New JBer Chris Lepock, whose recent dissertation is entitled *Metacognition and Intellectual Virtue* (with Adam Morton his supervisor) now has available [http://www.ualberta.ca/~clepock/home.htm]a draft paper that develops many of its central themes:

"My dissertation analyzes the notion of an intellectual virtue, or (broadly speaking) a truth-conducive trait of cognitive character. In previous chapters, I endorse Greco’s (2000a, 2003a) account that a disposition to form beliefs is part of one’s cognitive character iff it is stable and well integrated with the rest of the cognitive system. However, this notion of cognitive integration is too vague to do the work it has to do. Below, I argue that we should understand it in terms of metacognitive control."

To elaborate, Chris says he works mostly "on metacognition - the monitoring and control of cognitive processes - and its implications for epistemology, especially virtue epistemology. In my dissertation (with the entirely unobvious title Metacognition and Intellectual Virtue) I argue that intellectual virtues should be understood as metacognitive control capacities. There are two broad lines of argument for this. First, what distinguishes an ordinary perceiver from somebody like a reliable clairvoyant appears to be metacognitive control... Second, high-level virtues like "conscientiousness" or "humility" can be analyzed as properties of one's metacognitive capacities...The next step is to try to understand better how metacognition is involved in the production of knowledge, and in particular, to what extent typical knowers are really capable of avoiding relying on untrustworthy processes in belief-formation, especially in the case of knowledge derived from testimony...I'm also working on epistemic value, trying to find a suitable framework of epistemic desiderata for virtue epistemology."

Chris, I read both your "Metacogntiion and Epistemic Virtue" and your "Levels of Reliabilist Appraisal: How to make the generality problem work for you." Both are very impressive, and I'm excited about the research program you have going on in VE.

"Metacognitive control is what makes a bundle of disconnected processes into a system for acquiring reliable beliefs." You show that there is a literature on this, and that its an empirically well-informed one. I want to get back to this, but let me start with a comment on your "Levels of Reliabilist Appraisal" draft paper. There you argue that "When we look at how epistemic appraisals are formed, we find indications that there isn't a signle level of generality at which all our appraisals are directed. It seems that different types of appraisals pick out processes at different levels of generality, even when appraising the same belief." More specifically, "We can detect at least two uses of reliability in appraisal. One is based on the reliability of the porcess narrowly construed, and appraises the status of the particular belief. Another makes use of reliability of a more broadly construed process, and describes the creditworthiness of the agent in having formed the belief." So while the generality problem has been worrisome for reliabilists, (in that a complete reliabilist theory is exepcted to determine how to select which process types to evaluate to determine the status of beliefs,) you argue that the situation has a certain advantage, and is exploitable by reliabilists: "It makes it possible to appraise beliefs in different ways by adverting to reliability at different levels."

More specifically, you write, "We should now consider what sort of appraisals make use, respectively, of narrowly-typed and broadly typed reliability (NTR and BTR, for short), and what roles they might play in our understanding of epistemic practices. NTR tells us a great deal about the etiology of a particular belief or narrow range thereof...However, the epistemic status identified by NTR does not necessarily accrue to the agent. NTR tells us little about the agent's overall capacities or cognitive practices, because it reflects only the etiology of such a narrow range of beliefs on such a narrow range of occassions. Thus while NTR seems central to assessing the status of a single belief [and appears to be 'the sort of reliability centrally involved in knowledge'], it is less important for assigning credit or blame to believers."

I like here especially the connections between this distinction, and questions of epistemic credit.We are mor willing to praise or blamed believers for BTR, since processes at the broad level of generality syas more about their status as cognitive agents. This useful distinction you say resembles Goldman's earlier distinction between global reliability and local reliability. I want to think more about this and have no critical comment, but wanted to point out another closely related distinction I found in reading Bob Audi's draft paper "Reliability as a Virtue." There he distinguishes "sectorial" reliability, that of a person in a subject-matter (or behavioral) domain, from "focal" reliability, that of a particular belief. One instance of the importance of the distinction to to testimony-based knowledge: "Here we may have reliable (and reliably grounded) belief--in aretaic language, reliably believing--but not believing from reliability (as a virtue). Indeed, acquiring knowledge from testimony does not imply that the *person* is reliable at all. A testimony-based belief may exhibit only *focal reliability*, the narrow kind exhibited by a single belief."

