An Introduction and Critique of Daniel Dennett’s essay “Intentional Systems Theory”
Pseudonymously authored by Marsilio Gagliano (Aka, KMS) for Marsilio’s Reading Group, Open Habitat Project, Second Life © 2009 [Meeting Sunday, 16 August 2009) @ The Open Habitat Sim.
INTRODUCTION
I have written the following little essay to aid us in discussing together a reading graciously provided by Daniel Dennett entitled “Intentional Systems Theory” (2009) available at the Tuft University Website in pdf format.
Follow this link: http://ase.tufts.edu/cogstud/papers/intentionalsystems.pdf
Our previous discussions of Nagel’s “What it is like to be a Bat?” and Whitehead’s “Nature Lifeless” went very well, but you may have noticed from the transcripts that our discussions departed rather rapidly from a consideration of the texts themselves. One of the useful features of a reading group, in contrast to other sorts of discussion groups, is that reading groups owe consideration of the texts themselves. So, I wanted to delay our departure from Dennett’s text by what may at first appear to be an odd means – by attacking it. Based on what I hope is a careful and accurate consideration of Dennett’s thesis in sections I and II of the following essay, I argue, in sections II and IV, that Daniel Dennett is an ironic metaphysical chicken, a cowardly beast whose theory of intentional systems theory is logically consistent but incomplete. My argument can be summarized in four steps:
STEP ONE: Dennett defines ‘stances’ as methods of predicting the behavior of some object in order to inform our practical reasoning, that is, reasoning in order to achieve something good (or apparently good) for the one who reasons practically.
STEP TWO: Dennett argues that, from an epistemological point of view, it is reasonable to apply the intentional stance to an object whenever it is reasonable to consider that object as autonomous, that is, as acting in such a way as to achieve something good for itself rather than as something to be used to achieve the good of another.
STEP THREE: From a metaphysical point of view, Dennett’s position is ironic: Dennett thinks that it is reasonable to consider things as autonomous, as having ends in and for themselves, only if it renders them more useful as a means to accomplish goods or ends extrinsic to themselves – or – why Dennett makes me feel like a deontologist stranded on a desert island with a utilitarian.
STEP FOUR: Dennett’s Intentional Systems Theory is not logically inconsistent, but it is incomplete because it cannot account for intentional differences. Intentional differences exist between two intentional states A and B when ascribing either A or B to an object does not yield significantly different predictions about the future behavior of the object but does yield significantly different assessments of the meaning of that behavior – or why Daniel Dennett is an ironic metaphysical chicken.
My hope is that you will give Dennett the benefit of the doubt and read his essay carefully with an eye towards defending him against the charge of being an ironic, metaphysical, chicken. The only rule is that any defense you may offer must be based on the text itself, or your thoughts about it, rather than, for example, an appeal to some other text by Dennett or by some other author. Since Dennett responds to some objections to his theory later in the essay, this essay may provide the starting points for a defense against my objection. They may even provide ammunition for going on the offensive and arguing that I am a confused metaphysical bull, a reckless beast devoid of due philosophical caution.
What side you take doesn’t really matter for present purposes: you may be a pro-Daniel Denettian or a pro-Marsilio Ganglianian. Again, my hope is that, by making a direct attack on this specific reading based upon specific passages in the reading, we will, as a group, be motivated to read Dennett more closely than we otherwise might outside of this context. If someone called me an ironic, metaphysical chicken, I hope that someone would take up my cause if I were not there to respond in person. (How dare that Gagliano for dissing that decent Dennett dude!)
Whether you read Dennett’s or my essay first is up to you. My suggestion would be to review quickly the four steps in my argument so that you can be on the lookout for potential responses to my charges. That is, knowing the four steps will help you read with a purpose. Having read Dennett, you can then read the rest of the essay that follows to see who wins, the ironic metaphysical chicken or the confused metaphysical bull.
Reading both the bull’s and the chicken’s essays may prove a bit much; so, I hope to stretch this discussion over two periods. In any event, I have tried to tie this material to the previous materials we have read and also with the wonderful discussion of intentionality held in Second Philosophy a few weeks back.
May the history of our little linguistic community grow deeper. Take care.
