evaluation: “Is my belief that p correct?”) then maybe procedural metacognitive aspects could play an important role in analytical self-evaluation, even if this kind of evaluation is not feeling based? Could procedural metacognition be something like a pragmatic “context-framing-mechanism” which tells us when the analytical-concept-based reasoning process is pragmatically ‘enough’: Most people do not – and, as far as I can see, should not – try to prove that the world exists before they go on to reason about questions like whether all mushrooms are edible. There seems to be a pragmatic framing of a lot of explicit and concept based reasoning processes which itself is not explicit.

    2) How to think about a model of analytical evaluation in general?
    Prof. Proust: In the book you identify a “prima facie” difficulty with ‘attributivists’ accounts of metacognition if one tries to transfer the picture of other-evaluation to self-evaluation. Knowing that you believe that p is not sufficient for a evaluation of p, it rather “merely “[…]end up echoing the first-order evaluation” (Proust, The Philosophy of Metacognition, 2013: 47). This prima facie difficulty is due to the lack of the possibility to ‘exploit’ the ‘other-world-simulation’ and compare it with the real world. What do you think about the possibility of analytical-concept based evaluation? Is there a way out of the ‘echoing problem’ and (here again) what role can procedural metacognition play for such a solution?

    • Joelle Proust

      Thank you, Tobias, for your two important questions. You are right; the mere hypothesis of two systems, one devoted to procedural metacognition, the other to analytic metacognition, falls short of explaining how they interact. If one contrasts them under the opposition between evaluation and declarative knowledge, it is unclear how they can ever be used jointly. A common mistake that your question correctly signals is that, assuming the system 1/system 2 division in metacognition, a bottom-up constraint is supposed to activate procedural metacognition for low stakes, to be fully replaced by a more strategic form of control in case of error or when higher stakes are expected. Such a bottom-up hierarchy, if applied literally, makes little functional sense. I agree with you that the pragmatic constraints that bear on individual and to collective reasoning (i.e., conversation) are monitored by procedural metacognition (See my recent Mind_&_Lg article on this topic). There is a pending question that you point to: granting that reasoning strategies in system 2 need themselves to be selected and evaluated, what is the system able to do this? Evans postulated that it was a System 3. I prefer to build on an idea of Rolf Reber: “critical feelings” can be strategically developed in individuals through adequate training. Hence, system 1 is extensively used to deal with the strategic epistemic goals that system 2 is able to develop. This seems to be the function of higher education.

      This suggestion may be way of starting to address your second question. There is more to it, however, than mere pragmatic framing (thank you for focusing so clearly on the most difficult part of the issue). You ask whether an analytical (concept-based) form of metacognition is able to exploit by simulation a possible context of evaluation (for example based on felt certainty) and compare the cognitive (simulated) outcome with another context of evaluation (for example, an outsider’s critical view). How is system 2 able to achieve a meta-evaluation, if all it has is declarative knowledge?
      It seems clear that analytical metacognition, i.e. strategic conceptual thinking, when suitably trained, can perform these shifts and exploit not only each simulation, but select which one to pursue further, as is done, for example, in philosophical “disputationes”. It is plausible that even in this case, the ability to shift modes of evaluation, and compare their respective epistemic value, depends on the compilation of learnt strategies into epistemic affordances. The latter can be evaluated for their valence and their predicted value (as discussed in “the representational structure of feelings.”) Of course there is still a lot that remains to be studied by the psychology and the neuroscience of reasoning. My hunch is that even in system 2, evaluation relies on system 1 processing, appropriately recruited.
      Thanks again, Tobias, for these two stimulating questions.

Comments are closed.

Back to Top
%d bloggers like this: