For those who might have wondered, I’ve been absent lately because I’ve both been swamped and mightily depressed. Often it seems as if I cycle in and out of depression, going through periods where I’m extremely productive and then falling into periods where I can barely get out of bed, where I border on autism, where I completely lose interest in everything and cannot focus on anything, and where I fall into thoughts of self-loathing and frustration, only aptly described by Lars in his comedy-tragedy Spurious (but without the comedy or W.), leading to a plethora of destructive and suicidal thoughts. Fortunately it seems that I’m beginning to turn a corner. I would also like to say that I’ve very much appreciated the emails and comments of those who have contacted me during this time.
At any rate, in the last few days I’ve been preparing my reading lists for my Fall metaphysics and ethics courses. I’m pretty excited about both of these courses. For the last few semesters my metaphysics course has largely been about God and God’s relationship to the world, focusing on Augustine, Aquinas, Descartes, Spinoza, and Leibniz. Exhausted by this routine, I decided to take a different approach next semester. In my metaphysics course I will thus be teaching Whitehead’s Science and the Modern World, Jane Bennett’s Vibrant Matter, DeLanda’s Intensive Science and Virtual Philosophy, and ending the course with Andy Clark’s Being-There: Putting Brain, Body, and World Together Again. Clearly this course will focus on events, processes, relations, ecology, and cybernetic conceptions of the world. In my ethics course the theme will be similar. We’ll be reading Jane Bennett’s The Enchantment of Modern Life, Alfonso Lingis’ The Imperative, Dennett’s Freedom Evolves, and Badiou’s Ethics: An Essay on the Understanding of Evil. The courses should, I hope, be both timely and a lot of fun.
At any rate, I wanted to return to Andy Clark’s Being-There: Putting Brain, Body, and World Together Again. I hadn’t read this book since graduate school, but it rewards rereading and is, indeed, one of the most exciting books I’ve ever read on cognition, mind, and our relationship to the world. Clark’s book is written in a crisp, clear, entertaining style that moves fluidly between research in cognitive science, developmental psychology, artificial life research, and neurology. All of this is used to formulated an account of situated and embodied cognition that challenges models of mind that conceive cognition primarily as symbolic manipulation divorced from an environment. In what follows I’d like to outline some theses on cognition more or less inspired by Clark’s book.
1. The only legitimate investigation of mind must unfold within a naturalistic framework.
While many, perhaps the majority, of theorists today would claim that they are naturalists, I think naturalism is more easily said than done. I take it that a truly naturalistic position and ontology requires two things, the first pertaining to general ontology, the second to how we approach mind.
a. Naturalism and General Ontology: At the ontological level I believe a position must meet three requirements in order to count as naturalistic.
i. Genesis: First, it must endorse the position that nothing comes for free and that everything that exists is the result of a genesis or production. This means that every naturalism necessarily requires an account of how forms or patterns are generated. Species don’t fall from the sky fully formed but must evolve. Concepts, categories, and norms must evolve or develop in brains and societies, and do not exist a priori or fall from the sky fully formed. This is why any naturalism is necessarily opposed to Kantianism. While there can indeed be historical a prioris (genetic inheritances and cultural inheritances) there can be no “strong” a prioris that are unconditional, ahistorical, and universal. As a consequence, forms are always a product or result of a genesis and therefore cannot serve as conditions. Any form of cognition therefore requires a developmental account of how it was produced or acquired. I take it that this is what Deleuze was seeking with his transcendental empiricism. I also believe that this requirement of naturalism is perpetually ignored in discussions of mind, language, and norms. There’s a tendency to put the cart before the horse, treating the products of a stochastic evolution as the conditions. These abstractions don’t explain but are precisely the things to be explained.
ii. Materialism: I take it that the second condition of any true naturalism is a thoroughgoing materialism. All things that exist necessarily exist materially as either entities in their own right or as material aspects of other material entities. I won’t say too much about this condition, except to draw attention to the implications it has for interaction among entities. If materialism is true then it follows that there is no action at a distance. This, I believe, is a crucially important principle, especially for social and political thought. The claim that there is no action at a distance is the claim in order for two entities to interact there must be some material interconnection. If, for example, I say that I see my daughter’s tricycle over there, eight feet away, then there must be some material connection that allows this connection to take place (photons of light bouncing off the tricycle and interacting with my eyes). Similarly, in a social context, if we wish to speak of something like “society” or “social forces” we require some sort of material mechanism through which humans come to be related throughout this society, such that they come to share similar ideas and customs, thereby reducing entropy or the brownian motion to which these persons would otherwise be suspect. It’s not enough to talk about structure. The mechanism by which structure persists through time and the processes by which they came to be must be accounted for. It is not enough to appeal to “ideas”, there must be an account of how those ideas come to spread throughout a population and organize their actions.