Marsilio (marsiliof@comcast.net)
________________________________________________
PART ONE
STEP ONE: Dennett defines ‘stances’ as methods of predicting the behavior of some object in order to inform our practical reasoning, that is, reasoning in order to achieve something good (or apparently good) for the one who reasons practically.
As practical, we reason in order to act, to do, or to make something. Practical reasoning requires that we make predictions about the future behaviors of other persons and things. In order to be useful, these predictions must be not only correct but also timely, that is, if one cannot make the prediction in time, the information contained in the prediction may become useless. Often the demand for timeliness competes with the demand for correctness. The more quickly one makes a prediction, the greater is the risk that it will be incorrect in one or more respects.
In “Intentional Systems Theory,” Daniel Dennett refers to the basic methods we use to predict behavior as “stances,” of which there are three: the physical stance, the design stance, and the intentional stance. Using the physical stance, we predict an object’s future behavior on the basis of the “laws of physics and the physical constitution of the things in question.” (p.2) Using the design stance, we assume that the object in question is designed for a certain purpose, that it is sufficiently well designed to accomplish that purpose, and that it will not malfunction or break.
Imagine, for example, a child asks you what will happen when she presses the buttons on a calculator she found in her desk drawer in the following order: 2, x, 2, and =. Adopting the physical stance would require that you 1) apply the relevant laws of physics, 2) comprehend the physical characteristics of the calculator’s various components and 3) having carefully gathered current data about this particular device, calculate the future of particular groups of electrons as they wend their way through the circuitry of the machine. Adopting the design stance, requires that 1) you assume that the device was designed to correctly perform mathematical operations and that it was in good working order and 2) you know that multiplying 2 by 2 is four.
The physical stance is less risky and less timely than the design stance. If you have a thorough knowledge of physics and of the physical properties of the components of the particular calculator that the child found in her desk, then the average future of the millions of electrons, which are responsible for causing the elements of the liquid crystal display to darken and form the conventional symbol for the number four, can be known with statistical certitude. The design stance is more risky: the calculator may be broken or poorly designed. But from a practical point of view , the advantages of timely prediction outweigh the increased risk of making a false prediction.
Now suppose we select at random 500 human beings who have completed 12 years of education and are between the ages of 25 and 35. We want to predict how many of these individuals will resort to the design rather than the physical stance in predicting what will happen when the buttons on a calculator are pressed in the following order: 2, x, 2, and =. They are permitted to push any sequence of buttons before they make the decision except for the sequence 2, x, 2, and =. Each of them firmly believes a) that the calculator in question is a made by Casio, 2) that it has been recently purchased at Wal-Mart, and 3) that he or she will receive a $10 reward for making a correct prediction. They are given access to a well equipped physics lab and three days in which to make their final decision, but they cannot drink or eat until they have made their decision. Offhand, I would predict that the vast majority of our test subjects would either say “four” right away or say it after having made a few practice calculations and that they would make this prediction based upon the design stance.
Note that in making this prediction, I do not adopt the physical stance towards these people; I have no idea what might be happening in their central nervous systems. I do not adopt the design stance; I make no assumptions about what these individuals may or may have not been designed to do. Rather, I adopt an intentional stance. I assume that they have certain beliefs and desires and that they will act rationally. I assume few of them will have the requisite knowledge for adopting the physical stance, which would be error prone for anyone not well versed in electronics. I also assume that even if adopting the physical stance were less risky, the cost of its application (time without food and water) would outweigh the benefits of its application – a lousy ten dollars. In adopting the intentional stance, I treat these tests subjects as “an agent of sorts, with beliefs and desires and enough rationality to do what it ought to do given those beliefs and desires.” (p. 3) I make a lot of assumptions. My prediction is risky, but timely – very, very timely.
The intentional stance is a very fruitful way of predicting the behavior of things other than humans. Chess-playing computers, robots used in manufacturing, dolphins, flatworms and amoebas, even self-replicating macromolecules can be successfully handled using the intentional stance – that is to say, their behavior can be more or less successfully predicted if one thinks about them as agents with beliefs and desires that act rationally so as to achieve what is good for themselves.
