______________________________________________ CONTENTS F ______________________________________________ DECLARATION________________________________________________I ACKNOWLEDGEMENTS________________________________________II ABSTRACT__________________________________________________ III INTRODUCTION: FUNCTIONALISM AND EMBODIED, EMBEDDED MIND 1 Functionalism, in a few words ___________________________________________ 2 Embodied Cognition as Undermining Functionalism ______________________ 3 Embodied and Embedded Cognition as Extended Functionalism __________ 6 Where do we go from here? _____________________________________________ 8 CHAPTER 1: FUNCTIONALISM IN DETAIL__________________________ 9 1.1 The Background and the Spirit _______________________________________ 9 1.1.1 Behaviourism __________________________________________________________ 10 1.1.2 Mind-body Identity Theory ________________________________________________ 11 1.1.3 Materialism ____________________________________________________________ 13 1.1.4 Anti-reduction and methodological autonomy _________________________________ 13 1.2 Methods for defining functional roles ________________________________ 14 1.2.1 Turing Machines and Machine tables ________________________________________ 15 1.2.2 Lewis’ Ramsey-sentence method of funtionalization _____________________________ 17 1.3 Commonsense vs. Empirical functionalism __________________________ 18 1.3.1 Commonsense functionalism _______________________________________________ 18 1.3.2 Empirical functionalism __________________________________________________ 19 1.4 The liberalism-chauvinism dilemma _________________________________ 21 1.4.1 The charge of chauvinism _________________________________________________ 21 1.4.2 Functionalism and abstraction _____________________________________________ 24 1.4.3 Functionalism and Multiple Realization ______________________________________ 25 1.5 Functionalism without multiple realization and chauvinism? __________ 28 1.6 Sum up and Conclusions ___________________________________________ 33 CHAPTER 2: THE BODY-DETAIL MODEL: SHAPIRO’S ARGUMENT ____ 36 2.1 Body Matters ______________________________________________________ 38 2.2 The Body’s Role in Perception ______________________________________ 39 2.2.1 A skill-based model of perception ___________________________________________ 42 2.2.2 Sensory substitution and plasticity __________________________________________ 45 2.3 Body Concepts ____________________________________________________ 47 2.3.1 Concepts and Metaphors__________________________________________________ 48 2.3.2 Body and language ______________________________________________________ 48 2.3.3 Individual embodiment vs. social embeddedness _______________________________ 58 2.4 Leaky Minds _______________________________________________________ 60 2.5 Conclusions _______________________________________________________ 64 CHAPTER 3: EXTENDED FUNCTIONALISM _______________________ 70 3.1 Extended computation _____________________________________________ 71 3.1.1 Exploitative Representation _______________________________________________ 72 3.1.2 Wide computationalism ___________________________________________________ 73 3. 2 Ballard et al.: The computational role of vision ______________________ 76 3.3 Dispositional beliefs and extended functionalism ____________________ 78 3.4 Extended Functionalism and the Body_______________________________ 80 3.4.1 Body as a functional notion________________________________________________ 82 3.4.2 The negotiable body _____________________________________________________ 84 3.4.3 Body and Mind – a one-to-one match? _______________________________________ 86 3.5 Status _____________________________________________________________ 87 CONCLUSION _______________________________________________ 94 BIBLIOGRAPHY ______________________________________________ 96 ______________________________________________ INTRODUCTION FUNCTIONALISM AND EMBODIED, EMBEDDED MIND ______________________________________________ In recent years it has become close to orthodoxy in the philosophy of mind to view the mind and cognition as essentially embodied and embedded. This view has developed as an opposition to a tendency within cognitive science to view mind and cognition as a matter of purely internal computation. So far the focus has been on developing the theses within the embodied and embedded framework and with less attention being given to how this framework relates to the views that were prevalent in the immediately preceding philosophical debates on mind. An exception to this is Lawrence Shapiro’s book “The Mind Incarnate” (2004). Here Shapiro argues, among other things, that the embodied, embedded approach undermines a functionalist program. In contrast and as a response to Shapiro’s argument, Andy Clark, in his article “Pressing the Flesh” (2006), claims that the embodied, embedded approach can be seen as a form of extended functionalism. My aim in this project will be to explore the logical space between functionalism and the embodied, embedded approach. In this I will focus on the question of whether these two positions are compatible or whether the embodied, embedded approach undermines functionalism, as suggested by Shapiro. My answer to the latter question is going to be that this is not the case. However, before we get to this result there is a lot of ground to be -1- covered and a few philosophical obstacles that need to be pointed out and overcome. In the remaining paragraphs of the Introduction I will introduce the different positions, give an outline of the possible tension between them and develop and present the central questions of the project. Functionalism, in a few words In basic terms functionalism in philosophy of mind is the metaphysical claim that mental states are individuated in terms of the effect the world has on them (input), the effect they have on each other (internal interaction) and the effect that they have on the world (output) (Braddon-Mitchell and Jackson 2007). To take a simplified and standard example, consider the mental state pain. A functionalist account could characterise pain as a state that is often caused by tissue damage, that tends to produce the belief that there is something wrong with the body and the desire not to be in this state, and in the absence of stronger desires may cause moaning or even crying. According to functionalism, a creature that has something that plays this role is in pain. (Levin, 2004) What it is that occupies this role is of less importance. The case is similar in connection with other functional kinds. In this way, what is important about something being a thermostat is the way, it controls the temperature by turning heating or cooling apparatus on or off. Such a device might be constructed in different ways, however if it plays the appropriate role it is a thermostat. In the same way, what is important about being a left winger on a hockey team is how this position is coordinated with the rest of the team. It -2- seems reasonable to say that what it is to be a left winger does not depend on whether it is Fernando or Peter that plays this position. Who plays the role might affect the outcome of the match, of course, but it would not affect the concept of what a left winger is. In the same way whether a mental state is a thought, belief, desire etc. depends not on what it is that occupies the functional role, rather it depends exclusively on its relation to the bigger system of other mental states, sensory input and behaviour. In this way the functional identification is neutral to what kinds of states that occupy the role. As a result, mental states can be realized by many different kinds of states. This is what is called multiple realization. Different versions of functionalism have been developed since its first appearance in the 1960’s. Over time it has become a general approach to distinguish between commonsense functionalism and empirical functionalism. I will elaborate on this distinction and on functionalism in general in the first chapter: Functionalism in detail. Embodied Cognition as Undermining Functionalism The second chapter will concern Shapiro’s argument that functionalism is undermined by research within the area of embodied cognition. According to Shapiro, it is in the spirit of functionalism to characterise mind in a way that abstracts away from the hardware (the body), on which the mind is implemented. The result is that the body is neutral in the sense that “characteristics of bodies make no difference to the kind of mind one possesses.”(Shapiro (2004) p. 175.) This kind of body neutrality implies that similar minds can exist in bodies with very distinct properties. However, this assumption, according to Shapiro, is shown to be false in the light of recent -3- research in the area of embodied cognitive science - research that shows that a creature’s embodiment imposes constraints on what kind of psychological profile it can have. Shapiro takes his point of departure by distinguishing between three different lines of research that all, according to Shapiro, show that characteristics of bodies make a difference to the kind of mind one possesses. The first area of research is concerned with the body’s role in perception and the sense in which characteristics and specific actions of the body play a significant role in determining the nature of thought. Shapiro gives an example of this kind of thought concerning auditory perception. According to Shapiro, our ability to locate a sound source depends on facts about our ears. One such fact is that we have exactly two ears and not one or more than two. Another fact is that these ears are placed on both side of a head of a certain size. One result is that there will be a certain time difference between the sound received at each ear. To locate the source of a sound the human auditory system incorporates these facts in its processing. According to Shapiro, different setups of auditory sensory systems yield different setups of the processing involved in locating a sound source. Because the body, in this way, makes a “difference to how an organism hears”, Shapiro suggests, that the body cannot be neutral to the kind of mind an organism can have (Shapiro (2004) p. 189). The second direction that Shapiro refers to within the embedded, embodied research program concerns the way our concepts and the way in which we conceive of the world depend on details of our embodiment. The central idea is that organisms with different kinds of bodies “will conceive the world differently and will think differently.” (Shapiro (2004) p.183) As an example Shapiro refers to the work of George Lakoff and Mark Johnson on the role of -4- the body in structuring metaphors. The central idea of this work is that the type of our body plays a significant role in structuring our basic concepts and that these basic concepts in turn structure more abstract concepts and metaphors. According to this research, basic concepts like ‘front’, ‘back’, ‘up and ‘down’ are directly related to the shape and type of our bodies. Shapiro takes this to support the claim that the content of mental states and thereby the perspective that one has upon the world is dependent on the type of body that one has. The third direction within the embodied cognition research program, which Shapiro takes to undermine functionalism, is what he calls extended mind.1 The main feature of this view is that the mind and the body are so intimately tied together that it does not make sense to maintain the division between the two. Shapiro argues that if we give up this division then, naturally, difference in bodies will entail differences in mind. (Shapiro (2004) p.183) The work cited above shows, according to Shapiro, that a difference in body— even a difference in the fine details of embodiment – makes a difference in mind. If the body is involved in our psychological processes in this intimate way, this indicates, according to Shapiro, that it requires a humanlike body to have humanlike psychological capacities. I will call this line of embodied approach: the Body-detail model. According to this model, it is not enough to recognise that the mind is intimately connected to a body; we also have to specify the characteristics of this body. In the sense that Shapiro takes it as being in the spirit of functionalism to abstract away from exactly these facts, the body-detail model of embedded, embodied cognition undermines functionalism, according to Shapiro. 1 This label originates from Clark and Chalmers (1997). However, Clark and Chalmers intended the label for the view that the realization of the mind is extended not only to the body, but also to the physical and social world that surrounds us. -5- To get a better understanding of Shapiro’s claim it is useful to notice that Shapiro uses the term ‘body’ in a slightly different way to which it is commonly used. Normally when we talk about bodies the term refers to the entire collection of all body parts such as feet, muscles, entrails etc., including the brain and central nervous system. Shapiro’s use of the term body is intended to exclude the brain and the central nervous system (Shapiro (2004) pp. 70)2. This use of the term body is helpful when we are dealing with embodied and embedded approaches to mind, because these views take themselves to be opposing views that take the brain as the sole substrate of mind. What proponents of embedded, embodied approaches want to emphasise is exactly how our bodies and environment – as opposed to only our brain— are intimately linked to cognition. Because of this advantage I will follow Shapiro’s use of the term ‘body’ throughout this thesis. Embodied and Embedded Cognition as Extended Functionalism Contrary to proponents of the Body-detail model, Clark argues that there is a way in which the embodied approach can be seen as a form of extended functionalism. This will be the focus of chapter 3. On this line of thought, the body’s contribution is just a part of a bigger system consisting of body, brain and environment that as a whole sometimes implements the functional profile that constitutes mind and cognition. Whereas traditional functionalism analyses input, output and states internal to the individual, extended functionalism includes non-biological parts of the environment as what might be called the 'material realizers' of the mentality-fixing functional roles. 2 Shapiro (2004) treats the question about the relation between mind and body as distinct from the question of the relation between mind and brain (op.sit. 71) -6- One of the examples of such an extended functionalist approach could, according to Clark (In Press), be the following: in the process of solving some kinds of accounting problems, an accountant, call her Ada, uses a method of rapidly scanning columns on one page and copying certain numbers to a piece of paper and then shifting her attention between the two papers. Using this strategy, involving external props and the repeated shifting from one page to the other, the task is cut up into smaller and less demanding tasks. In this way the workload on Ada’s memory is significantly reduced. Clark suggests that this process is best analysed in “extended functionalist terms as a set of problem-solving state transitions whose implementation happens to involve a distributed combination of biological memory, motor actions, external symbolic storage and just-in-time perceptual access.” (Clark (In Press) p.12) Another embodied approach that, according to Clark, can be interpreted as a form of extended functionalism is the work of Ballard et al (1997). Their work is based on experiments in which subjects were given the task of copying a certain structure of coloured blocks. The subjects were placed in front of a computer screen with three clearly distinguishable zones: one area containing the original structure; a reserve area with blocks in different colours that could be picked up by clicking on the mouse, and an empty area where a model of the original structure could be built. Ballard et al found that the subjects solved this problem with the help of repeated rapid saccades to the original - fixating only on the part of the original necessary for the completion of the next part of the process: for example, picking up a block with a certain colour. According to Ballard et al, this suggests that the subjects were only storing smaller pieces of information about the colour or the position of the block, taking advantage of the fact that the information could always be retrieved by fixating on the original. Clark suggests that we should view this -7- as an extended functionalist approach because Ballard et al’s theory “analyses a cognitive task as a sequence of less intelligent sub-tasks, using recognizable computational and information-processing concepts, but applies those concepts within a lager organizational whole.” (Clark (in press). p. 11) By acknowledging the body’s contribution as in this way a part of a bigger system, extended functionalism leaves room for the possibility that creatures with different kinds of bodies can share similar minds. Where do we go from here? An important task is to investigate how well defined the body-detail model and the extended functionalist approach are as positions. Further, how different are they and what is the underlying difference that makes them come to apparently opposite conclusions about the relations between the embodied, embedded approaches and functionalism? Is there a possibility that the two positions could be reconciled? One possibility might be to formulate functional roles in terms of specific characteristics of the body. That is, to formulate a functional role so that only creatures with certain bodies and sensory systems will be able to play that role. This possibility is also considered by both Clark and Shapiro (Clark (In Press) p.16) (Shapiro (2004) p. 174). Would this be too much embodiment for the functionalist to admit? These questions will concerns trough the thesis, however, an obvious first step on our way to settle the question of the relation between the embodied, embedded approaches and functionalism is to explore the position of functionalism in further detail. This will be the task of the following chapter. -8- ______________________________________________ CHAPTER 1 FUNCTIONALISM IN DETAIL ______________________________________________ In this chapter I will survey a chart of the complex landscape of functionalism. Firstly, looking into the background and origin of the position will provide an insight into the intentions behind the position. This will also provide us with an idea of what ‘the sprit’ of functionalism might be considered to be. Secondly, I will investigate the notion of functional role and the models for giving such presented by proponents of functionalism. These models will provide the background for understanding the difference between the two dominant versions of functionalism: commonsense functionalism and empirical functionalism. Finally, I will discuss whether functionalism in its different versions can avoid the pitfalls of both liberalism (attributing mental states to systems that do not have them) and chauvinism (falsely denying mental states to systems that in fact have them)3. A special focus in this connection will be the relation between functionalism and multiple realization. 