iii. Contingency: This follows from (i). If it is true that all forms or patterns are the result of a genesis or production, it also follows that these patterns and forms are contingent. This is to say, there is no necessary reason that they exist as they do and they are capable of being otherwise. There is no necessary reason that this or that species exists. Rather, the existence of each species is the result of an evolutionary genesis that could have gone a very different way as a result of contingent events. Likewise with concepts, categories, norms and so on. These are the result of a genesis, both biological and cultural, and are therefore capable of being otherwise.
b. Naturalism and Mind: As a consequence of the foregoing, I take it that any account of mind and cognition must situate mind within a naturalistic context, seeing the powers and capacities of mind as a result of a genesis and, above all, situating mind within a biological context. Any account of mind that evades these questions of genesis is illegitimate. Moreover, questions of genesis should not focus on ideational and cultural genesis alone (Hegelianism), but must also take into account bio-genesis. What does this mean? Humans and other animals have minds for the sake of acting in the world, reproducing, and surviving, not for the sake of knowing or representing the world. This is not to suggest that we cannot and do not come to represent and know features of the world. That’s not the point. The point is that prior to raising any questions of epistemology, knowledge, cognition, etc., etc., etc., we must remember that we’re animals that evolved the minds we have for the sake not of representing and knowing, but of acting. Our ability to know, represent, reason, etc., is a happy side-effect of how our minds evolved, not the primary vocation of our minds. This must be taken into account when theorizing mind. Much of the literature that discusses mind treats it as a self-enclosed symbol manipulating device that merely represents the world. This thoroughly distorts the nature of mind and, indeed, presents a picture of mind that would be thoroughly unable to survive in the world. A number of consequences follow from this action-centered conception of mind.
2. Consequences of Action-Oriented Conceptions of Mind
a. Cognition is constrained by the pressure of time. Because our minds are geared towards acting in the world (not representing the world) and because our computational power is temporally limited, mind, in order to be viable, must deal with the constraints of time. As cognitive science repeatedly shows us, symbolic manipulation or representational thought, while certainly something humans are capable of, is extremely unwieldy and impractical in dealing with the world. In Being-There Clark gives an example of representational models of mind and thought based on the CYC project which was an attempt to program a “cognitive encyclopedia” containing a massive body of representations that the computer would then be able to manipulate through inferences from one frame to another. Such a map might look like this:
Capital: (Jefferson City)
Residents: (Andy, Pepa, Beth)
State of: (United States of America)
The idea is that reasoning consists in making inferences upwards and downwards from these frames or symbols. Thus, given the frame “resident” the computer might infer “Andy” or “Beth”. Likewise, given “Beth”, the computer might infer “resident”. Likewise, given “Jefferson City” the computer could infer “Missouri” or “Residents”. There are a number of problems with this model. The first is that the time required to sort through these symbolic networks is extensive, making real world action based on these maps or models very ineffectual. Another problem is that these internal models or semantic maps do an extremely poor job dealing with the noise of the world because that noise is not coded in these representational systems. A naturalistic theory of cognition, reasoning, and knowledge takes these computational time constraints seriously and examines the manner in which these constraints are surmounted. Not incidentally, the temporal constraints on cognition and the manner in which our minds have evolved, coupled with the action-orientation of our minds, goes a long way towards explaining why 1) we aren’t particularly good at 1) abstract reasoning, 2) logical reasoning (in the formal sense), 3) long-term planning. These are precisely the sorts of deficiencies we would expect of minds evolved to act and respond to an environment rather than represent an environment. At any rate, a number of consequences follow from the time-constraints on cognition.
i. Because computational power is limited and the time required to compute through representations potentially endangers the organism that needs to decide and act, cognition has evolved all sorts of shortcuts. I’ll have a bit more to say about this in a moment, but for now the lesson to draw from this point is that representational models of the world, due to their high expenditure and costly time demands, are of limited value in navigating the world about us. Take the example of cooking. Suppose cooking required a detailed representational map or symbolic system of all the procedures to be executed in cooking the meal. Here I’m not simply talking about the different stages of the recipe, nor the ingredients to be used, but rather a model of every aspect of the procedure, e.g., all the motions that have to be executed to sautee the garlic. It’s clear that this would be an extremely cumbersome way of cooking my dish. While there are indeed representations involved in this process, it’s clear that I don’t represent each and every activity in which I engage in the process of cooking the dish. Rather, as I saute the garlic, I move the garlic back and forth in the olive oil with my wooden spatula (which, for some strange reason, I deeply love), possessing no map of this activity, but rather responding to a series of positions of the garlic, color cues, scent, etc., etc., etc. There’s a continuous feedback relation between my action, the subsequent position and state of the garlic, and my subsequent actions without an overarching map or series of representations. Cognitively I minimize my “computations” of these actions, instead responding in the moment, and working on the fly.
