Version of October 21, 1997. This metacommentary was presented as part of a symposium, organized by Gregg Schraw, on "Constructing Metacognitive Knowledge." The papers in the symposium were:
The symposium took place at the Society for Research in Child Development meeting, Washington, DC, on April 6, 1997.
 At a symposium like this one, there is a role for a commentator--and a role for a metacommentator. As a commentator, I can respond to the specifics of each paper. And I will be happy to do so. But as the papers are quite varied in approach, and some are notably voluminous as well, I think I can dispense commentary more easily to each of the authors individually by e-mail, than to a larger audience in a 15-minute talk. Besides, a symposium on metacognition--especially one that has already featured first-level commentary by thinkers of the caliber of John Flavell and Deanna Kuhn--seems to be in need of metacommentary.
 At today's summit conference, representatives of the metacognitive and "theory of mind" communities have been trying to find common ground. Common ground, it turns out, isn't so hard to come by.
 Everyone here (this metacommentator excepted) has been pouring forth great quantities of varied and ingenious empirical work. And some of the presentations today hold out the prospect of successfully teaching metacognitive skills. But to me the most important thing that has been going on this morning is the theoretical discussion. Without broader theoretical ideas to situate our empirical work in a common landscape, each of the local research traditions represented here could keep right on functioning without paying attention to the others.
 Today John Flavell has mentioned the tendency of different research programs, with different ideological roots and different animating questions, to operate in isolation. His commentary is sound, but it calls for a metacommentary. Why has significant intellectual exchange between metacognitive researchers and "theory of mind" researchers taken so long to develop? We take it for granted that local research programs in psychology will remain walled off from one another, unless some special effort is made to bring them together. (I was about to say an interdisciplinary effort, but that can't be right. We're all in the same discipline!) None of us here had to build distant analogies or engage in genius-level flights of creativity to bring these research programs together. What is it, then, about our discipline, the intellectual tools it affords to its practitioners, and the system by which most of us are rewarded for our contributions, that makes isolation and factionalism seem natural?
 Psychologists usually assume that the theories needed to explain our research findings (as opposed to announcing our allegiances and rallying our troops) are specialized and local. We assume that the people who actually conduct the empirical studies are best placed to arrive at those theories. Indeed, we often attribute "ownership" of specialized research areas to certain figures--and in doing this we are not just recognizing their expertise. For many of our colleagues the model of a theoretical publication is still an article in Psychological Review, a suite of empirical studies on a par with the theory that is put forward to account for their results.
 Yet the work of at least three of our presenters this morning, not to mention the work of this metacommentator, is strongly indebted to a thinker who never conducted a single empirical study of "the child's theory of mind," nor of most of the issues currently labeled "metacognitive." Ideas elaborated to make sense of other kinds of data, or simply to answer broad questions about the nature of knowledge, have given us a basis for relating research on metacognitive strategies to research on "theory of mind," and (as I argued at another symposium 6 years ago) many other things as well.
 Of course, I'm talking about the ideas of Jean Piaget--one of them, anyway. All of today's efforts to bring these research traditions together are based on a simple notion. The notion is that there is a difference between being able to do something, and being reflectively conscious of how you do it. And in the normal course of things (we will need to explain this qualification later), you have to be able to do it before you can become conscious of how you do it. This idea goes well back in Jean Piaget's intellectual evolution. His writings from the 1920s are full of references to la prise de conscience, or the act of becoming conscious. By 1950 reflecting abstraction (and its slightly more sophisticated relative, reflected abstraction) were making their appearance in his theory, at least with regard to logical and mathematical knowledge. By the mid-1970s reflecting abstraction and other process considerations were vying with cognitive structures for the central place in Piaget's theory.
 The Piaget connection testifies to the enduring power of an idea that is not always tied to local data sets or specific empirical research programs. Yet Piaget's idea, in its original setting at least, poses some noteworthy problems.