1.1 The Background and the Spirit In order to get a feeling of what functionalism is and to illuminate what ‘the spirit of’ functionalism might be, it will be fruitful to look into how and on 3 ‘Liberalism’ and ‘chauvinism’ as used by Block 1980. -9- what foundations this theory was developed in the 1960’s. One major source of inspiration available at this time was the first hesitant beginnings of computer science and its early success. In particular, the distinction between software and hardware, between function and structure inspired the distinction between role and occupant found in functionalism. The relation between software and hardware was viewed as a particularly apt model for the relation between mind and body. An important source of inspiration from computer science was the invention by Alan Turing of abstract and theoretical machines defined by functions characterizing the relations between input, output and internal states of the machine. I will return to this issue in my exposition of Hillary Putnam’s early method for defining the functional role of mental terms. But first let us have a look at the philosophical background which functionalism can be seen as a reaction to. In this context functionalism can be seen as having developed out of dissatisfaction with the solutions provided by the dominating theories of mind at the time: Behaviourism and The Mind-Body Identity Theory. 1.1.1 Behaviourism Behaviourism, as a theory of mind, is the thesis that mental states can be defined in terms of relations between the input of stimuli and the output of physical behaviour. If, for example, I have a desire to drink a glass of water, this desire can be defined as having a set of dispositions to act in a certain way - one of these being the disposition to reach out for the glass, if I get the opportunity. Notice that this is a simpler approach than the one presented by functionalism where internal mental states are taken into consideration as well. According to proponents of functionalism, the problem with behaviourism is that it seems possible to think of examples where something has the right kind of behavioural dispositions to the right kind of stimuli, but without having the right kind of mental state linked with these. Think of a - 10 - creature, maybe extraterrestrial, who displays the particular behavioural dispositions of having pain when receiving stimuli of tissue damage, but who is only pretending to be in pain – acting as if in pain just to fool us earthlings (for examples of this sort see Putnam (1963)). Behaviourism would falsely attribute the mental property of having pain to such an actor. Accordingly, behaviourism is considered by functionalists and by the general philosophical community today to be too liberal in its ascription of mental properties to creatures and systems. In contrast, a functionalist account would not take the actor to be in pain. The reason being that such a creature would lack the proper internal relations, for example, it will not tend to believe that there is something wrong with its body. 1.1.2 Mind-body Identity Theory In contrast to behaviourism, the mind-body identity theory is considered by functionalists to be too chauvinistic in its ascription of mental states. The identity theory claims that in the same way as water is identical to H 2O and temperature identical to mean molecular kinetic energy, so are mental states identical to physical states. A very simplified although much used example is to identify the mental state pain with C-fibre stimulation in the brain.4 Identifying a mental state with a brain state, however, makes the theory vulnerable to the accusation of being chauvinistic (i.e. falsely denying systems mental properties). This has the result that only creatures with brains can be attributed mental states. In this light traditional research within AI, where the goal is to develop an artificial system with humanlike mental states, seems to be a non-starter. Further, this chauvinism may not be limited to creatures with brains but perhaps to humans only. If it is only humans who have C-fibre stimulation, only humans can feel pain. 4 This is of cause a very simplified identification and the expression C-fibre stimulation is only to be taken as a stand-in for a more complex brain state. - 11 - The challenge for the identity theorist, according to Putnam, is the following: “He has to specify a physical-chemical state such that any organism (not just a mammal) is in pain if and only if (a) it posses a brain of a suitable physical-chemical structure; and (b) its brain is in that physical-chemical state. This means that the physical-chemical state in question must be possible of a mammalian brain, a reptilian brain, mollusc’s brain (octopuses are mollusca, and certainly feel pain), etc. …. the brain state theorist [the mind-body identity theorist] is not just saying that pain is a brain state; he is, of course, concerned to maintain that every psychological state is a brain state. Thus if we can find even one psychological predicate which can clearly be applied to both a mammal and an octopus (say ‘hungry’), but whose physical-chemical ‘correlate’ is different in the two cases, the brain state theory has collapsed. It seems to me overwhelmingly probable that we can do this.” (Putnam (1967), p.436 (Reprint version)) In addition to the problems of finding a similar state within different species there seem to be problems identifying certain mental states with particular kinds of brain states, even within our own species. Research in the field of neuropsychology has shown that the same psychological process can be associated with distinct neurological structures. Examples are people who because of brain damage have developed speech centres in the right hemisphere (normally situated in the left side) or cases where the auditory cortex is used to process visual information. (Block and Fodor, (1972) and Shapiro (2004) p. 28 citing research of von Melchner, Pallas and Sur (2000)) Taking these as different realisations show that mind, or at least some mental - 12 - capacities, can be multiply realised. The mind-body identity theory, according to functionalists, has a problem accounting for this fact. Because functionalism can accommodate multiple realization, proponents of functionalism have argued, it is prima facie preferable to the mind-body identity theory as a theory about the nature of mental states. 1.1.3 Materialism What functionalists have in mind when they speak of multiple realization is that a mental kind can be realized by distinct physical kinds. In this sense, Functionalism is materialistic in spirit. Materialism is the view that, fundamentally, all things that exist are 'material' or physical. Arguing that the kind of states, which are most likely to be able to carry out the functional roles are physical states, David Lewis (1966) presents his functionalist account as an argument against dualism. However, strictly speaking, materialism is not implied by functionalism. In fact, functionalism is compatible with some kinds of dualism. Again, this is because the functional descriptions involved are neutral about the nature of the occupier of the role (This neutrality will be explained further in section 1.2). The only kinds of dualism incompatible with functionalism are positions like epiphenomenalism and parallelism that deny causal interactions from the mental to the physical. Properties lacking ability to affect the physical domain will not be suitable candidates for occupying a functional role. 1.1.4 Anti-reduction and methodological autonomy Characterizing the spirit of functionalism it is also important to note that functionalism has been seen by many as being able to provide a materialistic account that is non-reductive. The general line of argument is that because psychological kinds can be realized in diverse physical kinds (multiple realization) psychological kinds cannot be reduced to physical kinds. This has - 13 - led functionalist like Jerry Fodor (1974) to the conclusion that regularities at the psychological level, although implemented on a physical level, cannot be reduced to regularities on this underlying level. In this light, endorsing functionalism can be seen as a way to ‘save’ the methodological autonomy of psychology. Accordingly, psychology has its own level of description that is above the details of neuroanatomy, physiology, chemistry and physics. Psychology uncovers special laws which cannot be deduced from and thus not reduced to the laws of physics. 1.1.5 The spirit of functionalism In the light of the dissatisfaction with the mind-body identity theory and behaviourism, functionalism can be seen as an attempt to provide a materialistic and mechanistically inspired account of the mental that promises the methodological autonomy of psychology and that at the same time avoids the pitfalls of both liberalism and chauvinism. This seems to sum up the intentions behind functionalism and thus gives us a picture of what can be considered to be the spirit of functionalism. Now, we can begin to understand Shapiro’s claim that it is in the spirit of functionalism to regard the mind as “a programme that can be characterized in abstraction from the brain/body that realizes it” (Shapiro (2004) p. 175). Before we discuss whether this is necessarily a consequence of functionalism I will say a bit more about how we should understand functionalism and functional roles, and present some different versions of the position. 1.2 Methods for defining functional roles Over the years various different versions of functionalism have been developed. The best way to get an overview of this is to start by looking at the two main methods for defining mental states in a functionalist way. The - 14 - first method is set in terms of Turing Machines and machine tables. The second is Lewis’ method for defining terms using Ramsey sentences. 1.2.1 Turing Machines and Machine tables As noted, functionalism, in its early days was inspired by work within computer science. Hillary Putnam (1960 and 1967) introduced the idea of modelling a functional theory of mind over Turing Machines. A Turing Machine is a purely theoretical and abstract construction defined by two different functions. Function 1: From input and states to output. Function 2: From input and states to other states. A machine of this kind can be characterised in a machine table that, for every possible combination of states and input, lists the instructions for these functions. These instructions have form of conditionals of the following kind: If the machine is in state S and receive input I then it emits output O. A coke machine is an example of an ordinary machine that can be described this way: Coke Machine 50 pence input 1 pound input S1 S2 Emit no output Emit a Coke Go to S2 Go to S1 Emit a Coke Emit a Coke and 50 pence Go to S1 Fig: 1 Machine table of coke machine (UK version of example from Block 1980) This machine table specifies that if the coke machine is in S1 and receives an input of 50 pence the machine emits no output and goes to state S2. If the machine, while in S2, receives another 50 pence the machine will emit a coke and return to state S1 and so on. I will not go through the whole table but I - 15 - think that the idea should be clear from these conditionals. One thing to note is that the system’s response to input depends on its internal state when receiving the input. Putnam’s idea was that organisms with mental states can be described analogues to Turing machines.5 Hence, a mental state is equivalent to a Turing machine state. In this connection, it is important to notice that the states in a Turing machine are functionally defined by the way they are connected to each other in the way that the machine table describes. When we described the coke machine we did not have to say anything about what materials the machine was made of, only the relations between the internal states and the in- and output. Neither did we specify anything about the construction of the internal state of this machine. What it is to be in these internal states is specified implicitly in the relations between the states and the production of output from input as it is shown in the machine table. Any system that has internal states that interact in this way will have such states as S1 and S2. Putnam points out that an important feature of this account is that the physical realization of the sense organs responsible for the various inputs, and of the motor organs is specified, but that the ‘states’ themselves are specified only implicitly by the machine table (Putnam (1967)). This has the feature that the theory says nothing about the nature of the occupier of the role. The occupier could, in theory, be brain-states, silicon states, physical or even non-physical states. Like software the mind can be run on many different kinds of hardware. 5 In his article from 1967 Putnam models his functionalism over Probabilistic Automatons instead of Turing Machines. In Probabilistic Automatons transitions between states are defined in probabilities rather than being deterministic. - 16 - A problem with Putnam’s model, however, is that a machine state is a state of the whole system and consequently the model has difficulties accounting for systems that are in several internal states at the same time. (Levin 2004) 1.2.2 Lewis’ Ramsey-sentence method of funtionalization Another way to see how functional definitions are supposed to work is looking at Lewis’ (1972) method for providing functional definitions using Ramsey sentences. This method has the advantage of being able to account for systems with several simultaneous internal states. One could say that Lewis’ method defines a system’s mental state all at once. (Levin 2004) Take the example of defining pain as a mental state. In this case we have a theory of mental states (an extremely simplified one) where pain is the state which tends to be caused by tissue damage, causes worry and the emission of “Ouch“. Worry, in turn, causes brow wrinkling (Example from Block 1980). If we replace the mental state terms pain and worry with variables (x and y) and transform the theory into an existentially quantified sentence we get what is called the Ramsay sentence of the theory. x, y (x is caused by tissue damage & causes the emission of “Ouch” & causes state y & y causes brow wrinkling.) This reads: there are two states x and y such that x is caused by tissue damage and causes both the emission of “Ouch” and state y, which in turn causes brow wrinkling. Now we can define what it is to be in pain and worry in the following way: - 17 - Pain An organism or system S is in pain = there are two states x and y, x is caused by tissue damage and causes both the emission of “Ouch”, and the state y, and the state y causes brow wrinkling and S is in state x. Worry An organism or system S is worried = there are two states x and y, x is caused by tissue damage and causes both the emission of “Ouch” and the state y, and the state y causes brow wrinkling and S is in state y. Notice that in the above definitions the mental state terms are replaced with variables, while input and output remain specified. Again this says nothing about the structure of the involved mental states. Anything with an internal state that plays these causal roles is worried and in pain. 1.3 Commonsense vs. Empirical functionalism Having the methods for defining the functional roles in place the next concern is which kind of theory we should use as the foundation for our definition of these functional roles. There are two main directions that functionalism has taken on this question. So-called commonsense functionalists take an a priori approach, while empirical functionalists hold that we should take a posteriori empirical psychological theories as our point of departure. In what follows I will give a more detailed exposition of both these directions. 1.3.1 Commonsense functionalism According to commonsense functionalists (e.g. Lewis (1972) and Armstrong (1968)) our guideline for defining the appropriate functional roles for our mental states is our common knowledge about the relations between - 18 - particular input, internal causal interaction and output. This knowledge provides the background for what could be called a commonsense psychology. Lewis (1972) suggests that we think of "common-sense psychology as a term-introducing scientific theory, though one invented before there was any such institution as professional science." (Lewis (1972). p. 256) According to Lewis we construe this theory out of our extensive repertoire of platitudes about mental states — as for example, 'headache is a kind of pain'— and other platitudes about the causal relations of mental states, sensory stimuli, and motor responses (Loc cit). We have already seen pain analysed in such a commonsensical way. It is common knowledge that pain is caused by tissue damage or damage to the body; it is also general knowledge that this creates a desire to escape that situation and so on. This is similar to the way that we define ordinary terms like mousetrap, alarm clock or ball pen in broad functional terms. A mousetrap is something that has the functional role of catching mice; an alarm clock has the functional role of keeping and showing the time and making a loud noise at the right time and so on. The functional role of an alarm clock gives us the meaning of what it is to be an alarm clock. If something keeps and shows the time and makes a loud noise at a certain set time it is an alarm clock. Likewise, an a priori analysis of pain finds the common sense functional role associated with pain and this role gives us the meaning of the term ‘pain’. 1.3.2 Empirical functionalism This variety of functionalism holds that, instead of being guided by commonsense we should be guided by the findings of the science of empirical psychology (e.g. Putnam, Fodor). A typical way to express this view is that mental states are occupants of functional roles and that which functional role determines which mental states that a subject is in, is a matter for science (perhaps cognitive psychology or neuroscience) (Braddon-Mitchell and - 19 - Jackson (2007)). This yields a different kind of specification of input and output than that which is open to commonsense functionalism. Commonsense functionalism is restricted to specifying input and output in terms of what is externally observable and common knowledge (e.g. inputs that are characterised in terms of objects present close to the subject, and outputs in terms of bodily movements). In contrast empirical functionalism can use the knowledge that we acquire via experiments e.g. knowledge of neural signalling. (Block (1980)) The typical description of empirical functionalism replaces the commonsense functional role directly with the empirical functional role. However, as Braddon-Mitchell and Jackson (2007) points out, it seems that the commonsense functional role must enter into the picture at some point. This is because having only the neuroscientific facts about a person would not suffice to give ordinary people with no knowledge of neuroscience any knowledge about what that person thinks. They suggest that the commonsense functional role, instead of having the role of fixing the meaning of the mental terms, as it has according to commonsense functionalism, employs the function of fixing the reference of these terms. Braddon-Mitchell and Jackson distinguish between two general versions where the commonsense role fixes the reference. In the first, the commonsense role fixes the nature of the state that plays the role, e.g. on neurophysiological states. In the second version the commonsense role fixes another more detailed role that underpins the commonsense role. In this case “What settles that a subject is in M [a certain mental state] is the internal functional role which underpins the common sense functional role associated with M.” (Braddon-Mitchell and Jackson (2007) P.85) The interesting role for science becomes to investigate which internal functional roles play the - 20 - common sense functional roles. This latter version is the more sensible way to go, according to Braddon-Mitchell and Jackson. This becomes obvious when we realize that the first version contradicts the way in which we defined multiply realization. If the functional role refers directly to the neural state we end up with chauvinism, similar in kind to that which the mind-body identity theory has been accused of. However, as we shall see in the next paragraph, even the more promising versions of empirical functionalism have been accused of being chauvinistic. 