The overlap between the distinctions you and Audi develop only underlines the importance of your point of the need to exploit the fact that there isn't a single level of generality at which all of our appraisals are directed. In retrospect exploiting this might have advantaged me in the recent exchange I had with Jennifer Lackey over a testimony case similar to what Audi describes. I have some questions about what you say later in the paper about how what process types are relevant for waht appraisals, but I'm certainly taken by the importance of your distinction in illuminating how we appraise beliefs in different ways by adverting to reliability at different levels of generality.

Chris' draft paper "Metacognition and Epistemic Virtue" endorses much John Greco's account, but also takes issue with at a crucial point, leading to his own understanding of intellectual virtue. "I endorse Greco's...acount that a disposition to form beliefs is part of one's cognitive character iff it is stable and well integrated with the rest of the cogntive system. However, this notin of cognitive integration is too vague to do the work it has to do."

Part of the problem, we find, is too narrow a conception of virtue: "The trouble with treating virtues as belief-forming processes is that it seems to rule out any possibility of uniting the high-level and low-level virtues. Open-mindendess, intellectual courage, and the like are not dispositions to form beliefs...,One forms beliefs by believing the testimony of others, rather than by being oepn-minded. The open-mindedness is a disposition to treat the testimony of others in certain characteristic ways; it shapes one's secondhand belief-formation though it doesn't produce the beliefs itself."

To mark the distinction here, and make more unitary our account of the intellectual virtues, Chris proposes that we identify them with capacities for metacognitive control. Integration into cognitive character can be accomplished by metacognitive control. "That makes those capacities excellent candidates for intellectual virtues."

This leads eventually to his own analysis of intellectual virtue distinct from that of other extant truth-conducivist accounts:

"V is an intellectual virtue of a subject S iff:
(a) V is a stable capacity of S's to exert control over his processes in a way that allows S to form true beliefs and avoid forming false beliefs, and
(b) for all the processes under V's control, whichever of (i) or (ii) adverts to nearer possible worlds is satisfied:
(i) the process could be prevented from generating false beliefs in a large proportion of the nearest possible worlds in which it otherwise would;
(ii) the most similar possible processes that S might have had while remaining the same agent would not cause him to generate unreliably formed belief."

I won't comment on these latter, 'safety'-type conditions, though they seem conducive to safe-belief concerns shared by Sosa, Greco, Pritchard and others. My main worry is with the idea of control over processes that leads to the understanding of epistemic virtue as a capacity to control one's cognitive processes in a way that allows one to form true beliefs and avoid forming false ones. Having just read David Owen's book *Reason Without Freedom* which argues pretty strongly against our ability to "control" belief-formation, I view this claim warily, though of course I do find it highly advantageous from the responsibilist interest in research into the reflective virtues like those mentioned above. However difficult it is to make good philosophical sense of control over processes of belief-formation, I note that I don't mean to saddle Chris with holding that such control is necessarily self-reflective. metacognitive control is not always or necessarily akin to reflective access to reason, but is rather "what makes a bundle of disconnected processes into a system for acquiring reliable beliefs." This does appear to leave ample room for the automaticity of many of our belief-forming processes. As he writes,

"Even automatic belief-forming processes can be under effective metacognitive control, provided that they are more or less automatic in cases where this does not lead to unreliable belief-formation. (We cannot, of course, expect metaprocesses to entirely prevent unreliable belief-formation.) Thus, B can arise from intellectual virtue even if metaprocesses had no causal influence on its formation." This goes quite some distance towards alleviating my worries (as I treated automaticity in my own recent papers and used Nancy Snow's new paper on this), but doesn't completely erase the concern that more needs to be said to make sound philosophical sense of the claim that "successful cognizers with processes like ours need to control their processes in order to have reliably formed beliefs." Comments?

I'm very glad to hear that you liked the papers, Guy. Thanks for suggesting Audi's work on different types of reliability, which I'll have to look at. Regarding your comments on metacognition, I agree with your worries about the extent of our control over our beliefs. Part of the problem might be that "control" has the wrong connotations for what I have in mind. Maybe I should be calling it "management" or "regulation" instead.

Reflecting on one's processes, judging their deliverances, etc., are ways of regulating belief-formation, but there are many more subtle or indirect ways of accomplishing the same result. For instance, we take the salience of data to be an important signal of its reliability; we don't take indistinct visual images or weak, difficult-to-recall memories to be as trustworthy as vivid impressions. This is an important control process, because memory and perception are less reliable when they provide less salient data. But our disinclination to take hazy recollections at face value is often just as automatic as our inclination to believe clear and distinct ones.