 Around the same time that Piaget was putting together his latter-day volumes on prise de conscience and reflecting abstraction, Mark Bickhard was setting forth a framework now called interactivism. Interactivism encompasses ideas about evolution, microgenesis, and linguistic communication, among other things, but the part that comes to mind in present company is the theory of knowing levels. And if the single-minded science of citology [note 1] is to be trusted, the theory of knowing levels has had far more influence on developmental psychology than any other part of the interactivist project.
 The knowing-levels story is also straightforward. Knowing is a relationship between an organism and the environment that the organism interacts with. The knowing system also has properties that are interesting and useful to know. But knowing is irreflexive--which is a fancy way of saying that the knowing system can’t know its own properties. A higher-level system that sits atop the first-level system can know those properties. It can know them by interacting with the first-level knowing system, just as the first-level knowing system interacts with the external environment. And, once this higher-level system is in place, an unbounded series of further knowing levels can arise: Knowing Level 3 by interaction with Knowing Level 2; Level 4 by interaction with Level 3; and so forth. We call what gets you from one knowing level to the next reflective abstraction, in obvious homage to Piaget.
 The knowing-levels idea has been developed further by Mark Bickhard and myself, and by Dave Moshman in a series of articles. (A significant feature of knowing levels theory that was emphasized in Gregg Schraw's paper today--and Deanna Kuhn's--is the possibility of downward influences. Once you are functioning at a higher level, it becomes possible to modify knowledge at a lower level to make it more consistent with pre-existing knowledge at the higher level. Until recently, downward influences had received no attention, except in some discussions of the development of values.) Even in stripped down form, the knowing-levels model requires apparatus foreign to Piaget’s theory. But there a couple of reasons why the apparatus may be worth importing.
 The first reason is that while reflecting abstraction became more and more important in Piaget's thinking, it remained subordinate to equilibration. Piaget's Studies in Reflecting Abstraction is a rich and absorbing book, but it is also a difficult book--difficult by Piagetian standards. (One chapter, in which Piaget seems to have laid down a "microgenetic" trace of a major transition in this thinking, currently sports 70 more footnotes in English translation than it had in the French.)
 Piaget always wanted to establish a polarity between equilibration and "mere" learning, between reflecting abstraction and empirical abstraction. On the good, strong side of this polarity, we get creativeness and novelty and cognitive advances of all sorts. On the weak, bad side, we get mere generalizations about objects and their properties. And that is when we are not reduced to "noticing," "reading off" data--or absorbing classroom instruction [note 2].
 Now for Piaget reflecting abstraction is what gets you from one major stage to another. And in most of its incarnations, it is what gets you from just doing, to knowing about doing. So, do you have to ascend to a higher stage to come up with new knowledge? Do you have to climb up a knowing level to be creative? Can't you make new constructions without doing either?
 Knowing-levels theory enforces a sharp separation between learning (which, for interactivists, has been cleansed of the dreadful connotations it had for Piaget) and reflective abstraction. Learning is happening all the time within a level of knowing. Learning is quite capable of constructing novel knowledge. What it doesn't do is yield knowledge about our knowing system, other people's knowing systems, and their properties. I find it significant that this feature of knowing-levels theory is usually missed, even in sympathetic accounts. I think it is missed because of a powerful Piagetian tendency to conflate novelty, creativity, and reflection.
 The second reason takes us deeper than the first one. Piaget rejected empiricism, and "copy theories" about where we get our knowledge. Yet his theory held on to "figurative" knowledge, data that just get "read off," perceptual snapshots, and other empiricist leftovers. Had Piaget been more consistent, he would not have concluded that these kinds of knowledge are inferior. He would have concluded that they are impossible. Then, of course, he could not have characterized empirical abstraction as he did. Nor could he have opposed reflecting abstraction to it in the same way.