1.4 The liberalism-chauvinism dilemma Now that we have an overview of what functionalism is and different versions of it, it is time to ask ourselves how this picture fits with the picture that we drew of the spirit of functionalism, just a little while ago. Remember that one of the essential reasons for endorsing functionalism was to avoid the chauvinism of the mind-body identity theory. We have already seen that some versions of empirical functionalism seem to have problems avoiding chauvinism and providing us with the right kind of multiply realization. In his article “Troubles with Functionalism” Block argues that the dilemma for functionalism is “that any physical description of inputs and outputs ….yields a version of functionalism that is inevitably chauvinist or liberal” (Block (1980) p. 295). Let us take a look at why he thinks that this is the case. 1.4.1 The charge of chauvinism We’ll start by considering Block’s challenge to commonsense functionalism. Block observes that commonsense functionalists “tend to specify inputs and outputs in the manner of behaviourists: outputs in terms of movements of arms and legs, - 21 - sound emitted and the like; inputs in terms of light and sound falling on the eyes and ears….Such descriptions are blatantly species-specific. Humans have arms and legs, but snakes do notand whether or not snakes have mentality, one can easily imagine snakelike creatures that do. Indeed, one can imagine creatures with all manner of input-output devices, e.g., creatures that communicate and manipulate by emitting strong magnetic fields “(Block (1980) p. 294). On this account Block holds that the general version of commonsense functionalism leads to a kind of chauvinism: human body chauvinism. In this sense, Block is suggesting that because the identification of mental states is limited by the fact that the input and outputs are species-specific, we do not get the kind of multiple realization the functionalists were originally looking for. Turning to the question of empirical functionalism Block’s charge of chauvinism is a charge of internal organization chauvinism. Remember that empirical functionalism is concerned with the functional organization of the internal cognitive mechanisms and processes as are the objectives of empirical psychology such as input and outputs of the central nervous system described in neurophysiological terms (Braddon-Mitchell and Jackson (2007)). Now, Block invites us to imagine that we meet some Martians and find that in a commonsense functional way they are very much like us. However, in the empirical functional way we differ significantly. “When we get to know Martians, we find them about as different from us as humans we know. We develop extensive cultural and commercial intercourse with them. We study each - 22 - other's science and philosophy journals, go to each other's movies, read each other's novels, etc. Then Martian and Earthian psychologists compare notes, only to find that in underlying psychology, Martians and Earthians are very different… Now imagine that what Martian and Earthian psychologists find when they compare notes is that Martians and Earthians differ as if they were the end products of maximally different design choices (compatible with rough Functional equivalence in adults). Should we reject our assumption that Martians can enjoy our films, believe their own apparent scientific results, etc.?…..[T]here may be many ways of filling in the description of the Martian-human differences in which it would be reasonable to suppose there simply is no fact of the matter, or even to suppose that the Martians do not deserve mental ascriptions. But surely there are many ways of filling in the description of the Martian-Earthian difference I sketched on which it would be perfectly clear that even if Martians behave differently from us on subtle psychological experiments, they nonetheless think desire enjoy, etc. To suppose otherwise would be crude human chauvinism.” (Block (1980) pp. 291-93) The consequence of identifying mental states on the basis of theories of neural inputs and outputs as we humans have them seem to be that only creatures with neurons or with the right kind of neural set up can have mental states like ours. This is what you might call a neurophysiological chauvinism (term from Braddon-Mitchel and Jackson (2007)). - 23 - 1.4.2 Functionalism and abstraction One way for both commonsense and empirical functionalists to avoid the charge of chauvinism is to characterise inputs and outputs in more abstract ways. If, in our example with the coke machine instead of cokes, 50 cent coins and 1 pound coins, we had characterized these as output 1, Input 1 and Input 2, then every machine that had some inputs and outputs standing in the same functional relation as that of our coke machine will have the same internal states as this (Braddon-Mitchell and Jackson (2007)). In regard to mental states, any creature or system which had inputs and outputs and similar internal functional relations would be functionally isomorphic to you and have the same mental states as you. This would guarantee that systems without brains or with different bodies are not excluded from the realm of psychological subjects from the outset. That is, it would ensure the possibility of the kind of multiple realization with which Putnam and others were concerned in arguing against the mind-body identity theory. However, according to Block, the problem with such a characterisation is that it is too liberal. Block invites us to imagine that we set up all people in China in a network linked via radio communication that communicates with a body somewhere in the world. This system is set up in such a way that it is commonsense functionally equivalent to a human brain and body. Would such a system have mental states? Commonsense functionalists would have to say that it has. In the case of empirical functionalism, if someone arranged the economy of Bolivia in such a way that it had similar relations of internal states, input and output as a human being, we would have to say that it has mental states (Block 1980). Block argues that since these would be absurd assumptions, these versions of functionalism would be untenable. Block holds that it is our intuition that this kind of system would lack qualitative mental states. That is, there would not be something “it is like” to be these systems. - 24 - However, one might object that what is involved in the example seems to be outside the reach of our intuitions. It is possible that if we really understood the complex functional story of the Bolivian economy when organized this way or the China brain, the thought that it has mental states might not be absurd. At least it seems reasonable to say that because of our lack in knowledge of this kind our intuitions in this case can only be very vague. Bechtel and Mundale (1999) present an alternative option for functionalists to avoid the pitfall of chauvinism. I will give an exposition of this in a short while; however there is one thing that we need to shed light on first and that is the relation between functionalism and multiple realization. 1.4.3 Functionalism and Multiple Realization Often multiple realization is mentioned in the same breath as functionalism, and as we have seen multiple realization plays a major role in the general debate concerning chauvinism versus liberalism. But what exactly is the logical relation between these two notions? Does functionalism entail multiple realization? And does multiple realization entail functionalism? Let us start with the second question. The short answer is: No. It is possible for something to be multiple realizable even though it is not individuated in terms of functional role. Standard examples are geometrical kinds like square, cube, sphere and circle (Mandik (2007)). What makes something a cube is that it is a hexahedron with six equal squares as faces. This is an intrinsic property. Something can be a cube without having any causal relations to anything else. A cube is a geometrical, not a functional kind. Further, cubes can have different physical properties, be realized in different materials like foam, stone or wood. It can also have different sizes of squares, be solid or hollow - 25 - and have different masses. In this way cube seems to be multiply realizable. If we had an account on which mental states were individuated in terms of intrinsic properties, these could still be multiple realized. It may be that multiple realization of mental states, if true, gives us a good reason to prefer functionalism to mind-body identity theory, but it does not support functionalism definitively. Even more interesting for our current discussion is the second question of whether we can have functionalism without multiple realization. The idea of functional kinds being multiple realizable seems to be general. Take for example the functional kind of mouse trap. Mouse traps come in al sorts of shapes, materials and make ups. The same seem to be the case with corkscrew, watch and left winger. Taking mental kinds to be functional kinds in a similar way, functionalists have argued that mental states like pain or hunger can likewise be realized in different physical structures e.g. in brain states in mammals, electronic states in extremely complex (and not yet developed) computers or in some kind of plasma in extraterrestrial creatures. However, even though multiple realizability seems to be a property shared by many functional kinds, this does not necessarily show that all functional kinds are multiple realizable. According to Shapiro, “adoption of a functional perspective toward the mind does not entail that the mind is multiply realizable.” ((Shapiro 2004), p. 22). This is the case, according to Shapiro, because it is possible to have a description of a functional role that applies only, as a matter of necessity, to a single physical kind. One of his examples involves the drill bit of a machine designed to drill in extremely hard minerals. Drill bit is an example of an ordinary functional kind. What is significant to being of the kind drill bit is to have the capacity to drill a hole when rotated in to a surface. However, Shapiro invites us to consider that “if diamond are the only substance that in fact are hard enough - 26 - to drill through very hard surfaces, then drill bit picks out a physical kind no less than it refers to a functional kind”(Shapiro (2004) p.21). Notice, however, that whether or not diamonds, in this case, are the only substance hard enough for the drilling job is an empirical question. Thus, it concerns the nomological possibility of drill bit being multiple realizable. This is distinct from the question of the metaphysical possibility of drill bit being multiple realizable. Even if diamonds turns out to be the only substance that can realize the special kind of drill bit in our world and in all the possible worlds that share our laws of nature, this does not exclude there being other possible worlds where other physical kinds can do the job as well. The fact that the functionalist description of the functional role is neutral, at least to some extent, about the nature of the realization, seem to ensure the metaphysical possibility of a kind being multiple realizable. However, as we have seen Shapiro argue, this is not the case when we are concerned with the nomological possibility. The situation is similar when we consider the question in relation to mental states. To sum up: a functionalist approach does entail multiple realization when we are concerned with the metaphysical possibility; however, this is not the case when the question comes to the nomological possibility of multiple realization. It is the nomological possibility of multiple realization that concerns Bechtel and Mundale in the article “Multiple Realizability Revisited” (1999) In this article they develop a suggestion as to how we can have a functionalist approach that does not involve multiple realization or entail chauvinism. - 27 - 1.5 Functionalism without multiple realization and chauvinism? We have now seen that in a certain sense functionalism without multiple realization is a possibility. As we can formulate a functional role for drill bit that exclusively can be occupied by diamonds, so we could formulate a functional role for pain that could only be occupied by humans. However, in our previous discussion of the liberalism-chauvinism dilemma we saw that this seemed to steer us straight into chauvinism. In the article “Multiple Realizability Revisited”, (1999) Bechtel and Mundale suggest, a way forward for denying multiple realization of mental states, but nevertheless keep a functionalist approach that has the advantage of not being chauvinistic. The key feature of this account is their argument discrediting the claim that mental states are multiple realizable in the way suggested by functionalism. An exposition of this argument will essentially provide an insight in how they think that it is possible to reject multiple realization and at the same time avoid chauvinism. According to Bechtel and Mundale the prima facie case for multiple realization stems from a misguided understanding of what a brain state is. When Putnam advances his argument against the mind-body identity theory his main claim is that it seems implausible that we would find the same physical-chemical state in the all different species that share the ability to have psychological states like pain or hunger (see above quote p. 12). The conception of a brain state as physical-chemical states has been taken as a general point of departure for most philosophical discussions of multiple realization. However, according to Bechtel and Mundale this notion of brain state is not shared by researchers in the neurosciences. The closest we get to something that could count as brain states in the neurosciences, are areas of - 28 - activity in different parts of the brain. This is a notion of brain state that is much more abstract and coarse-grained than the one found in the philosophical arguments. Bechtel and Mundale accuse the philosophers for failing to note that “the same degree of variability is tolerated by neuroscientists in identifying types of neural processes that they accept in identifying psychological types” (Bechtel and Mundale (1999) p. 202) Bechtel and Mundale point to the fact that it is only on a very abstract level that humans and octopuses can be considered to share a psychological state like hunger. In this case, hunger could be characterized as something that is caused by lack of food, causes discomfort and food-seeking behaviour and so on. However, if we look at hunger in these different species with the aim of asserting the individual differences, we see that there are differences in, for example, what kind of food that is sought, how it is done etc. The usual practice, in identifying psychological states, involves abstracting away from such details. On similar lines neuroscientists abstract away from details on physical and chemical levels. In fact, neuroscience is more functionalistic than it has traditionally been thought to be, according to Bechtel and Mundale. They point to the fact that the neurosciences both in their tools and project rely heavily on considerations of behaviour and psychological function. This has the result that it is possible to identify brain states across species and thereby, according to Bechtel and Mundale, the consequence of making multiple realization less likely. To give a simple example of this approach in neuroscience Bechtel and Mundale refer to Ferrier’s (1886) comparative investigations based on weak current stimulation of brain areas of, among others, dogs, rats, rabbits and - 29 - monkeys. Figure 2 shows the areas on the left hemispheres of monkey (left) and dog (right) where motor response was educed when stimulated. Fig. 2 Illustration from Ferrier 1886, reprinted in Bechtel and Mundale 1999 Each number refers to a different kind of motor response e.g. Number 1: Opposite hind limb is advanced as in walking, Number 5: Extension forward of the opposite arm (as if reaching or touching something in front) Although Ferrier’s research is of older date it is illustrative of the approach that is also present neurophysiology. in Although today’s research contemporary in neuroanatomy research employs and further developed techniques such as single cell recording, it shares the point of departure in the assumption that we can compare across species and its dependence on psychological or behavioural function. When it comes to research on humans techniques such as position emission tomography (PET) and functional magnetic resonance imaging (fMRI) has been developed. Here the dependence on psychological function is even more evident, because it relies on the cognitive tasks that the subjects are asked to perform in the experiments while scanned. - 30 - According to Bechtel and Mundale, what we should learn from all this is that the way we ask questions in the neurosciences is constructed in a way that makes multiple realization less likely. “…neuroscientists employ behaviour and psychological function in setting the context for identity of brain states, with the result that they identify areas and processes in brains across species, as well as in different brains within the same species. Not surprisingly, psychologists employ similar contexts for identifying psychological states, resulting in a one-to-one mapping between brain states and psychological states” (Bechtel and Mundale (1999). p.203) In short, functional criteria are relevant in identifying both psychological states and brain states, according to Bechtel and Mundale. Multiple realization of psychological states is about judging similarity and difference: firstly, similarity of psychological states and secondly, difference of physical realizations. The standard arguments for multiple realization have used a coarse-grained criterion when judging the sameness of psychological states and a fine-grained criterion when discriminating between brain states. Bechtel and Mundale’s claim is that if we choose the same grains of analysis for both psychological and neuroscientific investigations then the plausibility of multiple realization disappears. “One can adopt a relatively coarse-grain, equating psychological states over different individuals or across species. If one employs the same grain, though, one will equate activities in brain areas across species, and one-to-one mapping is preserved…Conversely, one can adopt a very fine-grain, and differentiate psychological states between - 31 - individuals, or even in the same individuals over time. If one similarly adopts a fine-grain in analysing the brain, then one is likely to map psychological differences onto brain differences, and brain differences onto psychological differences.” (Bechtel and Mundale (1999) p. 202) Which level of grain that is appropriate in judging these similarities and differences depends on the context of inquiry. If, for example, one takes an evolutionary perspective a coarse-grained analysis would be appropriate. (Bechtel and Mundale (1999) p. 203) However, according to Bechtel and Mundale, on neither of these different levels of abstraction is it likely that psychological states are multiply realized. On the coarse-grained approach realizations in form of brain states will be similar across species and on the fine-grained approach we do not share psychological states. Finally, how does this approach avoid chauvinism? Taking a coarse-grained approach to both psychological states and brain states makes it possible to ascribe psychological states to cats, dogs, rat, rabbits and possibly many other species and not only to humans. But what about computers and extraterrestrials composed of different types of components? Well, we might find that these things are alternative realizations, but according to Bechtel and Mundale, this is not obvious and maybe it is not even likely. Computers that have psychological abilities are likely to be very different from the computers that have been invented up till now. Bechtel and Mundale suggest that it at least seems possible that such computers would have areas that could be identified as being responsible for processing different sensory inputs or motor output. According to Bechtel and Mundale, ”this might provide a basis for a common taxonomy of the physical processing states underlying the psychological function” (Bechtel and Mundale (1999) p. 204). The situation I imagine would be similar in the case of extraterrestrials. - 32 - At this point one could ask whether this is not exactly what we took as standard cases of multiple realization? That is, for example, realizations in silicon-based systems. An alternative way of reading Bechtel and Mundale’s point about the notion of brain states in the neurosciences is that this notion in itself is a functional notion that actually permits multiple realization in the standard terms. So the important issue here is perhaps not, as we first thought, about whether or not the account discredits the plausibility of multiple realization. The really important thing to recognize is that neuroscience is more functionalistic in its outset than first expected and that the central issue is about choosing level of grain for one’s analysis appropriate for the context of inquiry. 