The differences between the GPS and MapQuest concern how this external information is stored. MapQuest works by providing you with an algorithm: a set of instructions such that if you follow them faithfully you will arrive at your destination-at least in theory. Leaving your residence, you turn west on 156th. Then at 77th you turn north. At the second light, you turn east on 144th, and so on. When MapQuest wants you to turn left, it needs to provide you with information that specifies precisely where you are to do this (e.g., at the intersection of 77th and 144th). There is no need for your GPS to do this. With your GPS, the time and place of the utterance is inextricably linked to the meaning of the utterance. The information contained in the GPS instructions is situated in a way that the information contained in the MapQuest algorithm (typically) is not. The GPS information is both external and situated. The MapQuest information is external but not situated (or, at least, not as situated).
We all know the problem with MapQuest directions. The chances are there are too many instructions for you to remember, and so you have to take them with you in printed form. And then, consulting your sheet to find out what you should do next, you slam into the back of the Camry that's stopped in front of you. Your GPS, on the other hand, gives you easily digestible snippets of information as and when you need themwhich it can do precisely because the information it provides is situated. Nevertheless, although they accomplish it in different ways, both the GPS and MapQuest have essentially the same function. You need to accomplish a task-getting from A to B. If you couldn't avail yourself of MapQuest or a GPS or some other form of external information storage system, then you would have to remember how to get from A to B "all by yourself." That is, the task would have to be entrusted to your naked biological-that is, neural-memory. However, to the extent that you are able to avail yourself of an external information source, the complexity, and therefore difficulty, of the task you have to accomplish in your head is correspondingly reduced. The task that you would have had to accomplish in your head is, in part, off-loaded onto the environment. Equivalently, we might say that the task is distributed onto the environment.
It is the possibility of this sort of off-loading that provides the starting point for the new science of the mind. However, recent technological developments of the sort that result in MapQuest or GPS systems are only the most recent manifestations of a process that began almost as soon as humans became humans. Human cultural development is, in part, a process of creating external information-bearing structures: structures that could be used to enhance our ability to accomplish important tasks. One of the most significant of these structures is the development of written language. In a classic early study, the Soviet psychologists Alexander Luria and Lev Vygotsky (1930/1992) identified the implications of this development for human biological memory.
Consider an early form of written language-a simple system of visuographic representation. Imagine two people (the example is Luria and Vygotsky's). One is an African envoy, entrusted with the task of remembering word for word the message of his tribal chief. The other is a Peruvian kvinu officer-an "officer of knots"-who uses a conventional system of knots tied in string to achieve the same purpose. As Luria and Vygotsky point out, for each new message, the African envoy must employ anew his (biological) memory. The kvinu officer, on the other hand, need employ his memory resources only once-in learning the "code" that allows him to access the information contained in the knots. Once he has done this, a potentially unlimited amount of information becomes available to him since a potentially unlimited number of knots can be presented to him. The African envoy faces a comparatively difficult task. But, for the kvinu officer, part of the difficulty of this task has been off-loaded onto the environment. By making use of an appropriate environmental structure, the kvinu officer significantly reduces the complexity and difficulty of the task that he must accomplish in his head (see also Rowlands 1999, 134-137).
Suppose you have to do something; it doesn't matter what. And suppose successfully doing this thing requires you to put in a certain amount of work. Then, this is a truism: if you can get something (or someone) else to do some of this work for you, you will have correspondingly less work to do yourself-as long, and here is a crucial caveat, as the work you put into to getting that something or someone to do the work for you is less than the work they thereby do. There is an old adage that captures this idea quite nicely. It says: why keep a dog if you are going to bark yourself? If there is some barking to be done, and you have a dog that will do at least some of it for you, then you have correspondingly less barking to do yourself (Rowlands 1999, 79-80). At the conceptual heart of non-Cartesian cognitive science is this barking dog principle.13
Any cognitive science, whether Cartesian or non-Cartesian, will be concerned with, at the very least, the following sorts of tasks: (1) perceiving (visually, aurally, etc.) the world, (2) remembering perceived information, (3) reasoning on the basis of information perceived or remembered, and (4) expressing this information, or listening to information expressed by others, in the form of language. Perceiving, remembering, reasoning, and the processes involved in the production and comprehension of language may not exhaust the legitimate business of cognitive science; but they certainly lie at the core of this science.
The central idea of non-Cartesian cognitive science is that cognitive tasks are not, in general, the sort of thing that need be accomplished only in the head or by a brain. If we had to accomplish a cognitive task using only our brains, then complex and difficult neural maneuvers might be required. However, if we are able to use relevant structures in our environment, then some of the complexity and difficulty of this task might be reduced: we off-load at least some of the task onto the environment around us. Roughly: we get the environment to do some of the work for us; and this reduces the work that we need do. What makes an environmental structure relevant? Again roughly: it carries information relevant to the task that we need to accomplish, and by using this structure, or acting on it in the right sort of way, we are thereby able to appropriate-make available-and employ this information in the accomplishing of the cognitive task in question.