PART TWO
STEP TWO: Dennett argues that, from an epistemological point of view, it is reasonable to apply the intentional stance to an object whenever it is reasonable to consider that object as autonomous, that is, as acting in such a way as to achieve something good for itself rather than as something to be used to achieve the good of another.
Dennett says that applying the intentional stance is sometimes “well-nigh obligatory – when the artifact in question is much more complicated than an alarm clock” (p. 3) In predicting the future behavior of a chess-playing computer, it would be impractical to resort to they physical stance or design stance. If you assume that the computer knows the rules of chess and can pick out those moves which are tactically and strategically useful for achieving its desire, victory, your predictions would be less certain than those that fall back upon the design stance (which requires sifting through millions of lines of computer code) or the physical stance (which requires calculating the flow of electrons), but practically speaking, if your purpose in predicting the computer’s move is to win a fun game of chess, then the intentional stance is the only stance that is reasonable to adopt. (p. 4.) Of course, if your purpose is to design a better chess playing computer, the story might be different.
Sometimes applying the intentional stance is silly. Though I can successfully predict that wood cutting implements will have sharp edges of some sort, it would be unreasonable to make this prediction on the basis of their having a desire to cut wood. So we might ask ourselves, when is it reasonable to apply the intentional stance. Here is what Dennett has to say:
“Whereas our simpler artifacts, such as painted signs and written shopping lists, can indeed be seen to derive their meanings from their functional roles in our practices, and hence not have any intrinsic meaning independent of our meaning, we have begun making sophisticated artifacts such as robots, whose trajectories can unfold without any direct dependence on us, their creators, and whose discriminations give their internal states a sort of meaning to them that may be unknown to us and not in our service.” (p. 8)
Please note, then, that in order reasonably to apply the intentional stance to some objects requires that it be reasonable to consider that object as capable of entertaining something that is intrinsically meaningful to it independently of any meaning it may have for us. To underscore this point, look at what Dennett has to say about flatworms and amoebas:
“Consider a simple organism – say a planarian or an amoeba – moving non-randomly across the bottom of a laboratory dish, always heading to the nutrient-rich end of the dish, or away from the toxic end. The organism is seeking the good, or shunning the bad – its own good and bad, not those of some human artifact user.” (p. 9)
In this passage, the phrase “its own good” does the same work as the phrase “intrinsic meaning independent of our meaning” did in the first passage. Having one’s own meaning or “seeking one’s own good is a fundamental feature of any rational agent.” (p. 9) Elsewhere, Dennett refers to the capacity to possess intrinsic meaning or one’s own good as autonomy: “What matters in the identification of the agent to whom beliefs and desires are properly attributed is autonomy, not specific structures.” (p. 16.) To identify something as an agent, and thus as an object to which the intentional stance may be applied, requires that there be something good for that object in itself and apart from the purposes of an extrinsic user. The goodness of a hammer is summed up in its usefulness to the one who hammers; there is no good for the hammer in itself. But there is something good for an amoeba, regardless of its utility for us. The amoeba is autonomous (from the Greek autos = itself; nomos = law); it is in some sense a law unto itself. It has its own purposes and ends; and, it is in virtue of this autonomy, argues Dennett, that it is reasonable to treat it as an intentional system (as a rational agent with beliefs and desires seeking rationally to secure its own good) when we want to predict its future behavior.
In short, then, here is the answer to our previous question about when it is reasonable or not to apply the intentional stance: it is reasonable to apply the intentional stance in predicting some object’s behavior iff it is reasonable to attribute some degree of autonomy to that object, and it is reasonable to attribute autonomy to an object iff there is something that can be described as intrinsically good for it, as opposed to its utility for another.
Note that this question was an epistemological one. It was a question about how we go about knowing thing, about when it is reasonable to apply this or that kind of explanation to this or that object.
PART THREE
STEP THREE: From a metaphysical point of view, Dennett’s position is ironic: Dennett thinks that it is reasonable to consider things as autonomous, as having ends in and for themselves, only if it renders them more useful as a means to accomplish goods or ends extrinsic to themselves – or – why Dennett makes me feel like a deontologist stranded on a desert island with a utilitarian.