1.6 Sum up and Conclusions My aim in this chapter has been to provide the tools for evaluating whether an embedded and embodied approach is indeed compatible with functionalism. To sum up, the development of functionalism was connected to dissatisfaction with the solutions provided by behaviourism and the mind-body identity theory. Behaviourism was too liberal in its attribution of mental states, whereas the mind-body identity theory was viewed as chauvinistic. From this we concluded that what might be thought of as the spirit of functionalism was to develop a theory that avoided these extremities, and at the same time kept a materialistic approach. In order to get a better understanding of what is involved in defining a functional role we looked at two different methods of functionalization: The machine method and the Ramsey sentence method. Both methods providing definitions of functional roles that avoid the use of mental state terms and which are neutral about the physical structure of the mental - 33 - states. Using different sources a priori or a posteriori psychology yields different versions of functionalism: commonsense functionalism and empirical functionalism respectively. However, both versions have been accused of being either too liberal or too chauvinistic. In my discussion of the liberalismchauvinism dilemma on the charge of liberalism I argued, with BraddonMitchell and Jackson, that Block’s examples involving the Chinese brain and the reorganized economy of Bolivia is outside the reach of our intuition and ,accordingly, that it seems to be at least an open question whether or not such a brain would indeed have mental states. Thus, one way to turn for the functionalist seems to be towards more abstract characterizations of inputs and outputs. Another way out of this dilemma, is to recognise the important insight delivered by Bechtel and Mundale, that the original debate on functionalism and multiple realization has been cast in different levels of abstraction for psychological states and brain states –- coarse-grained and fine-grained respectively. If we are willing to acknowledge that the notion of brain state is itself a functional notion in the way that Bechtel and Mundale argue, this could be a way out of the dilemma. Bechtel and Mundale also offer us another observation worth keeping in mind: which level of grain that is appropriate is determined by the context of inquiry. Judgements of sameness and difference are always relative to one’s reasons for wanting to make the comparison in the first place. Could we use our newfound distinction between a coarse-grained and finegrained analysis to track down the distinction between commonsense functionalism and empirical functionalism? To a first approximation this would seem to be the case in the sense that there seems be more grounds for a more detailed analysis working from the outset of scientific psychology than from commonsense psychology. However, it is - 34 - worth noticing that even scientific psychological analysis can be more or less coarse or fine-grained depending on the context of inquiry. Making a coarsegrained analysis as opposed to a fine-grained analysis, in Bechtel and Mundale use of the terms, is just to focus on the commonalities as oppose to the differences between individuals. In this light, Bechtel and Mundale classify characterisations of mental states at the level of scientific informationprocessing psychology as generally coarse-grained. Accordingly, in some cases the analysis in which an empirical functionalism can take its point of departure can be characterised as coarse-grained. In this way, it seems coarse-grained and fine-grained is not a matter of all or nothing, but a continuum. Although, the distinction does not completely track the distinction between commonsense and empirical functionalism it will be useful to keep in mind in the following discussions of the body-detail model and Extended Functionalism and our further investigations into the relation between embodied cognition and functionalism. - 35 - ______________________________________________ CHAPTER 2 THE BODY-DETAIL MODEL: SHAPIRO’S ARGUMENT ______________________________________________ Shapiro argues that research within embodied, embedded cognitive sciences conflicts with the spirit of functionalism. According to Shapiro, it conflicts with the body neutral functionalist model of mind that is associated with the claim that “characteristics of bodies make no difference to the kind of mind one possesses” (Shapiro (2004) p. 175). In his book “The Mind Incarnate”, Shapiro depicts these body-neutral views as what he calls naturalistic versions of Gilbert Ryle’s dogma of the ghost in the machine, which despite their naturalistic point of departure hold that mind is autonomously residing in and contained by the brain and the body. In “The Mind Incarnate” these views are represented by what he calls the seperability thesis (henceforth ST). ST is construed as the view that the mind is independent of the body in the sense that the very same mind can exist in many different kinds of bodies with different properties and that, as a result, the question of understanding mind can be separated from the question of understanding body. As Shapiro formulates it, ST views the mind as a “fairly self-contained organ, like a stomach or a kidney, whose properties and operations can be understood without having to attend much to the anatomical and physiological properties of the rest of the body”(Shapiro (2004) preface x). ST is thus the metaphysical claim that identical minds can exist in different kinds of bodies and the epistemological claim that we can understand mind without attending to the - 36 - facts of the body. As even classical functionalist accounts can allow that minds cannot be understood without having an understanding of bodies and the world (the epistemological claim) my focus will be on the metaphysical claim of ST. In opposition to ST, Shapiro interprets research in embedded, embodied cognitive science as supporting what he calls the embodied mind thesis (henceforth EMT). EMT is the thesis that “minds profoundly reflect the bodies in which they are contained” (Shapiro (2004). p. 167.) According to this thesis, a creature with a humanlike mind will have a humanlike body; and a creature with dolphin mind will have a dolphin body and so on. In fact, EMT claims “only humanlike bodies will have humanlike minds”. (Shapiro (2004) p. 182 my italics). The following scenario makes the opposite thesis clear: imagine that we, via radio, get in contact with some creatures from another solar system. We talk and interact with them in much the same way as we did with the Martians in the earlier example from Block. On this basis we assume that they have something very much like a human mind. As we have only encountered our new friends over the radio, we have never seen a picture or video of them. What kind of body would such creatures have? Proponents of ST would say that we could not know this. As far as we know our new friends could be spherical and judge distance with the help of sonar. Proponents of EMT on the other hand would expect these creatures to have a body very much like the human body (Shapiro (2004) p. 166). The Human Mind and the Human Body Before we turn to the research presented by Shapiro in favour of EMT, there are a few important points worth noting about the notions of the human mind - 37 - and body employed by Shapiro. The first is that Shapiro’s argument is concerned with entire minds and not just states. A human mind in the way Shapiro uses the term refers to a collection of cognitive capacities such as for example, memory, perception and use of language (Shapiro (2004) p.69). Combined with EMT, this means that creatures with dolphin bodies cannot have humanlike minds, but it does not imply that we cannot share mental states like for example, hunger or pain. However, whether this is the case depends, according to EMT, on whether they have body properties that are sufficiently similar to the relevant properties of human bodies. To understand the claim of EMT it is also important to be familiar with Shapiro’s notion of what it means to have a human body. Shapiro states that when he talks about human bodies he “will usually have in mind facts about our gross morphology. Human beings walk upright on two legs, have two forward-facing eyes in their head, and are bilaterally symmetrical. Our ears are on the sides of our heads, and we have five fingers on each hand” (Shapiro (2004) p. 71). In this way, EMT states that for a creature to have a human mind it must share the gross morphology that is generally shared among human beings. 2.1 Body Matters Shapiro argues against ST, referring to research in the field of embodied cognition. Shapiro presents three lines of research that he takes to support EMT. From the introduction we have already had a taste of what these involve, but here I will present the issues in further detail. - 38 - A. Perception and thought: the idiosyncratic aspects of the body play a unique and ineliminable role in perception, such that body form and structure determines type of mind B. Conception and content: “The content of the mind – the manner in which the mind conceptualizes the world – is a function of body type.” (Shapiro (2004) p.183) C. The division of mind and body: the mind is realized throughout the body, making the division between mind and body impossible to maintain; therefore body type determines type of mind I will present these three claims and the research that Shapiro offers in their support under the following headings: I) The Body’s Role in Perception, II) Body Concepts and III) Leaky Minds. Along the way I will give some concerns questioning whether these results suffice to support the strong claims of A,B,C and EMT. 2.2 The Body’s Role in Perception According to Shapiro, the body plays a significant and indispensable role in perception. In the introduction we have already encountered Shapiro’s example on auditory perception, where the size of the head and the placement of the ears matters for determining the location of a sound source. Another example is vision. Shapiro considers our depth perception. According to Shapiro, our abilities to move our bodies in certain ways and the particular characteristics of our eyes make an essential difference to our visual processing. - 39 - Shapiro points to the fact that we human beings have two eyes that are placed with a certain difference between them. When seeing objects in front of you, each of your eyes projects slightly different coordinates of these objects to the respective retinas. From the difference in coordinates your brain then calculates the relative distance of the objects. “The squirrel looks closer than the tree, which in turn looks closer than the fence, because the retinal image of the squirrel appears on the temple sides of each retinas whereas the image of the fence appears on the nasal sides of each retina, with the tree’s image nestled between the two”(Shapiro (2004) p. 187). According to Shapiro, what is important, in this case, is that the processing of the brain is fitted to receive the information from exactly two eyes – no more and no less. In the light of this, visual processing in a spider must be remarkably different from that of humans (most spiders have eight eyes, some 6. But there are spiders with no eyes, 4 and up to even 12 eyes). Accordingly, if we imagine that we could exchange the brain of a spider with a human brain, the processing done by the human brain would not be apt for providing depth perception. (Shapiro has an analogy of using a submarine instruction book for flying an airplane. And as he points out “this will almost certainly shorten your flight”.) Shapiro takes this to indicate that “Human vision needs a human body” (Shapiro (2004) p.189). Shapiro finds additional support for the above statement in work by Churchland, Ramachandran and Sejnowski (1994) on the issue of parallax. If you move your head or body in order to take a look around, you will notice that the things that are closest to you, will appear to move whereas the background continues to appear stable (This is the phenomenon of parallax). - 40 - These movements, according to Shapiro, are not just an aid to depth perception but indispensable parts of the visual processing. According to Shapiro, the picture is similar when we turn to our other perceptual abilities. Shapiro takes this to show that the body does not merely influence our mental properties (this would seem to be a trivial observation), but that its characteristics play an indispensable role in our mental operations: “..the point I draw from my comments about eyes and ears goes beyond the obvious claim that perceptual processes are tailored to body structures. The point is deeper—that psychological processes are incomplete without the body’s contribution” (Shapiro (2004) p.190) Shapiro further suggests that “this means that a description of various perceptual capacities cannot maintain body neutrality, and it also means that an organism with a nonhuman body will have nonhuman visual and auditory psychologies.” (Shapiro (2004) p. 190) Does this mean that, for example, people with only one eye or bad eyesight have non-human visual psychologies? I believe that this is the way we should interpret the above statement from Shapiro. To call it non-human may be a bit drastic, but Shapiro’s conclusion seems to imply that people with this kind of visual embodiment will have a different kind of perceptual psychology than what is standard among human beings. I think that this is likely to be true, but the question is what conclusions we should draw from this observation (more on this later p. 45). Shapiro takes the above research to show that the body plays such a specific and important role in perception that its - 41 - characteristics determining for our perceptual capacities. The result, according to Shapiro, supports EMT in the sense that it shows that we cannot separate mind from body, and that organisms with bodies of a particular type will have a matching mind. 2.2.1 A skill-based model of perception According to Shapiro (Forthcoming), we learn a similar lesson from Alva Noë’s account of perception. According to this view perception constitutively depends on the body and its relations to the surrounding environment. To be more precise, perception, according to Noë, occurs when an organism has implicit knowledge of how movements of its body or body parts will affect the flow of sensory stimulation that it receives, what Noë calls sensorimotor contingencies. For example, take my experience of a coffee cup on the table in front of me. According to the sensorimotor account, this experience is partly constituted by my implicit knowledge of how the stimulation would change if I, for example, moved my head to the right and thus looked at the cup from a different angle. Along the same lines, my experience also depends partially on my knowledge of how stimuli would change if I were to reach out and grab a hold of the cup. What is important in this context is that the sensorimotor contingencies would be different for a creature with a different body. Accordingly, reaching out for the coffee cup would be associated with a different set of sensorimotor contingencies for a creature with wings instead of hands. Because, according to Noë, sensorimotor contingencies are constitutive of experience, differences in these make differences to what experience one has. Hence Noë’s conclusion: “only a creature with a body like ours can have experience like ours” (Shapiro Forthcoming citing Noë p. 19). To what extent does perception on this account depend on the details of our embodiment? Noë and O’Regan (2001) have an example of the kind of characteristics that can make a difference to experience: - 42 - The human eye has a blind spot of 5-7 degrees in its field of vision. This corresponds to the place where the optic nerve goes through the eyeball and connects with the brain. The result is that small objects become invisible when in this field. Usually we do not notice the blind spot and as such it is not a part of our conscious visual experience. However, according to Noë and O’Regan, even though we do not experience the blind spot as such, it does make a difference to the phenomenology of our visual experiences and is part of what constitutes them. How can this be? According to Noë and O’Regan the presence of the blind spot produces certain sensorimotor contingencies. Take the experience of watching a bird flying in the distance. If there was to be no change in the sensation when the bird enters the blind field, “then the brain would have to conclude that the object was not being seen, but was being hallucinated.” (Noë and O’Regan (2001) p. 951) The brain gains information from the fact that there is a blind spot in the following way: “Monitoring the way the sensory stimulation from the retina changes when the eye moves to displace an object in the vicinity of the blind spot, is, for the brain, another way of gaining information about the object.” (Noë and O’Regan (2001) p. 951) Accordingly, experience of watching the bird fly in the distance is, according to Noë and O’Regan, partly constituted by my implicit knowledge of how the flow of stimuli is linked with my bodily actions. Now, if we were to imagine a creature that instead of having one blind spot on each eye had two such blind spots, this creature would associate the experience of watching a bird in the distance with other sensorimotor contingencies. - 43 - This illustrates that, according to the sensorimotor account, even differences in smaller