 We're drawing near the heart of interactivism, a set of ideas that remain short on citological appeal. Interactivism is at root a conception of knowledge. In particular, it is the view that knowledge takes the form of possibilities for interaction. Knowledge cannot take the form of structures in the mind that correspond to structures in the environment. Such correspondences aren't knowledge and won't give us knowledge--unless we already know them, and we already know what is on both ends of them. Knowledge is an emergent property of the functional organization of a system that is capable of goal-directed interaction with an environment.
 When we make the sort of statement about the nature of knowledge that I just made, we are putting forward metaphysical hypotheses. We can't evaluate metaphysical hypotheses by collecting and analyzing more data. Yet they are clearly relevant to our current concerns. I will give just one further example here. On several occasions, Mark Bickhard and I have presented the following argument:
 If we are right, then an account of "the child's theory of mind" that relies on the notion of metarepresentation, while endorsing an account of representation by correspondence, is internally inconsistent. Our argument, if valid, poses difficulties for conceptions of development as diverse, and as worthy, as Josef Perner's, Annette Karmiloff-Smith's, and David Klahr's.
 Now you'd think that the sort of claim we've been making would have touched off a major controversy. Several people would have knocked the stuffing out of it, or found a way around it, or modified their conceptions in order to handle it. No such luck. None of those things have happened.
 They haven't happened, I submit, for the same reason that we "naturally" expect research programs in psychology to remain isolated from one another. In contemporary psychology, researchers do not agree on the answers to metaphysical questions (such as what human nature is, or what sorts of constraints on human development there are). That's no crime; it's the condition of our knowledge. The problem is that many of us do not place much value on study, or analysis, or rational argument concerning these questions.
 There are philosophical reasons for this state of affairs. Academic psychologists are still emerging from the empiricism that dominated earlier in the 20th century. We've gotten to the point where we realize that our conclusions aren't just based on data. We’ve realized that we can’t do without metaphysical ideas and assumptions. But it is still easier to treat these metaphysical claims as objects of faith or commitment, than as objects of analysis and rational argument.
 There are institutional reasons, too. Money, prestige, and recognition most easily flow to those who bring in external grants and train graduate students. Big empirical projects cost a lot of money and need grant funding; generating new theoretical ideas, and evaluating existing ones, usually do not. Work can easily be found for apprentices on empirical projects; not so easily on theoretical projects. Institutional imperatives favor the continued production of empirical research within a recognizable paradigm. And so long as grant funding, graduate student maintenance, and publication are not contingent on responding to in-principle arguments, there are few incentives to be bothered with such arguments. It is easier to dismiss them as mere kibitzing and nay-saying, mere comments from the peanut gallery. Attending to such comments from the peanut gallery would distract us from the continuous production of empirical studies (in which we are investing so much time and effort) [note 3].
 We are here today because we realize that people outside our local research program, people who don't belong to our faction, may have things to say that are valuable to us. We may not always agree with them, but we know that what they have to say needs a response. Let's try to spread that attitude.
 In conclusion, it's worth taking a moment to ponder some current realities. Universities are now a "mature" business, as they would say in the Wall Street Journal. Academic employment in our field has quit growing exponentially, probably for good. Our chief source of external funding is $5.3 trillion in debt (and has run up some $17 trillion in unfunded liabilities). Many of us are, or soon will be, under pressure to devote more of our time to teaching undergraduates. There will be fewer opportunities to pour forth streams of empirical studies--and fewer pressures to do so. Under those circumstances, the skills required to develop and to respond to in-principle arguments may come in very handy.
1. The quantitative tracking of citations. [Return]
2. Piaget's references to l'apprentissage scolaire are usually scathing. [Return]
3. In his commentary, the ever-generous John Flavell said that the dearth of communication between metacognition researchers and "theory of mind" researchers was "not due to provincialism, or lack of vision on the part of the researchers concerned." Actually, I think it was. But not because any of the researchers in question were inherently or inevitably small-minded. Rather, they had philosophical reasons and institutional incentives for acting provincially, and in those respects they were no different from many other academic psychologists. [Return]