What about the metaphysical questions? Are chess playing computers really rational agents, are they really things that have beliefs and desires and for which some things are intrinsically good or bad? Intentional terms, Dennett points out, “are spread across the spectrum” and are applied to things as diverse as evolution, cuckoo chicks, computers, male drivers, female bakers, and fictional detectives. (See pp. 11-12) Do all such things really have beliefs and desires? Are they really intentional systems? Or do they simply behave “as if” they are intentional systems?
Dennett answers the metaphysical question in the first paragraph of our reading. The main thesis of “Intentional Systems Theory” is: “Anything that is usefully and voluminously predictable from the intentional stance is, by definition, an intentional system.” (p. 1) If it looks like a duck, walks like a duck, and quacks like a duck, it’s a duck. If treating something as an intentional system leads to a lot of useful (that is more or less accurate and timely) predictions capable of guiding our practical reasoning, then it is an intentional system. If it is useful to think about something as if it were some sort of thing called X, then it is an X. To quote Dennett again: “the intentional stance works (when it does) whether or not the attributed goals are genuine or natural or ‘really appreciated’ by the so-called agent . . . . Does the macromolecule really want to replicate itself? The intentional stance explains [predicts] what is going on, regardless of how we answer that question. . . .Seeking one’s own god is a fundamental feature of any rational agent, but are these simple organisms [planarians and amoebas] seeking or just ‘seeking’? We don’t need to answer that question. The organism is a predictable intentional system in either case.” (p.9)
Note the irony of Dennett’s main thesis, a thesis that relies on the implicit claim: if it is useful to think about something ‘as if’ it were some sort of thing called X, then it ‘is’ an X. In the present context, the claim is more specific: If it is useful to think about something ‘as if’ it were an intentional system, then it ‘is’ an intentional system. But 1) to say that something is useful is to say that it serves my practical purposes, and 2) to say that something is an intentional system is to say there is an intrinsic good for it regardless of how it might be good for me. So the irony is this: it is reasonable to think of something as having intrinsic purpose only if thinking so serves some purpose extrinsic to it, some purpose which is not its own, some purpose that is to be defined in terms of its utility for me or my kind.
In calling Dennett’s position ironic, I stop short of calling it logically inconsistent. If there is something wrong about what he is saying, it is more subtle than an error in logic. In order to better express this irony, I want to turn to meta-ethics for a moment. Meta-ethics is the study of moral theories. Moral theories are theories which lead to conclusions about what ought or ought not to be done. Two moral theories are especially prominent in our culture: utilitarianism and deontology – theories associated with John Stuart Mill and Immanuel Kant respectively. Roughly speaking, Mill enjoins us to act always in such a way as to promote the greatest good for the greatest number. Kant enjoins us to never treat other persons as a mere means to our own ends, but as ends in themselves – that is, as autonomous rational agents whom we must respect as good in themselves and as enjoying goods for themselves. In the case of act utilitarianism, circumstances may render it permissible to take the life of an innocent person as long as that action maximizes the greatest good for the greatest number; in the case of deontology, taking the life of an innocent person is always forbidden, for it is to treat that person as a mere means.
Now suppose you and I were stranded on a desert island, and you ask me whether I believe it would ever be permissible for me to take your life. I respond that I believe it is not permissible. You are a Kantian and conclude that I am a Kantian; I respond that you are mistaken. I argue instead that in present circumstances, the greatest good for the greatest number – the two of us – will be achieved only if we can rely on one another not taking the other’s life and that the only way we can rely on that is by adopting the rule that, regardless of our future circumstances, neither of us ought to take the life of the other. I do not promise not to take your life because I believe that I would be treating you as a mere means; rather, I promise not to take your life precisely because you and I are the best means to each other’s survival. As a Kantian, you would agree with my conclusion – it would wrong for either of us to take the other’s life. You may be satisfied that you can reliably predict my behavior in regard to taking your life. I will never take your life under any circumstances; but you may find the reasons I have for not taking your life highly ironic. From your point of view, I vow never to treat you as a mere means to an end because such a vow is the best means to secure an end, survival.