details of one’s body and sensory system will make a difference to the experience – even such details as the way our retinal image changes when moving our eyes. The differences that are provided by the differences in sensorimotor contingences make a difference on a sub-personal level in the information received and “expected” by the brain. This in turn makes a difference in the phenomenological character of the experience for example whether or not I experience watching the bird as a hallucination. According to the sensorimotor account, every difference in embodiment will make it seem different to the perceiver. (Noë (2004), Noë and O’Regan (2001)) Further, Noë (2004) rejects the idea that we can, as ST states, separate what David Marr has termed the algorithmic level of description of cognitive phenomena from the level of implementation: “In general it is a mistake to think that we can sharply distinguish visual processing at the highly abstract level, on the one hand, from processing at the concrete implementational level, on the other. The point is not that algorithms are constrained by their implementation, although that is true. The point, rather, is that the algorithms are actually, at least in part, formulated in terms of items at the implementational level. You might actually need to mention hands and eyes in the algorithms!” (Shapiro (forthcoming) p.15 citing Noë, original italics and excitement) The claim that Noë makes in the above paragraph is that the psychological processing is fitted to body type – hands and eyes and other characteristics of body are involved in the computational processing. His point is similar to the one intended by Shapiro with his analogy of using a submarine instruction - 44 - book for navigating a plane. In this way Noë’s account, according to Shapiro, seems to support the claim of EMT – that human minds require human bodies. (Shapiro (forthcoming)) 2.2.2 Sensory substitution and plasticity A challenge to Shapiro’s (and Noë and O’Regan’s) analysis above has been provided by arguments that hold that the human brain is remarkably plastic. That is, it has an ability to adjust its organization and structure in order to receive alternative kinds of stimuli—making it possible to substitute parts of the body with alternative sensory equipment. One of the more known examples of such sensory substitution requiring that the brain take in information in a new way is Tactile-vision sensory substitution (TVSS) (see fig. 3): Figure 3 - 45 - The way that TVSS works: An image is selected by a video camera that is usually positioned on the subject’s forehead. The data from the camera is transmitted via cable to a display unit. The display unit transduces the video data into a pattern of low-voltage pulse trains, each corresponding to a pixel in the image. These trains travel via cable to electrode arrays that are placed on an area of the skin. The electrodes then stimulate the sensors of the skin. An area that has proved particularly apt for this is the tongue. Source: Bach-y-Rita and Kercel (2003) p. 543 Using this system subjects are able to learn to “make perceptual judgements using visual means such as perspective, parallax, looming and zooming, and depth estimates” (Bach-y-Rita and Kercel (2003) p. 543), thus, seemingly having an ability of gross visual discrimination similar in kind to that of ordinary vision. Another interesting fact is that after training with the TVSS for a while, the subjects report perceiving the image out in the world and not on the skin. These capacities stay unchanged even if we switch from one motor system (the camera places on the head) to another (for example a hand held camera). The effect is similar when the electrode array is moved to different locations on the skin (for example from the tongue to the thigh). This last fact seems to support the subjects’ reports of not perceiving the image on the skin, but as out in the world as is also the case in normal vision. What has been said above gives us reason to argue that just because characteristics of body are significantly involved in determining perception it does not entail that it could not have been different, and that alternative organisations of sensory systems could not have a similar effect– a point made by Clark: “[F]rom the fact that (as seems highly likely) our human experience really does depend in part on many idiosyncratic aspects of our embodiment, it does not follow that only a creature thus embodied could have those very experiences. The very most that follows is that, - 46 - for a creature like us, all other things being equal, we would not have that experience were it not for such–and–such an idiosyncratic fact.” (Clark (forthcoming) p.7) This also connects with our considerations about people with disabilities in their visual systems. People with bad eyesight probably have different perceptual abilities compared to what is standard among human beings, but this is only with all other things being equal. However, things are not equal and people with bad eyesight can wear glasses or contact lenses. Do bodyspecifics really make such a special difference to mind that such aids do not make a difference in determining mind? I think that what has been said above gives us reason to doubt the strong claim that characteristics of our sensory system determine our perceptual experiences in such a way that all differences in body entail differences in mind. 2.3 Body Concepts A second area where Shapiro claims to find support for EMT is the role of the body in structuring our concepts and the way we think about the world -the content of our thoughts. Here Shapiro draws on the work of both G. Lakoff and M. Johnson (1980; 1999) and A. Glenberg and M. Kaschak (2000). Ultimately Shapiro takes this research to show that “the basic categories an organism develops in order to think about the world are a function of how its body interacts with the world, and thus differences in body type will result in differences in how and what an organism thinks or is capable of thinking” (Shapiro (2004) p. 183). - 47 - 2.3.1 Concepts and Metaphors The central idea of Lakoff and Johnson’s work is that the form and structure of our body play a significant role in structuring our basic concepts and that these basic concepts, in turn, structure more abstract concepts and metaphors. According to this research, basic concepts like ‘front’, ‘back’, ‘up and ‘down’ are directly related to the shape and type of our bodies. The claim is that these basic concepts are understood through the experience of the world that we have with a human body. Lakoff and Johnson hold that such concepts would not make any sense to beings which have bodies with no front or back. Imagine that all beings on this planet were spherical and able to perceive in all directions. What would ‘front’ and ‘back’ be for such creatures? According to Lakoff and Johnson, they would have nothing like our concepts of ‘front’ and ‘back’. (Example by Lakoff and Johnson cited by Shapiro (2004) p. 199) The basic concepts, in their turn, structure our more abstract concepts. A much used example is the concepts of happiness and sadness. Happiness is structured in terms of the basic concept ‘up’ while sadness is structured by the basic concept ‘down’. Hence for creatures with no concepts of up and down happiness and sadness would, according to Lakoff and Johnson, be differently structured. All this, according to Lakoff and Johnson, goes to show that the content of our psychology, the way in which the mind forms concepts is, at least partly, determined by body type. And as they conclude: “[T]he peculiar nature of our bodies shapes our very possibilities for conceptualization and categorization” (Shapiro (2004) citing Lakoff and Johnson (1999), p. 200). 2.3.2 Body and language The work of Lakoff and Johnson is not based on experiments and, in order to support his claim with empirical evidence, Shapiro draws on work of the psychologists Arthur Glenberg and Michael Kaschak(2000) to show that the nature of our body is reflected in the way that we conceive the world. - 48 - In order to understand the experiments and findings of Glenberg and Kaschak it is important to be familiar with the notion of affordance as it has famously been developed by the psychologist J. J. Gibson. An affordance in Gibson’s terms is that which is offered to an organism by the features of its environment. The crucial idea is that what is offered is relative to the properties of the perceiver - objects offer different things to different organisms. (Shapiro (2004) p. 203) A ceiling might provide a place for resting for a fly, but not so for a turtle or a pelican. Likewise, a small stone might be an object that you can throw, use as a tool or in jewellery for humans, whereas it might be an obstacle for an ant. Now let us turn the work of Glenberg and Kaschak and how it is interpreted by Shapiro. A central experiment in their research involved detecting people’s ability to understand sentences containing conventional verbs and innovative denominal verbs in different contexts – contexts that provide affordances and contexts that do not. Denominal verbs are verbs that are made out of nouns as for example ‘couch’ in the sentence: “Vera couched in front of the TV”. An innovative denominal verb could be something like ‘chaired’ in “Rachel chaired the scientist his mail” where chaired is suppose to refer to the act of transporting something by the means of a chair (example from Kasckak and Glenberg (2000)). Prior to reading the sentence containing the innovative denominal or conventional verb, the subjects read a text that was designed to set up a situation where the character of the story had it as a goal to transfer something (e.g. transfer the mail to the scientist.) See Table 1. Each text had two versions: an afforded and a non-afforded. The afforded version contained an object that easily afforded the transfer (e.g. a chair with good wheels). The - 49 - non-afforded version contained the same object, but where some of the properties had been altered so that it no longer provided easily affordance of transfer (e.g. a chair missing the wheels). At the end of each text, the critical sentence would contain an innovative denominal verb (e.g. chaired) or a conventional verb (e.g. brought). The sentence comprehension was measured by the time it took the subjects to read this critical sentence. Table 1: Sample stimuli from Kaschak and Glenberg (2000) Rachel worked for a scientist in a research firm. As part of her duties, she was required to bring the scientist’s mail to his office so he could open it after lunch. On this particular day, Rachel encountered three large boxes among the mail addressed to the scientist. The boxes were way too big for her to carry. Affordance manipulating sentence: In the corner of the room, though, Rachel noticed an office chair with four good/missing wheels Critical sentence Rachel brought/ chaired the scientist his mail. The result of the experiment showed that the subjects were faster reading the critical sentence when the context was afforded than when it was nonafforded. Interestingly, the difference in reading time between afforded and non-afforded versions was larger in the case of innovative denominal verbs than conventional verbs. The results concerning the innovative denominal verb are especially interesting because these are the cases where learning of a new word is involved. Shapiro takes these results to show that language comprehension involves the knowledge that an organism has about how its body can interact with its environment. Shapiro concludes: “[A]n ability to understand sentences seems, at least in many cases, to incorporate an - 50 - organism’s knowledge about how its body might interact with objects in its environment. Differences in body will, presumably, create differences in ones ability to understand sentences, and from fact about sentence comprehension one could predict fact about the body” (Shapiro (2004) p. 212). Shapiro is carefully trying not to overstate his conclusion, but what does it mean that differences in body create differences in one’s ability to understand sentences? And in what sense does ability to understand sentences incorporate an organism’s knowledge about the interaction possibilities of its body and environment? The first thing to note is that the first conclusion seems to rely on the latter: it is only the case that differences in body can make any differences in ability to understand sentences if it is the case that the body, in some sense, is involved in linguistic understanding. Thus, a good place to start our investigation is to consider how we should interpret the conclusion that the ability to understand sentences incorporates an organism’s knowledge about the interaction possibilities of its body and environment. We can start by recognizing that the word “incorporate” seems to allow at least two different interpretations of the claim: Strong: Linguistic understanding requires having knowledge of one’s own embodiment and possibilities of interaction with a certain environment. - 51 - Weak: Linguistic understanding can use, but does not require having knowledge of one’s own embodiment and possibilities of interaction with a certain environment.6 The strong reading advocates the view that linguistic understanding is constituted by the knowledge one has of the possibility of one’s body’s interaction with one’s environment. In contrast, the weaker reading suggests that that the use of this kind of knowledge in understanding is only contingent. How should we read Shapiro’s claim? On the one hand Shapiro’s use of the phrase “at least in many cases” in the original formulation indicates that there is a possibility of having linguistic understanding without having knowledge of ones own embodiment. If this is the case then the appropriate reading is the weaker version. On the other hand, this might not be enough to support the strong claim of B: “The content of the mind – the manner in which the mind conceptualizes the world – is a function of body type” (Shapiro (2004) p.183). Likewise it might not be enough to support EMT, that only creatures with humanlike bodies can have humanlike minds. Shapiro (forthcoming) suggests that “If differences in body type are to suffice for the fracturing of psychology [that psychology must be species specific] the body must have a role that is deeper than mere realization” (Shapiro (forthcoming) p. 19). What Shapiro is aiming at here is that the body must in this case play a constitutive role. However, in what follows I will follow Weiskopf (Unpublished ms) in arguing that the weaker claim is the most plausible. 6 This distinction between a strong and a weak reading of the claim is inspired by Weiskopf (Unpublished) where he considers the possible relations between linguistic understanding and the process of generating a sensory motor representation of a linguistic described situation. - 52 - The minimal view of understanding Part of the question, Weiskopf argues, lies in the way in which we interpret the notion of linguistic understanding. In this connection, Weiskopf presents a view of linguistic understanding under which it is straightforward that knowledge of one’s own embodiment is not necessarily involved in the understanding of sentences. According to what Weiskopf dubs the minimal view, understanding a sentence (only) requires understanding the truth condition of this sentence. This involves being able to make certain logical inferences from the sentence in question. Take the sentence: ”Felix enjoyed watching the festival shows in August”. Understanding this sentence under the minimal view involves being able make inferences of the following kind: Someone enjoyed watching the festival shows in August and Felix watched the festival shows. Weiskopf’s central claim is that such inferences do not involve knowledge of affordances let alone knowledge of one’s own embodiment. To show that this is the case Weiskopf uses the example of the following nonafforded sentence: “Adam pulled out of his golf bag a ham sandwich and used that to chisel an inch of ice off his windshield.” (Originally from Glenberg and Robertson (2000)) In this connection, Weiskopf points out that although a sandwich does not (usually) afford chiselling off ice of windshields there is a way in which it seems reasonable to say that the sentence is understandable. From the sentence we can infer for example that Adam used something to remove the ice from his windshield. Further, by assuming that the pronoun “that” refers to the ham sandwich we can infer that the thing that removed the ice was stored in the golf bag. We can even infer something about the sandwich, namely that the sandwich is the kind of thing that can chisel ice. Combining this knowledge with - 53 - general knowledge that things that chisel ice have to be hard we can infer that the sandwich is hard. Neither of these inferences seems to rely on one’s own embodied encounters with the kind of objects involved. Thus, Weiskopf suggest it is not obvious that linguistic understanding under this view requires knowledge of embodiment. This thereby speaks for the weaker reading. It does not, however, completely cement the case for the weaker reading, because the notion of understanding that Shapiro has in mind seems to be more complex than the minimal view. Understanding as involving sensibility judgements On the notion of understanding that Shapiro seems to have in mind, understanding a sentence involves being able to judge whether the sentence is sensible or not. This view of understanding seems to be present in the following quote from Shapiro: “Bodies like ours can push chairs, but cannot swallow them; we can sit on chairs but cannot balance them on our heads;… All of these facts about chairs are contingent on facts about human bodies. Likewise the perspective from which a human being judges the sensibility of a sentence is one that reflects these facts.” (Shapiro (2004) p. 212) In this way, it might be argued that judging the sensibility of sentences like “Rachel chaired the scientist his mail” and “Adam used the sandwich as a chisel” requires knowledge about facts of the world; about the affordances of chairs and sandwiches for human beings. This certainly seems to be the result of the experiments of Glenberg and Kaschak. However, it is a further question whether knowledge of affordances, in turn, requires knowledge of one’s own embodiment? - 54 - First of all it seems reasonable to say that Glenberg and Kaschak’s experiments, in the first instance, show that ability to understand a sentence, in terms of judging its plausibility, depends not on the reader’s knowledge of his or her body’s abilities to interact with objects but instead on the understanding the reader has of how the body of the character in the story can interact with the environment provided by the text. We assume that the character in the story, Rachel, is a normal human being and as such she is able to push a chair. This provides the information that we need in order to understand the sentence: “She chaired the scientist his mail.” I take it that this is the point that Shapiro has in mind when he writes: “If a subject were led to believe that Rachel were two inches tall, or weighed only five pounds, or had no appendages with which to grasp, then, quite clearly, the suggestion she chair the scientist his mail would be no less perplexing than the suggestion that she tennis ball the scientist his mail, or shingle him his mail, or grass him his mail (Shapiro (2004) p. 212 italics original). But, if it is the case that our ability to understand the sentence depends on knowledge of the character’s body, where does knowledge of one’s own body come into the picture? One might