Or suppose that in the heat of debate S has difficulty recognizing when others have presented cogent arguments against his views. (Maybe he is unwilling to admit defeat, or maybe he just has difficulty coming up with intelligent responses to his interlocutor and updating his belief-set at the same time.) But suppose further that S has a habit of reflecting on conversations after the fact, and in these relaxed circumstances he is good about revising his beliefs to take account of contrary arguments. S might have no idea that this habit allows him to take account of others' arguments; he might even not be self-aware enough to realize that he is too obstinate around other people. It's nonetheless a very effective way of keeping his belief-formation from being subverted by a potentially vicious tendency.

It's these non-reflective, non-volitional ways of regulating belief-formation that lead me to think that metacognitive control is better understood in a virtue framework than in traditional forms of internalism.

Chris,
I definately agree that non-reflective, non-volitional ways of regulating belief-formation should lead one to think that metacognition is better understood in a virtue framework than in traditional forms of internalism. I think with the language of "control over processes" I'm just vaguely worried about a kind of category error that such language suggests. Methods and strategies seems to be one thing, and processes in the brain another. Its not that the one doesn't exert control over the other, but that the control is indirect, I suppose I'd say, in a way that makes talk of metacognition, metacognitive "management," metacognitive reflection, metacognitive strategy selection, etc. very interesting and indeed vital, but talk of metacognitive control OF processes somewhat suspect.

The problem I'm suggesting seems intimately connected with your concern to distinguish and then re-combine "low-level" virtues (involved in the production of knowledge), and "high-level" virtues (such as open-mindedness, conscientiousness, and the like). So maybe I'm just underlining one way to read the implication of your very valid point that "the latter [which have largely been our focus here @JB] seem to be involved in different sorts of appraisals of agents." I mean I'd definately agree with you that "Open-mindedness, intellectual courage, and the like are not dispositions to form beliefs, though they are (loosely speaking) dispositions to ...engage in inquiry in certain ways, and that these virtues, in the fullest sense of the term, are directly concerned with metacognitive management and monitoring, whether self-reflectively or habitually/automatically. But I suppose I'd say that this shows all the more that the task which you set yourself of "uniting the high and low level virtues" (faculty and character virtues are other similar terms we've used here, or reliabilist and responsibilist interests in explanation), need to be kept distinct, while the language of "control over processes" seems to over-run such distinctions.

Guy,

I think you can treat methods and strategies as fairly general belief-forming process types, of which the narrow sorts of types that we advert to when we consider whether someone has knowledge are themselves tokens. So high-level virtues are involving in managing bundles of processes, rather than individual processes. In addition, ordinary evaluation seems to refer to characteristics of bundles of control processes rather than control processes themselves. High-level virtue-words, in particular, seem to describe features of wide-ranging control capacities that are involved in managing lots of low-level processes, and that potentially consist of bundles of different control processes that share certain features.

For instance, take conscientiousness to be a word for a trait that prevents one from forming beliefs too hastily, or on too little evidence. (This is a crude characterization of the virtue, but it'll do for my purposes.) So it describes a characteristic of one's control processes, which is that they effectively prevent a certain type of error over a pretty wide range of underlying processes. One's conscientiousness might arise from a single control process that can influence a wide range of object processes, though it seems more likely that it would be realized in a bundle of different control mechanisms that all tend to prevent this sort of error.

The control over any given process that conscientiousness provides is pretty limited because it only prevents certain types of mistakes and not others. So manifesting the virtue doesn't entail reliability. But it is still of great value because it applies to such a wide range of processes, and thus the conscientious person is more likely to be right over a wide range of beliefs in a wide range of environments.

If you're interested, I put a chapter of my dissertation on my website that gives more detail on how I'm trying to unify the different types of virtues. It's the paper called "The Structure of Intellectual Virtue".

Chris, that's a useful clarification. Yes I'd be very happy if you posted more of your dissertation. I've actually been meaning to ask if you have more on the different explanatory roles of NTR & BTR in respect to the generality problem--the different questions that they each address, and how those hang together into a broader conception of agency. I really liked that discussion alot in your paper on the generality problem, and wanted to know where you take it from there. If you have more on that from your dissertation or papers, please do post it or send me a copy.

P.S. I hope you got my letter about the JB Dinner on Friday. Hope to see you there unless you have other plans!