If this sounds a little abstract, just think of our MapQuest instructions that we have printed out prior to our journey. This printout contains information relevant to the task we need to accomplish-getting from A to B. I am able to make this information available and use it by manipulating the page in the right way-for example, by picking it up, turning it the right way up, and holding it in front of my eyes. In the absence of this sort of manipulation, the information I require is present on the page but unavailable to me. Because I am able to act on the external structure-the page-in such a way as to make the information it contains available to me, the number and complexity of neural operations that I must perform to accomplish the task of getting from A to B is accordingly reduced (given that we are encultured in linguistic communities, we generally find reading novel information easier than remembering it). Acting on external structures in such a way that the information they contain is transformed from the merely present to the available lies at the heart of non-Cartesian cognitive science. This sort of action is, according to the non-Cartesian conception of cognition, part of what cognition is. That is, action of this sort-action that transforms information that is merely present in external structures to information that is available-forms a properly cognitive part of an overall cognitive process.
Consider another example that makes much the same point. Imagine how fiendishly difficult it would be to do jigsaw puzzles if we weren't allowed to pick up and manipulate the pieces (Kirsh and Maglio 1994). If we couldn't do this, then we would have to form detailed mental images of each individual piece, and by a process of mental rotation try to work out with which other pieces it fits. In such circumstances, jigsaws would be even less fun than they already are. But this, of course, is not how we do jigsaws. We don't do them this way because the pieces themselves contain information about which of the other
pieces they are going to mesh with and which they are not. By picking up individual pieces and "trying them out"-bringing them into close proximity with other pieces and adjudicating the likelihood of their fitting together-we make this information available to us. Therefore, we don't need to construct this information in the form of mental images and processes of mental rotation. We get the world to do some of the work for us, and we do this by acting on the world-manipulating and exploiting structures that contain useful information and thereby making this information available to us. Each piece of the jigsaw contains information that uniquely specifies the pieces with which it will fit. By manipulating the pieces, we transform this from information that is merely present to information that is available. And this, according to non-Cartesian cognitive science, is, in part, what cognition is.
5 Non-Cartesian Cognitive Science: Framework
We are beginning to glimpse at least the general contours of the nonCartesian conception of mental phenomena that underlies the prospective new science:
1. External structures carry information relevant to the accomplishing of a given cognitive task (or task that has a cognitive component). This information is present in these structures.
2. By using such structures in appropriate ways, I can transform the information they contain from information that is merely present to information that is available-available for detection by my sensory apparatus, and for deployment by my subsequent cognitive operations.
3. The information thus made available is, therefore, information that I need only detect rather than information I need construct or store.
4. Detection of information is cheaper than construction or storage.
And, finally, the money ball:
5. Action on external structures that transforms the information they contain from the merely present to the available (and so amenable to detection rather than requiring construction or storage) is part of what cognition is.
To see these principles in action, let us once more revisit MapQuest. The external structure in question is the printout of the directions. This printout contains information relevant to the accomplishing of a given task: getting from A to B. This information is present in the printout. Moreover, this is a task that is part locomotive, but also part cognitive (that is, it has a clear cognitive component). Therefore, condition 1 is satisfied. By using the structure in appropriate ways-for example, by picking it up, turning it the right way up, and holding it up in front of my eyes-I transform the information contained on this page from information that is merely present to information that is available for detection by me and deployment in my subsequent cognitive operations. This is condition 2. The information that is made available is information that I need only detect, for example, through perception. I do not need to store this information in the form of memory. Nor do I need to construct this information by way of a process of reasoning. This is condition 3. We find perceiving of information easier to accomplish than storing or constructing it. Typically-not always, not necessarily, but typically-perceiving makes fewer demands on our neural apparatus than either reasoning or storage. This is condition 4, and we shall look at reasons for thinking it is true later on. It is condition 5 that is the controversial one. We will look at this principle, and why it is controversial, in the next section.
The first condition of this general framework appeals to external information-bearing structures. In this context, to call something "external" is to say that it is outside the brain (or central nervous system). The structure in question is not a neural state or process. The structures I have looked at so far are also ones that are outside the body as a whole-GPS devices, printed pages, knots, and so on. However, like its Cartesian counterpart, non-Cartesian cognitive science is a broad church-as we shall see later, perhaps too broad. In some cases, the external structures in question might be bodily ones: things located inside the body of the organism but not part of the brain or central nervous system. This means that the notion of "use" referred to in the second condition must also be understood in a very broad sense. The way in which we use bodily structures in accomplishing our cognitive tasks is typically very different from the way in which we use structures outside our body. We shall return to this point in a later chapter.
At present, however, we need to look at a more fundamental question, one that pertains to condition 5.