argue that understanding someone else’s perspective depends on one’s knowledge of how one’s body can interact with the world. To understand the sentence, “Rachel chaired the scientist his mail”, I have to know something about what is needed in order to be able to use a chair as the means for transporting a large box of mail. This knowledge, it would seem, I have, at least partly, from my own interactions with the world. In this sense, knowledge of my own body’s possibilities of interaction could be said to be incorporated in my ability to understand sentences. However, this is only in a derived sense and further it seems to rely heavily my - 55 - assumption that Rachel is a normal human being. Let us imagine that I instead had been told that Rachel is alien with superpowers that allow her to move chairs only using her mind. In this case it still seems reasonable to claim that I can judge the sensibility of “Rachel chaired the scientist his mail”. However, it is no longer clear that it matters to any extent whether I myself have arms or wings, feet or fins. It seems that the knowledge I have of the affordances involved in this case, is the kind of knowledge that I have about the world in general – together, of course, with the inferences I can make from the knowledge of Rachel’s special powers. In contrast, it does not seem to involve the knowledge I have of the precise details of my own embodiment. This suggests that comprehension of sentences and thus knowledge of affordances of the objects involved does not require the knowledge of the precise details of one’s own embodiment. Weiskopf (Unpublished) offers an analysis that can help us get a deeper understanding of why this is the case. Weiskopf suggests that the kind of arguments supporting the claim that knowledge of one’s own embodiment is required for knowledge of affordances, make the mistake of conflating “having information about humans’ physical structure and representing that information in a particular way (e.g., a way that only a creature with humanlike sensorimotor capacities could enjoy).” (Weiskopf (Unpublished ms), p. 23) Weiskopf’s claim is that there has not been enough attention paid to the difference between psychological content and representational vehicle. The representational vehicle is the mental particular that caries the content. Importantly, this content can be carried by a wide range of different kinds of representational vehicles. To clarify his point Weiskopf draws on Dretske's analysis of content in terms of information (Dretske (1981)). Information can be represented in a number of different ways. Weiskopf uses the example that if one - 56 - wants to inform somebody that the cat is sleeping, one has lots of different ways of doing this: one could, for example, draw a picture of it or write it down, or signal it, mime it or tell it. It is central to Weiskopf’s claim that from knowing the information it is not possible to infer anything about the vehicle. Knowing that you have the information that the cat is sleeping it would not be possible for anyone to tell whether you had seen it yourself, had been told it or had read about it. According to Weiskopf, the picture is similar with information about affordances; it too can be represented in different ways. Firstly, it can be part of the content of a perceptual state, as when I immediately perceive the affordance of the chair in front of me. Secondly it can be part of the content of an informational state (for example a belief) that is immediately derived from perception. Thinking about a chair might immediately bring up thoughts about what this object affords, how it might be used. Thirdly it could be part of the content of informational states that are not immediately derived from perception, but instead via abductive reasoning. That is, it could be converted into conceptual content. The immediate consequence of this is that what Glenberg and Kaschak's results really shows is that in judging whether a sentence is nonsensical, we employ states that carry information about affordances, but not that this information is represented in a way only be available to creatures with a certain body type. In fact Weiskopf suggests that in many cases “coming to judge that… scenarios are or are not sensible might rely on general abductive competence — the ability to make inferences about real-world situations drawing on relevant knowledge — rather than any specifically sensorimotor capacities.” - 57 - In this light the weaker reading of the relation between linguistic understanding and knowledge of one’s own embodiment seems to be the most plausible. Now we might ask: is this enough to establish Shapiro’s second claim, that differences in body create differences in one’s ability to understand sentences and in this case, what is involved in the notion of difference? My claim is that because knowledge of one’s own body is often involved only in a derived way, small differences in one’s body do not create differences in one’s ability to understand sentences. Only humans can understand language so it’s difficult to test on other creatures, but what has been said above suggest that a humanlike body is not required to understand sentences. This is the case even when it is a question of the understanding involves appreciating facts about how the world affords itself to human beings. Thus, sameness of understanding does not seem to require sameness of even gross bodily structure. 2.3.3 Individual embodiment vs. social embeddedness In support of this conclusion we can also draw on Harry Collins’ ((Forthcoming))work on the relation between the individual, its body and its language. Collins argues that in the sphere of language the type of body of the individual has no influence. Instead, what matters is the brain and the social embeddedness. He argues that the ability to speak a language of a community fluently can be acquired without the individual being able to take part in the bodily activities of that community. In Collins’ terms one can have interactional expertise (fluency) without having contributory expertise (being able to engage in the bodily activities of the community). The claim is that on - 58 - a test solely based on language someone with maximal interactional expertise and no interactional expertise will be indistinguishable from someone with both sorts of expertise. As an example, Collins refers to cases where colour blind people have passed as colour perceivers (see also Collins et al. 2006). Following this line of argument Collins endorses what he calls the Minimal Embodiment Thesis, according to which the precise details of the body of the individual does not have relevance for the linguistic abilities of this agent. Another way to state this is that “an individual needs only a minimal body (ears, mouth, larynx, and maybe something else), to become fluent in the language of the surrounding society” (Collins et. al (Forthcoming)). Collins suggests that the language and thoughts available to the surrounding society may be determined by the typical bodily form of its members (This is called The Social Embodiment Thesis by Collins). This, however, does not affect the fact that the precise detail of the individual members’ embodiment does not determine the language available to that individual. Following this line of argument Collins agrees with Wittgenstein in saying that ‘if a lion could talk we would not understand what it said to us’. However, Collins’ claim is that this is not because of the body shape of that particular lion. Rather it is because lions typically are brought up in communities of lions. Thus if a (talking) lion was brought up in a human community we would be able to understand it (Collins (Forthcoming) p. 5). Again this is against the strong reading of Shapiro’s claim that differences in one’s individual body type create differences in language comprehension. Further, it gives us a new way of reading Lakoff and Johnson’s claim that “the peculiar nature of our bodies shapes our very possibilities for conceptualization and categorization” (Shapiro (2004) citing Lakoff and - 59 - Johnson (1999), p. 200). We could interpret this as saying that the typical body type of our global human community limits the possibilities for conceptualization and categorization. Thus a community of spherical beings might not have concepts of FRONT and BACK, but there’s no reason why an individual spherical being could not have this e.g. if living in a human community. What has been said above casts doubt as to whether the research presented in favour of EMT by Shapiro can support the claim in the required way. Again, what has been said above is a way of saying that just because body characteristics can and do play an important and significant part in determining aspects of our mentality, it does not follow that this is always the case and that every little difference will make a difference to mind. 2.4 Leaky Minds As a last argument against ST, Shapiro refers to neurological research presented by Damasio in “Descartes Error” (1994). This research suggest that the brain and body are so intimately connected that it is impossible to distinguish processes that realise mind from bodily processes. This is the case because there are parts of the brain such as hypothalamus, the brain stem and the limbic system that are involved in all the neural processes on which mental phenomena are based, but which are also involved in regulating bodily functions. Further these parts of the brain are, at least partly, regulated by the complex chemical activity of the body. Damasio concludes: “Body regulation, survival and mind are intimately interwoven” (Shapiro (2004) p. 217 citing Damasio (1994) p.123). From this Shapiro draws that mind is realized in a complex process that spreads throughout the body and is not - 60 - only in the brain as it has been commonly assumed. In this way the realization of mind leaks out into the body. How does this apply to the case against ST? According to ST the mind can be separated from the body and exist in many kinds of bodies. This view, according to Shapiro, is associated with the view that mind, like a resistor in a TV, is a “relatively independent and selfcontained” component that only interacts with the body through well defined interfaces (quote from Haugeland (1995) cited by Shapiro (2004) p. 212). For the mind to be a component it does not necessarily require that it can be spatially disconnected from the rest of the body. Under Haugeland’s interpretation of the notion, something can be a component without being spatially confined. Whether we can interpret something as a component of a particular system depends on the nature of the internal connections between it and the system. An example could be the Human Resource department in a multinational firm. The employees of such a department might be stationed in different countries on different continents. What makes the department a single component, regardless of the spatial distance between the individual members of staff, is among other things, the fact that they work with the same problems and have a high level of interaction and communication between them. For something to be a component, in this non-spatial sense, it has to have clear boundaries and interfaces to what surrounds it also. According to Shapiro, Damasio’s research shows that there are no clear-cut boundaries and interfaces between what realizes the mind and the rest of the body. The point is not that we cannot remove the brain from the body and therefore we cannot separate mind and body: we could imagine that we could remove the brain and connect it to the body via wires. However, the point is precisely that if we move the brain in such a way, we have not thereby moved - 61 - the mind, because part of what realizes the mind is the brain and body combined. Further, the mind is not a component in the sense of a Human Resource department either, because the processes that realize mind are so closely connected to the processes of the rest of the body, that it does not make sense to make this distinction. The result, according to Shapiro is that mind cannot be separated from the body as ST claims. Further, Shapiro claims that this supports EMT in the respect that if we cannot separate the realization of the mind from the body “then of cause differences in bodies will entail differences in mind” (Shapiro (2004) p. 183) In the end of his discussion of the issue of this kind of extended realization in “The Mind Incarnate” Shapiro considers the thought that the realization could extend beyond the body and into the environment. Shapiro suggests that in such cases, “[p]erhaps anything with a humanlike mind must live in a humanlike natural and social world. Although not conceived in these terms, research in embodied cognition by people like Clark (1999, 2001), …and Wilson (2004) promotes such a view” (Shapiro 2004 p. 224). In the following chapter we shall see a different way of interpreting these arguments. First, however, I will accompany Grush (2003) in questioning Damasio’s neurological findings as evidence that the mind is not a component.. Shapiro bases his argument on Haugeland's criterion of what makes something a component. As we saw, according to Haugeland’s criterion, whether something is a component is dependent on the level of interaction between it and other 'parts' of the system (Grush dubs this The Bandwidth Criterion). - 62 - Grush questions this criterion, arguing that it is false and that it gains its prima facie plausibility from a connected but deeper criterion: The Plug Criterion. According to this criterion something is a component if it can be plugged into and unplugged from different systems. It is often that these two criteria correlate, because from a design point of view it is often easiest to place the plug interfaces at points with low bandwidth interaction. However, according to Grush, there is no necessary connection between the two and in situations where they come apart the plug criterion has the upper hand. To show that this is the case Grush uses the example of the relation between a pilot and her plane: “A pilot sits in a cockpit. Her hands in contact with the joystick, minute forces and resistance it offers are skilfully processed and aid her in putting the appropriate forces on the stick. The visual scene is complex, providing a great deal of information in the optic flow concerning the movement. Even her sensitivity to forces of acceleration and gravitation are part of this high-bandwidth interaction. In short, let us agree with Haugeland that this activity is high-bandwidth all the way through. Is the pilot a component? She can get up, unplug her highbandwidth interfaces from the flight simulator, and if she so chooses engage instead with a real aircraft through those same interfaces.” (Grush (2003) p.80) Grush takes this example to show that even though there is a high bandwidth interaction between the pilot and the plane we can still consider the pilot as a component because she can be plugged and unplugged from both the plane and the flight-simulator. The point to emphasise here is that if, instead of taking our point of departure as the bandwidth criterion, we take it is the plug criterion it no longer seems obvious that Damasio’s findings imply that - 63 - the mind is not a component. Whether we in fact can unplug the mind in the way suggested by the plug criterion is of course a further empirical question. Additionally, even if it turns out that the processes that realize mind and those of the rest of the body cannot be distinguished as separate components we may not necessarily be able to conclude that sameness in of mind must require sameness of body. Damasio’s results may show that some bits of the body are bits of the mind too. If that’s so, then differences in those bits of the body will, in at least one sense, entail differences in mind. However, this does not show that we need sameness of gross bodily morphology in order to have sameness of mind. 2.5 Conclusions Shapiro’s overall claim is that if the research in the area of embodied cognition and mind is correct then it undermines abstract versions of functionalism associated with body neutrality— the idea that characteristics of body make no differences to the kind of mind that one posses. He argues that research of embodied, embedded cognition and mind shows that characteristics of embodiment determine the characteristics of the mind. In this way, his argument is representative of what I have termed the body-detail model. This reading of the research of this area has the consequence that the very same kind of mind cannot exist in bodies with different characteristics, and thus it requires a humanlike body to have a humanlike mind. On this ground Shapiro also characterises his position as chauvinistic, a label which further highlights its suggested opposition with the spirit of functionalism. - 64 - The body-detail model and functionalism At this point it is worth to explore how exactly the position presented by Shapiro, if correct, would conflict with a functionalistic approach. In exploring this question it will be helpful to use Bechtel and Mundale’s distinction between coarse-grained and fine-grained analysis. It seems obvious that what Shapiro is presenting us with is first of all a fine-grained analysis of the body. Even though Shapiro talks about gross morphology as the thing which has to be shared, his primary examples are about details of placement of ears and eyes. The attention to detail is also evident in the sensorimotor account. This fine-grained analysis of the body is matched with at similarly fine-grained analysis on the psychological level. Now, as we have learned from our exposition of Bechtel and Mundale in chapter 1, judgements about differences and sameness are always dependent on the context of inquiry. The appropriateness of grain size is dependent on what one is interested in investigating. Further, even if differences in brain states match differences in mind on a detailed level this does not exclude that on a more abstract level differences in body does not entail differences in mind. There seems to be no reason why this should be any different when we are concerned with the relation between body and mind. If this is correct, then it is not the case that a fine-grained approach in itself undermines the abstract versions of functionalism. This is only the case if we claim that a fine grain is the only appropriate grain under which mind should be investigated. Shapiro claims that if the research of Embodied Cognition surveyed above is correct then this would show that the fine-grain analysis is the generally appropriate one. Because body type is constitutive of mind, according to Shapiro’s analysis of this research, differences at the body detail level will entail differences at the psychological level. Noë’s claim is that differences in body necessitates differences in mind and under Shapiro’s reading of for example Lakoff and Johnson it is in virtue of having a human body that the human - 65 - concepts mean what they do. If this is the case, then according to Shapiro we have “no choice but to give up hope of generality in favour of species-specific regularities of the organism over which they range” (Shapiro (Forthcoming) p .20) – a way of saying that psychology is species-specific. A fine-grained functionalism? It is important to note that although arguing that a fine-grained analysis is the right level of analysis is in conflict with the abstract versions of functionalism, it is not necessarily conflicting with functionalism as such. In this connection, Shapiro considers the possibility that his position could be compatible with a version of functionalism where the “specification of the inputs