ii. Any cognitive system must be capable of dealing with noise. The problem with mental maps is that, as Bateson liked to say, the map is not the territory. The territory is always more complex than our map of the territory. This means that critters such as us are constantly encountering noise or unexpected events in the world about us. Mental maps and representational systems are unable to encode this sort of noise precisely because its, well, noise or the unexpected. One nice example Clark gives is of robots navigating an office with the task of removing empty soda cans. Those robots premised on the idea of representational maps and inferential webs do a particularly poor job at these sorts of task because the varying perspectives of the cans, situations that don’t fit with the representational map, etc., lead to all sorts of uncertainties in the system. By contrast, those artificial forms of life that rely very little on detailed representational maps, instead interfacing between minimal models and stochastic sensory phenomena do an exceptionally good job at removing the cans. Less is more.
iii. Clark does not discuss this issue, but I would also add that any advanced cognitive system must be capable of dealing with paradoxes arising out of system self-reference. A major difference between a tornado and more advanced autopoietic systems like the minds of bats is that the latter distinguishes between system-states and other-references. That is, cognitive systems like those found in bats distinguish between events that take place within them and events that take place in their environment. It is here that self-referential paradoxes emerge precisely because it is the system itself that distinguishes between what takes place inside of it and in its environment. In other words, the distinction between internal state and other-reference is a distinction drawn internal to the system. As such, such distinctions are inherently paradoxical insofar as they are simultaneously internal to the system and draw a distinction between the inside and outside of the system. We find these self-referential paradoxes all over the place. For example, the distinction between legality and illegality is a distinction drawn within the legal system. Likewise, the distinction between the normative and that which violates a norm are drawn from within the system of normativity (here we encounter the profound affinity Cary Wolfe outlines between Luhmann and Derrida). More advanced cognitive systems perpetually face the problem of how to resolve or unfold these paradoxes. The answer, of course, is time. Were these cognitive systems atemporal symbolic logic processing machines these paradoxes would spell their computational death. By contrast, the fact that operations within systems unfold in time is what allows these paradoxes to withdraw into the background and function in an operational efficacious way within the system. More on that another time.
b. The world is its own best model. This is perhaps my favorite thesis in Clark’s work. With this thesis, Clark discovers the entire field of contemporary media theory (at least that branch of media theory heavily indebted to McLuhan) in a rigorous framework of developmental psychology, cognitive science, and neurology. In section 2ai it will be recalled that I remarked that in order to function efficaciously in the world, cognitive systems must find shortcuts to deal with the constraints on time and computational power. Clark’s thesis that the world is its own best model is a big part of his answer to the question of how mind finds these shortcuts. In short, Clark’s thesis is that far from using cumbersome representational maps or model of the world, we instead use the world about us as our model of the world. In other words, in our knowledgeable activity much of the world isn’t mapped at all precisely because the world is already there. A number of implications follow from this.
i. The use of complex representational maps to navigate the world is an exception rather than the rule. Rather, we primarily use the perceived world itself as our model of the world. Here we encounter a profound overlap between Clark’s understanding of cognition and the manner in which Deleuze makes use of Riemann’s non-Euclidean geometry in his development of the concept of multiplicities. It will be recalled that a Riemannian multiplicity uses purely local information to plot the characteristics of a space. For example, rather than embedding a sphere on a plane to map the properties of the sphere, Riemann instead uses differential calculus and local points on the sphere to map the sphere. It dispenses with embedding space to map the properties of the space. This is very similar to what Clark is after in his discussions of using the world itself as its own map. In the text he gives the wonderful example of a person on an inclined plane in pitch darkness, i.e., they can’t see the incline. Their goal is to get to the bottom of the incline. How are they able to do this without a map of the territory or visual cues about the space? Each step provides local information about the properties of the incline allowing the agent to make decisions as to how to move in real time. A step in this direction is kinaesthetically registered as upward while in the other direction it is registered in that direction. In this way the agent is gradually able to solve the riddle based on local information alone without any global picture or representation of the properties of the incline itself. We thus get a Riemannian interaction with world as a technique for navigating the world based on continuous feedback loops between world, sense, body, and cognition.
ii. Cognition requires scaffolding from the world. Clark presents the regularities of the world as a scaffold for cognition and interaction with the world. The fact that many features of the world around us (in this environment) are reliable allows us cognitively to dispense with the need to represent or symbolize these features of the world while nonetheless reliably acting in the world. I’ll have more to say about this scaffolding in a moment, but one of Clark’s most innovative claims is that mind is prosthetic such that we actually think in, with, and through this scaffolding. This, of course, entails that changes in scaffolding also generate changes in capacities for action and cognition.