In reading Dennett, I sometimes feel like a Kantian who has been stranded on an island with a utilitarian. Dennett is willing to consider me an intentional system (who can enjoy intrinsic goods that are somehow independent of his aims and purposes), but he does so only on the grounds that considering me as an intentional system allows him to make useful, that is timely and reasonably accurate, predictions about my behavior as he pursues his aims and purposes. I would not fear for my life if I were marooned on a dessert island with a utilitarian of the sort I have described. Given this utilitarian’s beliefs about the way things work and given his or her moral theory, I can reasonably predict his or her future behavior. Take it as a near certainty that he will not kill me, no matter what the circumstance. Practically speaking, I can regard him as if he were a Kantian who respected me as an end in myself, and not a mere means in his survival project; but as to the meaning of his actions towards me, whether or not he is a utilitarian or a Kantian seems to make a really big difference. Let me call this difference ‘the intentional difference’. Intentional differences exist between two intentional states A and B when ascribing either A or B to an object does not yield significantly different predictions about the future behavior of the object but does yield significantly different assessments of the meaning of that behavior.
PART FOUR
STEP FOUR: Dennett’s Intentional Systems Theory is not logically inconsistent, but it is incomplete because it cannot account for intentional differences. Intentional differences exist between two intentional states A and B when ascribing either A or B to an object does not yield significantly different predictions about the future behavior of the object but does yield significantly different assessments of the meaning of that behavior – or why Daniel Dennett is a metaphysical chicken.
I take it for granted that I know what it means to talk about meaning, until, of course, I ask myself explicitly what it means for something to have meaning. Dennett has insisted that some of our more sophisticated artifacts, like robot poker players, can engage in activities that have “intrinsic meaning independent of our meaning . . . whose trajectories can unfold without any direct dependence on us.” So whatever the meaning of “meaning,” both Dennett and I must account for it in some fashion. Dennett has argued that it is reasonable to consider objects as intentional systems (that is, systems for which meanings can exist independently of those ascribed to it by external agents) whenever considering them as such systems yields rich and voluminous predictions about their behavior for the sake of accomplishing our own ends. Though ironic, Dennett’s position is not logically inconsistent; it is, however, incomplete. It fails to account for intentional differences, where intentional differences are defined as different intentional states that lead one to the same conclusions about the future behavior of a system but different assessments of the meaning of that behavior. So I do not disagree with Dennett because I think it is wrongheaded to ascribe intentional states like desire to things like macromolecules; such ascriptions would be wrongheaded only if one adopts some version of matter like that described in Whitehead’s “Nature Lifeless”, which we read a few weeks ago. I do not think matter is necessarily lifeless. But I do think Dennett’s account is incomplete because some objects seem to be the kinds of things for which things can matter and others do not. Either winning doesn’t really matter to a chess computer because it is not the sort of thing to which winning can matter, or winning does matter to the chess computer, in which case the computer is the sort of thing to which winning can matter, that is, for which winning can mean something. If something can mean something to a computer, then there must be something that it is like to be a computer; the autonomy which Dennett ascribes to intentional systems implies that there must be some point of view, called the chess computer’s point of view, from which point of view winning has meaning and matters. If computers are or can be the sort of thing to which something can mean something, then computers are or really can be quite different in themselves than from how I conceive them to be. I don’t things could matter to a computer; I could be wrong. But then I would be really wrong and my error would not consist in or manifest itself as an error in my predictions about the computers future behavior in so far as it is relevant to my own practical projects; I would be really wrong about what computers are or can be in themselves.
Pausing to consider how things might be in themselves apart from the predictions we make about their future behavior for the sake of accomplishing our own goals is the essence of metaphysical speculation. Metaphysical speculation requires a certain amount of courage; we are on risky grounds when we attempt to transcend, however briefly, the narrow confines of our day to day pragmatism. Dennett is willing to assert that all sorts of objects in the world are the sorts of things to which other things can matter and for which they can have meaning; but he reduces what being an intentional system means to its cash value in our practical dealings with the world. He lacks the courage to go the next step and ask “what must something be in itself if something else can have meaning for and/or matter to it”. Daniel Dennett is a metaphysical chicken.
Marsilio Gagliano, Aka. Kms (© 2009, 9 August)
May 18th, 2010 at 22:07
??? ??????? ??? ????????. ??????? ???????….
INTRODUCTION
I have written the following little essay to aid us in discussing together a reading graciously provided by […….