and outputs that define mental states are chauvinistic because it restricts attributions of minds to just those organisms with bodies or brains of a particular sort” (Shapiro (2004) pp. 174). However, he does not develop this thought in much detail. As the current project has the overall aim of exploring the logical space between functionalism and the embodied, embedded approach, it is nevertheless interesting to investigate this possibility. Can the positions presented by and Shapiro (2001, 2004), Noë (2004) and Noë and O’Regan (2001) be read as functionalistic, in a chauvinistic and fine-grained sense? In Pressing the Flesh Clark suggest a prima facie way of reading these positions as treating “the body or sensory apparatus as making a unique kind of functional/computational contribution, one that cannot help but sensitively impact certain aspects of mind” (op. cit p.16). In the same way that we, in Chapter 1, considered a way of formulating the functional profile of the kind drill bit in such a fine-grained way that only diamonds could play the role, this is a way of formulating the functional profile of the mental in such a way that only very similar characteristics of embodiment can play this role. In our distinctions in terms of fine-grained and coarse-grained this would be a - 66 - version of functionalism that applied a fine-grained analysis on both the level of the definition of the functional role and at the level of realization matching every difference in structure with difference in mental states. This seem like a plausible interpretation where we keep the insights of functionalism concerning the way mental states are defined by inputs, outputs and inner interactions among states and the thought that every difference in the relevant implementation makes a difference for mind. However, as we all know (and almost expect, at least in philosophical contexts) first looks can be deceiving. As Clark (In press) points out: It seems reasonable to require from a functional account, that it leaves it an open question whether some role can be fulfilled exclusive by one specific kind of implementation at least until the matters have been solved empirically. To make this clear consider again our example of functionally defining the kind drill bit. In this situation it is not the case that we, prior to empirical investigation, will be able to say that it must be the case that only diamonds can fulfil this role. There is, at least in principle the possibility that there are alternative implementations. Clark (In Press) argues that Noë and Lakoff and Johnson’s accounts seem to be in conflict even with this fundamental requirement of functionalism. Clark’s claim is that these accounts seem to have foreclosed the possibility of alternative realizations. Thus, Clark makes the following observation about the sensorimotor account: “By simply identifying experience with implicit knowledge of the full suit of contingencies defined at the sensorimotor surfaces, the strong sensorimotor account leaves no room for compensatory downstream adjustments to yield identical dissimilarities.” (Clark (In Press) p.21) - 67 - experiences despite surface I take Clark’s overall point to be that we do not, at this time, have the empirical knowledge to warrant the conclusion that only human bodies or sensory systems can realize human minds and perception. It would not be in line with any kind of functionalism to close the possibility when empirical matters are unsolved. In this way the body-detail model seems to be incompatible with functionalism on a broader scale than first anticipated. The above discussion has revealed the deeper opposition between the bodydetail model and functionalism: It lies in the fact that the body-detail model does not leave room for alternative implementation from the outset. This is also where most of our critique against the position comes in. The critique is that there seem to be no warrant on empirical grounds to support the strong conclusion – in fact it even seems likely that alternative implementations might be available. Rejecting the body-detail model In this chapter I have shown that the research material presented by Shapiro does not warrant the strong conclusion of the Body-detail model. Although, it seems reasonable on the basis of the research on perception to conclude that body does play a very important role in cognition, this does not imply that the characteristics of body type are indispensable and that nothing can substitute them. Wearing a pair of glasses can make up for weak eyesight and research on sensory substitution and brain plasticity gives us reason to believe that alternative implementations could be made available. This suggests that at least not every difference in characteristics of body and gross morphology will result in difference in mind. The situation is similar if we evaluate the research presented on the issue of language comprehension and conceptualization. Just because knowledge of one’s embodiment can be involved in language comprehension this does not mean that it is necessarily - 68 - required. In fact we have reasons to believe that in most cases this is not the way things are. Information of affordances can be carried by informational states and be the object of abductive reasoning. The fact that we are embedded in a language using community has the consequence that the bodily characteristics of the individual are not the determining factor for the limits of that individual’s language comprehension and conceptualization. On the third area of the mind leaking into the body our claim was that the bandwidth criterion did not suffice to tell us if something is a component or not. That the processes of mind and body are intimately tied does not necessarily make it the case that the distinction between mind and body cannot be maintained. The introduction of the plug criterion challenged the high-bandwidth criterion and Shapiro’s interpretation of Damasio’s research. Further, even if the realization of mind is leaking to the body this does not necessarily imply that the very same mind could not have been realized in a different kind of body. Rejecting this strong reading of the embodied, embedded mind research results suffice to reject Shapiro’s claim that research within the area of embodied, embedded cognition and mind undermines the abstract functionalistic model of mind. Further, it paves the way for a weaker reading of this kind of research where we keep the insight that the body plays an important role in cognition and mind, but which, at the same time, is compatible with non-chauvinistic versions of functionalism. This will be the subject of the next chapter where I will present Clark’s interpretation of the embodied, embedded approach as extended functionalism. - 69 - ______________________________________________ CHAPTER 3 EXTENDED FUNCTIONALISM ______________________________________________ In contrast to the body-detail model, Clark (In Press), argues for a way of understanding the mind as embodied and embedded which is not only compatible with functionalism but is a way of extending it. He argues that there is a way of recognising the body as significant for and constitutive of the mind, that does not entail the conclusion that, in all cases, differences in body characteristics creates differences in mind. To make his point Clark draws on arguments, from the research area of embodied and embedded cognition, which supports the claim that “aspects of the body and world can, at times, be proper parts of larger mechanisms whose states and overall operating profile determines our mental states and properties.” (Clark (In Press) p. 4) This is what Clark and Chalmers (1998) have labelled Extended Mind or what Clark (In Press) calls The Larger Mechanism Story. According to Clark, these arguments generally appeal to the computational role in online problemsolving played by processes and events of the body and the environment (Clark (In Press) pp. 9) In the introduction of this project, I have already presented a couple of examples, which Clark takes to exemplify an extended functionalistic approach. In what follows I will use these and additional examples to go into further details of the position. My focus will be on illuminating what is involved in what Clark denotes an extended functionalist approach and how the examples, presented by Clark, are exemplary of such an approach. - 70 - Subsequently I will investigate how this view contrasts with the body-detail model and its consequences for the issues of chauvinism and multiple realization. 3.1 Extended computation One of the examples, presented by Clark, with which we are already acquainted, is the example concerning Ada. Ada is the accountant who uses pen and paper to perform complex accountancy tasks. Using pen and paper in the way that she does allows her to offload some of the work from her biological memory to external processes – what have been called minimal memory strategies. “Instead of committing complex dependencies to biological memory, Ada follows trails through the numbers, creating external traces every time an intermediate result is obtained. These traces are in turn visited and re-visited on a ‘just in time, need to know’ basis, moving specific items of information into and out of short term bio-memory.” (Clark (In press) p. 12) Clark suggests that this accounting process is best analysed in “extended functionalist terms as a set of problem-solving state transitions whose implementation happens to involve a distributed combination of biological memory, motor actions, external symbolic storage and just-in-time perceptual access” (Clark (In Press) p. 12). As a framework for understanding the above example, Clark suggests using Robert Wilson's notions of Exploitative Computationalism. - 71 - Representation and Wide 3.1.1 Exploitative Representation Representation, according to Wilson (2004), is an activity performed by individuals “extracting and employing information that is used in their further actions” (Wilson (2004) p. 183). This prepares the grounds for the view that representing often involves exploiting information that is already directly available. Wilson calls this Exploitative Representation. The strategy of exploitative representation is exemplified by the way in which an odometer keeps track of how many miles a car has driven (Example by Wilson). An odometer is usually engineered in a way so that it exploits the fact that each rotation of the wheels is x meters. Thus, for every wheel rotation the odometer simply record x meters. This exploitative strategy stands in contrast to another a more complex strategy, which could be to let the odometer first count the wheel rotations and then calculate the result from the assumption that each rotation = x meters. In this second strategy the information would first be encoded and then computed. In contrast the first strategy, simply exploits the information that is already there (Wilson p.163 also cited by Clark (In press) p.13). According to Clark, this is the form of computational strategy involved in the minimal memory strategies employed by Ada when performing the accountancy task. In Clark’s words “Ada’s biological brain does not create and maintain persistent encodings of every figure she generates” (Clark (In Press) p. 13). Instead, it relies, in its computational strategy, on the possibility of embodied action and the presence of pen and paper to make the information available when needed. - 72 - 3.1.2 Wide computationalism The notion of exploitative representation prepares the ground for what Wilson labels Wide Computationalism. The idea is that computational systems can be widely realised in the sense that parts of its realization are located outside the head of the individual organism. In this connection, Wilson distinguishes between entity-bounded realizations and wide realizations. In the case of entity-bounded realization all the parts of the realization are located entirely within the individual. In the case of a wide realization some parts of the realization are not located within the individual. (This requires that the relevant system S of which the total realisation is a state extends beyond the individual). The important shift accompanying the idea of wide realizations is that realizations are within systems (as opposed to in individuals) and that these systems are located within in a broader environment (Wilson (2004) pp. 117). Wilson’s position is captured nicely by the following: ”[W]hy think that the skull constitutes a magic boundary beyond which true computation ends and mere causation begins? Given that we are creatures embedded in informationally rich and complex environments, the computations that occur inside the head are an important part but are not exhaustive of the correspondent computational systems. This perspective [(viewing representation as an active and exploitative process)] opens up the possibility of exploring computational units that include the brain as well as aspects of the brain’s beyond-the-head environment.” (Wilson 2004 p. 165) The difference between the traditional and wide conception of computation is illustrated by Wilson with an example concerning multiplication. On the traditional view multiplying 675 with 765, would involve encoding of input - 73 - from world symbols into internal symbols and a subsequent internal computation of these internal symbols. On this view representation is seen exclusively as a form of encoding. In contrast, on the wide computationalist view multiplication involves active exploitation, manipulation and computation of symbols. Importantly, these symbols can be both internal and external. In the example above the computation is modelled between the relevant symbols that constitute the computational system whether they are in the head or in the world (e.g. on a piece of paper) (Wilson and Clark (Forthcoming) p. 11-12). According to Clark, the example involving Ada is an exemplary case to analyse in wide computational terms. We can analyse Ada’s brain and body together with the pen and paper as constituting one single wide computational system. To understand exactly why this is an example of an extended functionalist approach, we have to take a look at the two key steps involved in arguing that something is a wide computational system. (1) The first step is to appeal to a coarse-grained computational/functional role that supports the ascription of the cognitive state or activity. In the multiplication example this function is simply transforming the inputs of 675 and 765 to the output 516375. In the case of Ada it is a more complex accountancy functional role, but still within the grasp of common knowledge among ordinary human agents. (2) The second step is to investigate, at a more fine-grained level, what it is in the particular case that realizes this functional role and then argue that part of this is played by non-neural processes (Clark, personal correspondence). In the case of Ada, Clark suggests, that an investigation of actual flow of processing reveals that the functional role played not only by processes internal to Ada’s brain, but additionally by parts of the body and the environment. - 74 - Note that the first step in the analysis above seems to be identical to what we find in standard commonsense functionalism: What defines the cognitive process is its commonsense functional profile (a relation between inputs, inner transitions and outputs). It is on the second step that the insights of the research in embodied cognition make their entrance. The main intuition is that if mental states are functionally defined it seems arbitrary to stop looking for the realization of this coarse-grained functional role at the boundary of the skull. As we saw Wilson put it “[W]hy think that the skull constitutes a magic boundary beyond which true computation ends and mere causation begins?” (Wilson (2004) p. 165). The argument is that the difference between being part of the brain or not does not in itself make a relevant difference to whether it plays a computational role or not. Where it gets done does not matter for identifying what it is that gets done. This view leads the way to what has become known as the Parity Principle: “If, as we confront some task, a part of the world that functions as a process which, were it done in the head, we would have no hesitation in recognizing as part of the cognitive process, then that part of the world is (so we claim) part of the cognitive process.” (Clark and Chalmers, 1998 p. 8) At this point we begin to see in what sense it is that these arguments, for extended mind and wide computational systems, are extended functionalists approaches. The extension simply lies in recognizing that there is nothing in the functional description that limits the location of the realization to the boundary of the brain and then arguing that there are in fact cognitive systems that extends this boundary. The following quote illustrates this point: - 75 - “Just as, for the standard approach, we need not care (within sensible limits) exactly where within in the brain a given operation is performed, so too (it might be urged) we should not care whether, in some extended computational process, a certain operation occurs inside or outside some particular membrane or metabolic boundary.” (Clark (In Press) p.17) In this way, it now seems clear that what Clark has in mind when he uses the term extended functionalism: Extended functionalism is an argumentative extension of general commonsense functionalism. Clark argues that this line of argumentation is also present in other arguments from the research area of embodied and embedded cognition. Another example suggested by Clark is Ballard et al’s computational model of vision. 3. 2 Ballard et al.: The computational role of vision To refresh; Ballad et al’s experiments showed that subjects, when performing the task of copying a structure of coloured blocks, used a surprisingly high number of rapid saccades to the original block structure. To Ballard et al this suggested that people, with each saccade, merely took in the information needed for the immediately following action. This information could be either the colour or the position of a block, but not both. This indicated that the subjects were using fixation as a kind of “deictic pointing” device, marking locations where specific information was available. To test this hypothesis Ballard et al. developed further experiments where they, while the subject was looking else where, changed the colour of some of the still uncopied blocks in the original. In most cases this change was not noticed by the subjects and this was taken to confirm the hypothesis that only small bits of - 76 - information, was in use at a certain point. This, in turn indicated that vision was used as a form of information retrieving or recruitment device. The subjects were employing a strategy that relied upon the information being available in the world when needed. According to Clark two points are especially worth noting about this account: 1. It shows that visual fixation plays an identifiable computational role (In the sense that repeated fixation provide the specific information when needed for guiding action.) 