iii. Representations are never mere representations or mirrors of the world, but are also action-sequences. We already know this from Heidegger and Bergson, but it’s worth repeating. Bergson famously argued that perception is not representation but rather potential action. I perceive that, Bergson argues, that I am potentially capable of acting upon. What you perceive would thus be what you’re capable of acting or being acted-upon by. This is much of what Spinoza called “affect”. In Heidegger, the ready-to-hand is that dimension of entities such that in encountering them we encounter them as being potentials for use in some way or another. Thus, for example, when I perceive my spatula I don’t just represent it as a spatula, but rather perceive it as a set of potential actions or activities. What I perceive in the world about me is thus a set of potential action-sequences. This is a big part of the reason that object-oriented ontology is so difficult and why it is so difficult to regard things, as Spinoza suggested we do, in-themselves and for-themselves rather than in terms of our action-sequences. Here, probably, we encounter one of the major conflicts between our biological dispositions and the development of our sciences. To think something without thinking a potential use of that thing is damned hard. To appreciate a rotting apple in its own perfection and for the sake of what it itself is (a nod to the ethics of generosity developed by Bennett; what a beautifully inhuman and monstrous mind she has!) is exceedingly difficult.
iii. Thought doesn’t reside between the ears. This, I think, is Clark’s most exciting thesis and is one that deeply anticipates the sort of work Tom Sparrow is doing in the context of Merleau-Ponty and Levinas (and here I can’t praise Sparrow highly enough for developing the first truly novel theory of sensation we’ve seen in decades, if not centuries, where sensations are no longer locked in the mind, but are unleashed in the world as realities in their own right. He’s broached, without knowing it, an entirely new form of OOO. Listen to this guy, he’s undergoing some serious lines of flight.). For Clark a mind that doesn’t interface without the world does not do much cognizing at all. To think we must perpetually interface with the world about us. But that’s not all. The consequence that follows, for Clark, is that thought does not exist between the ears, in the isolated inner space of consciousness, but rather that mind is the assemblage of brains, bodies, and the world about them.
What does this mean? Take the example of complex multiplication. Being as feeble minded as many of us are, when confronted with a multiplication problem like “3276 X 765,899″ we need to resort to pencil and paper to solve the problem. For Clark mind is not is what is here going on in the brain, but rather is the entire assemblage of brain, body, pencil, paper and numbers. In his subsequent book, sadly entitled Supersizing the Brain, he recounts a story about Feynman who insisted that his notes were not a record of his work, but were the work itself. It was this interface, this interaction, that was mind and the thought. And if this is so, then it is because with scaffolding, with things like writing, paper, pencils, and so on, we become capable of forms of action and thought that would not otherwise be possible.
The implications of Clark’s distributed mind– also discovered by McLuhan –are profound. What is entailed here is that minds are a function of their prostheses. To think and to have mind is not simply to have symbolic activity go on in the brain, but rather to interface with a variety of different artifacts. But if that’s the fact, then the artifacts with which our brains form assemblages (spears, pencils, guns, drones, computers, factories, cooking utensils, etc, etc., etc.) mark different minds. These assemblages have different powers of acting, inferring, and cognizing based on the artifacts with which they engage. As Marx recognized long ago in the Manifesto, a farmer and a factory worker are not two variants of one and the same sort of thing (humans), but are entirely different species. Why? Because of the manner in which they are mediated as a consequence of the tools that they use and by which they are used. And this caveat of “by which they are used” is not merely a flourish, but the fundamental point. The causality between tools and, for lack of a better term, human bodies is bivocal, going in both directions at once. The technosphere is as much something used by persons as it is something that carries persons up, creating new goals and aims, pursuing their own teleological goals that might not mesh very well with our particular social and biological aims. Stiegler, here, is exemplary.
I’ve gone on way too long in this post. I have not sought to offer a theory of knowledge. That’s not the point. The point is rather that for such a theory to even be properly posed it must be posed in a framework akin to the one that Clark describes and that takes seriously the brute facts of cognition that he outlines. Above all, it becomes clear that knowledge is not merely a matter of symbolic manipulation, representational maps and the inference patterns that arise from them. That’s a part of the story, of course, but if it is indeed the case that mind is distributed and that the world is its own best model than it turns out that the mind/world opposition is woefully inadequate for posing these questions or thinking these issues. No, we must take into account distributed minds and the technospheres they inhabit and see knowledge not as an issue of a mind relating to the world, but rather of a society relating to a world with which it constantly interfaces. Luhmann, Kuhn and Foucault pointed the way to what epistemology must be to be epistemology (private minds and communicating minds are largely irrelevant to the issue of scientific realism), yet few have followed this route.