6 The Big Question (Or, How Many Es Are Enough?)
It is easy to understand the attraction of utilizing external forms of information storage in the ways we have encountered above. The attraction, bluntly put is: thinking is hard. The perils and pitfalls of irresponsible brainuse were, as far as I am aware, first pointed out by the philosopher Alfred North Whitehead, early in the twentieth century:
It is a profoundly erroneous truism, repeated by all copy-books and by eminent people making speeches, that we should cultivate the habit of thinking of what we are doing. The precise opposite is the case. Civilization advances by extending the number of operations we can perform without thinking about them. Operations of thought are like cavalry charges in a battle-they are strictly limited in number, they require fresh horses, and they must only be made at decisive moments. (Whitehead 1911, 55)
As if we needed reminding, this idea has been the subject of extensive empirical confirmation in recent years (e.g., Baumeister et al. 1998). Cognizing is hard, especially when we use only the naked brain, and is something one should do only at decisive moments. The rest of the time, it is a good idea to cognitively delegate. As Andy Clark (1997) once put it, we make the world around us smart so we don't have to be. Underlying this delegation, of course, is the barking dog principle. Delegating works, as long as the work you put into the delegation is less than the work your delegates then actually do for you.
However, this idea begs one crucial question: what reason, if any, do we have for thinking that the sorts of extraneural processes invoked by non-Cartesian cognitive science are cognitive processes?14 Take, for example, manipulating the MapQuest printout-picking it up, turning it the right way around, holding it up in front of my eyes, and so on. This is an action that transforms the information merely present in the printout into information that is available. But what reason, if any, do we have for regarding this as a part of my cognitive processing? Intuitively, this seems far-fetched. It seems far more natural to divide up what is going on in the following way. First of all, there are the processes of perceiving, recognizing, and understanding the words contained on the page. These are genuinely cognitive processes, and they occur in the brain. That is, they are processes occurring somewhere in the brain's visual and language modules. The processes are facilitated by actions that I perform. It is, of course, much easier to read the page if I turn it the right way up. Everyone knows that. But that doesn't mean that this manipulation forms part of my processes of cognition. The right way to think about things, surely, is that my genuinely cognitive processes-ones occurring in my brain's visual and language modules-are supplemented and supported by actions that I perform on the world. To be supported by actions that I perform on the world is not the same thing as to be partially constituted by those actions (Rupert 2004; Adams and Aizawa 2001, 2010).
We can make the same point about the jigsaw puzzle. The real cognitive processes involved here are ones of visual perception and mental image formation and rotation. Of course, it is true that if I were not able to pick up the pieces and manipulate them, then I would have to engage in far more complex processes of image formation and rotation. The need to do this is considerably reduced by the fact that I can pick up pieces and manipulate them-bring them into close proximity and see if they fit together. Nonetheless, the need for image formation and rotation is not eliminated. At any given time, I select which piece to pick up on the basis of an antecedent sense of whether it is a likely candidate for fit. And I get this sense through prior processes of at least partial image formation and rotation. So, it is natural to think that the process of completing a
jigsaw puzzle can be divided into two sorts. First, there are the real cognitive processes: perception of the pieces, and subsequent image formation and rotation. These occur exclusively in the brain. Second, there are bodily actions I can perform on the individual pieces: picking them up and "trying them out." These are not real cognitive processes at all; but they do provide a useful supplement to those processes-a supplement that reduces the extent, and perhaps alters the nature, of the real cognitive processes that we must perform.
These questions are obvious ones. The inference from the claim that external operations I perform on the world supplement or complement my cognitive processing to the claim that these operations therefore form part of that processing is unsound. We cannot afford to run together the question of whether things going on in the environment drive (i.e., causally contribute to) a cognitive process with the question of whether things going on the environment partially constitute a cognitive process. The answer to the former question is, of course, yes. Things going on in the environment do causally drive cognitive processes. Even Rene Descartes could have agreed to that-assuming he could work out his problems with understanding how mental-physical interaction took place. The idea that things going on in the environment causally drive cognitive processes is an utterly mundane claim that anyone should accept. Cartesian cognitive science would not only accept this idea, it would insist on it.
Indeed, not only is this question an obvious one that must be answered in any development of a non-Cartesian alternative, it also makes it clear why we shouldn't be in any rush to blithely identify non-Cartesian cognitive science with 4e. In effect, the question begins the process of whittling down 4e into something that is leaner and, I hope to show, meaner. To see this, consider a way of rephrasing the objection developed above. In the case of MapQuest and the jigsaw puzzle, we might say that the external operations I perform provide a useful scaffolding or framework within which the real cognitive processes can operate. That is, the actions I perform on the world provide a scaffolding or framework within which my cognition can be usefully embedded. And this is the thesis of the embedded mind. This scaffolding in which my cognition is embedded considerably reduces the amount of (real) cognitive processes that I must perform in working out how to get from A to B, or in completing the jigsaw puzzle. But this doesn't mean that the framework is literally part of the cognitive processes in which I engage when I am navigating or jigsaw-completing.
The New Science of the Mind Page 3