2. The repeated saccades gives the subject an opportunity of minimizing the requirements allocated to the biological memory7 ( In this way the world can work like an external storage space for information (Clark citing Wilson 2004)) Having these key points in mind a parallel of this example and the Ada example begins to be visible. The best way to capture this parallel is by illustrating how Ballard et al’s account also tracks Wilson’s notions of explorative representation and wide computation. According to Ballard et al.'s theory, eye movement is a sort of pointer that directs our attention in the world, thus making the world like a store house where we can go and pick out information when we need it - we recognise this as a form of exploitive representation in Wilson’s terms. Additionally, according to Ballard et al bodily actions in the form of fixations plays a computational role and thus in part constitute the overall cognitive process of copying blocks. In Clark’s terms: “Bodily actions are thus part of the means by which certain computational and representational operations are implemented.” (Clark (In Press) p. 11) In this way the theory recognizes the computational system 7 It is worth noting that at this time we cannot conclude that it is not the case that biological memory also encodes these details. What is present in the experience is does not necessarily reveal what is represented in the biological memory or not (Henderson (Forthcoming)). - 77 - carrying out the block copying task as wide in the sense that it extends beyond the brain of the individual. Again, this can be interpreted as a functionalist approach according to Clark, because; “what makes the cognitive process the one that it is is simply its functional profile (the set of state transitions mediating input and output.)” (Clark (In Press) p.11) Additionally, it is an extended functionalist approach because it recognises this functional profile as belonging to “not to the neural system and its inputs and outputs alone but to the whole embodied system located in the world.” (Clark (In Press) p.11) 3.3 Dispositional beliefs and extended functionalism So far all the presented examples have exclusively involved problem solving, but what about beliefs, desires and so forth? To expand the claim to mind on a more general term, Clark and Chalmers gives an example of how a dispositional belief can have an extended realization base. Compare the following two situations: Situation 1: Otto suffers from Alzheimer’s disease and to store important information he has a notebook which he always caries with him. One day he hears about an exciting new exhibition at the Museum of Modern Art and decides to go and see it. He consults his notebook which contains the information that the museum is on 53rd Street and goes to the museum. - 78 - Situation 2: Inga, who has a normally functioning memory, has also heard about the exhibition at the Museum of Modern Art and decides to go see it. She recalls that the museum is on 53rd Street and goes to the museum. According to Clark and Chalmers, these two situations are analogous with respect to belief. In the second situation it seems clear that Inga has the belief that the museum is on 53rd Street. She goes to 53rd Street because she wants to see the exhibition and believes that the museum is on 53rd Street. Additionally, we would also normally acknowledge that Inga had this belief all along, even at the time before she consulted her memory and the belief became occurent. According to Clark and Chalmers, the case is the same with Otto. He also goes to 53rd Street because he believes the museum is at this address. Otto and his notebook make up a coupled system, where the notebook has a function that is similar to the function that the biological memory has for Inga. Thus, it seems reasonable, according to Clark and Chalmers, to claim that Otto also had this belief prior to consulting his notebook. “To say that the beliefs disappear when the notebook is filed away …[would be] to miss the big picture in just the same way as saying that Inga's beliefs disappear as soon as she is no longer conscious of them.” (Clark and Chalmers (1998) p.13) There may be differences between the case of Inga and Otto, such as the time it takes to retrieve the information. However, these differences, Clark and Chalmers argue, are irrelevant to what it is to have a belief. Whether it takes more or less time to retrieve the information does not change the character of the belief. Whether Inga has a fast or a slow memory does not change the fact - 79 - that she has the belief that the museum is on 53rd street and so neither should it do so in the case of Otto. What makes some information a belief, according to Clark and Chalmers, is the role it plays, that it is there when needed, available to consciousness, available to guide action and that it is trusted by the subject (Clark and Chalmers (1998) p. 13). Again this can be seen as and extended functionalist approach. The first step of the analysis appeals to commonsense functional roles in identifying the mental state in this case the dispositional belief. This is a coarse-grained description where certain differences are abstracted away from, such as for example the difference in processing time. Again the second step involves recognizing that Otto’s note book play the same role as Inga’s biological memory. We have now seen in what way Clark interprets these arguments for extended mind as extended functionalist approaches. Our next concern is the view of the relation between body and mind that such an extended functionalist approach yields and how this relates and contrasts to the view on these issues associated with the Body-detail model. 3.4 Extended Functionalism and the Body A similarity between an extended functionalist approach and the body-detail model is that the extended functionalist account also argues that there is a sense in which the body makes an important contribution to mind and that brain algorithms involve exploiting bodily structures and opportunities. This is what is shared by the accounts and marks them both as embodied accounts. Where they come apart is whether this contribution is special in such a way - 80 - that only a body with certain characteristics (the ones that we humans typically have) can accommodate a humanlike mind (The Body-detail model) or whether the body is (only) a part of a larger organizational structure which determines mind (Extended Functionalism). Extended functionalism, in Clark’s words, depicts the body “as just one element in a kind of equalpartners dance between brain body and world, with the nature of the mind fixed by the overall balance thus achieved” (Clark (In Press) p. 25) As a consequence of assigning the body to this more modest position of not being the sole determinant for mind, it is consistent with extended functionalist approaches that the very same mind can exist in bodies with very different properties. As a first example we can take the example of Otto and Inga from before. They share the same kind of mental state even though the brain of Otto works differently from that of Inga. This is because there is something in Otto’s environment that can substitute for having a biological memory. Illustrating a similar point, but with focus on the whole body and not just the brain, Clark invites us to consider the example of Adder. Adder is an intelligent being with a snakelike body. Adder has the advantage of lying on top of a very advanced touch-screen like environment. By moving its body in a particular way e.g. by wriggling the tale at a certain speed, Adder is capable of causing certain external symbols to appear in a certain area of the screen. At the same time Adder is able to read these external symbols owing to a kind of Braille. Now picture Adder using this device to do the same accounting job as we saw Ada do earlier. According to Clark it seems reasonable to say that the cognitive states that are relevant for carrying out the accounting are similar in the case of Adder and Ada. The same abstract functional description of the accounting task can be attributed in the two situations. It might even be the case that when we investigate the realization base the two situations exhibit the same labour division between internal and - 81 - external processes. Clark further suggests that we could even imagine another creature that would be able, by internal computation, to do what Ada and Adder use hands and touchpad screens for. Taking its point of departure in a commonsense description the extended functionalist account does not give into a body detail kind of chauvinism. On this account differences in body do not entail differences in mind. In fact, according to Clark extended functionalism leaves even more room for multiple realization than ordinary (commonsense) functionalism (Clark (In Press) p. 25). The possibility of realizations being wide in Wilson’s terms opens up for the possibility of a larger variation in the organization of the complete realization. 3.4.1 Body as a functional notion As a solution to the tension between extended functionalists accounts and views that apply the body-detail model, Clark (In Press), in the concluding remarks, suggests viewing the body itself (at least in scientific contexts) as something that is defined by a functional role. 8 According to this view, what counts as the body is whatever plays a certain complex role in a larger, unified information-processing economy. As a label for this position I first considered extended, extended functionalism. However, this label has the unwanted effect of glossing over the fact that the second extension is a different kind of extension than the first. The first extension as we have seen is an argumentative extension. The second extension is a question of making the realization base of the commonsensical functional role functionally identified as well. Thus a less catchy, but more appropriate label might be: Extended Body Functionalism. 8 Clark offers this as a solution to the tension between what he calls the Larger Mechanism Story and the Special Contribution Story. As noted arguments in favour of the Larger Mechanism Story are the ones that have an extended functionalist approach. Arguments in favour if the Special Contribution Story are the ones that we have survey under the label of the Body-detail model - 82 - What is the functional role that identifies the body according to Clark? Answering this question Clark takes his point of departure by observing that the body is the first point where willed actions meet the world: It is the point where I turn my thought about having a cup of tea, looking out the window or climbing Ben Nevis into action. At the same time it is also the point where we receive sensory inputs and where the flow of sensory inputs change as a consequence of willed actions. Thus the body has the role of being the “common and persisting locus of action and sensing". (Clark (In Press) p. 23) This is connected with our sense of bodily presence and location in the world. Clark refers to research, on the issue that has become known as telepresence, which has shown that sense of presence is closely connected to an ability to interact with the environment. It is the ability to act upon the objects in the close environment and the fact that such willed sensorimotions results in certain expected new sensory inputs, that gives one a sense of presence in that environment. As an example of this, Clark, in his book “Natural Born Cyborgs”, mentions the experiences reported by people operating the kind of tele-operating system developed to handle radioactive or other dangerous materials. Here the operator is able to control a robotic device by wearing special gloves that transmit movements of the operator’s hands to the robot and tactile feedback back to the operator. Operators have reported to feel a shift in the point of view as if they were present in the environment containing the dangerous materials (Clark (2003 p. 92)). Interesting example that illustrates this point is the Japanese researcher Hiroshi Ishiguro experiences with his newly built robot “twin”. The robot looks just like him and it is even capable of simulating the idiosyncratic characteristics of his gestures. He can control this robot in real-time and thus via the robot interact with, for example, a lecture environment kilometres away. Ishiguro controls the robot from a standard control room containing two screens showing the view from the robot’s perspective. From here he can control its movements - 83 - and he can speak to the audience via a microphone. In a recent interview he reports “When I am tele-operating it is interesting what happens. The true situation is that I’m controlling a robot, but I feel something. It’s as though I am there. If I look at the robot’s body from this viewpoint, I feel as if the Robot’s body is my body. If someone walks up and touches the robot face, I feel something” (Interview by Alun Anderson (July 2007)). In addition to being the point of sensorimotor confluence the body, according to Clark, has the further important role of being a “gateway to intelligent offloading”. This is the role that was especially emphasised by Ballard et al. in their analysis of fixation. The body gives us a possibility of exploiting the fact that information can be stored directly in the world. In Clark’s term; "The body's role in such cases is that of an instrument enabling the emergence of a new kind of information processing organisation." (Clark (In Press) p. 23) According to Clark, the body should thus be identified with whatever plays “the role of being the locus of willed action, the point of sensorimotor confluence, gateway for intelligent offloading and the stable (though not permanently fixed) platform whose features and relations can be relied upon in the computation of certain information-processing solutions.” (Clark (In Press) p.24) Notice that this is a rather abstract characterization of the functional role of the body, which abstract away from structural differences such as chemical and physical differences. 3.4.2 The negotiable body Viewing body as a functional notion has some interesting implications. In this light Clark, suggests: - 84 - “ It is merely a contingent (and increasingly negotiable) fact about human embodiment that the body is both the metabolic centre and the bridge to sensory presence and intelligent action.” (Clark (In Press) p.23) Viewing body as a functional notion opens up for the possibility of something counting as a body without it being “a single persisting body in ordinary 3space”. (Clark (In Press) p.24) This presents new possibilities of having genuine embodiment that is spatially scattered or genuine embodiment in virtual or mixed realities. In our example of Adder from before we could perhaps count the touch screen as being part of Adder’s body. In less imaginative cases it could be that we would say that a blind person’s cane, being a point of sensorimotor confluence, would count as a genuine part of that person’s body. In Clark’s slogan formulation “the body is essential, but negotiable” (Clark (2003) p. 114). It is important to notice that on the extended body functionalist account the body is still important and significant for mind. It recognises the importance of motions, touch, vision etc. and the ability to interact with an environment. It rejects the view of mind and cognition as disembodied. Clark sees it as a resolving the tension between the body-detail model and extended functionalist approaches in the following way: “The cognitive importance of the body…. is fully exhausted by its ability to play a certain functional role in an intelligent organisation. The distinctiveness and importance of this role is what explains the intuition that the body makes a special and pervasive cognitive contribution. But because it [the body] is nothing but a complex functional role, there is nothing cognitively significant about the bodily contribution that is not fully captured by reflection upon its several … - 85 - computational and information-processing contributions.” (Clark (In Press) p. 24) In this way Clark analyses the position as joining the insight from the Bodydetail model that the body is special and the abstract characterisation of mental states from extended functionalism. 3.4.3 Body and Mind – a one-to-one match? The suggestion of viewing body as a functional notion can be seen as an extension of the technique applied by Bechtel and Mundale that we explored in our investigations of functionalism. Bechtel and Mundale introduced the thought that brain states in themselves are rather functional notions defined in a coarse-grained way. What Clark is suggesting is that the case is similar when we are concerned with the body. On Bechtel and Mundale’s account viewing brain states as functional states resulted in a one-to-one mach of brain states and mental states. Do we get a similar one-to-one match between body and mind on the extended body functionalist account? If we look back at our examples of Ada and Adder we see that even though they carry out the same accountancy task, it would not be right, even under the extended body approach, to claim that they have the same kind of body. Just because we count the touchpad as a part of Adder’s body it does not mean that Adder has hands and feet. The consequence of understanding body as a functional notion is that in both cases, what it is that counts as the body is determined by what it is that plays the functional role as laid out above (Clark, personal communication). At this point it seems relevant to ask what it is exactly that we gain from taking the extended body functionalist approach instead of just taking the extended functionalist approach. Clark suggests it as a resolution of the tension of a body-detail model and an extended functionalist approach - A - 86 - functionalist way of ascribing a distinctive and important role for the body, abstracting away from differences on the level of chemistry and physics. I think that this is an interesting suggestion, and I think that it is important to recognise that the intuitions behind the body-detail model are something that a competing position should try to account for. However, as far as the body already has an essential role under the ordinary extended functionalist approach an interesting issue for further research might be how it is exactly that it offers us a reconciliation of the Body-detail model and extended functionalism. As it stands now, I do not see that extended body functionalism provides us with an alternative that is sufficiently different from that provided by extended functionalism to warrant endorsing this position for our present purpose of rejecting the body-detail model. There might be other reasons for endorsing the extended body functionalist approach e.g. to explain the findings involved in tele-presence. However, endorsing this position might also leave one vulnerable to additional objections e.g. it might give rise to objections not face by the extended functionalist approach e.g. that it might be problematic for the position to set the boundary of the body in distinction from the environment. These issues are outside the scope of the current project, but would of course be obvious issues for further research. 3.5 Status We have now seen that there is a way of acknowledging the importance of body in mind and cognition that incorporates the insights of good old fashioned commonsense functionalism. I think that the above reading of the embodied and embedded approach is on the right track and that it manages to show how such an approach and functionalism are compatible. In order to - 87 - further compare this approach to the body-detail model and to elucidate possibilities for further research it is now time to take a step back and gain an overview of the arguments so far. Shapiro and Noë suggested that a fine-grained analysis of the body should be accompanied by a similarly fine-grained analysis of the mental. The suggestion was, as we have seen, that having certain kinds of mental states required having a certain kind of body with certain properties. Shapiro analysed this as conflicting with the spirit of an abstract, body neutral functionalist approach on account of its chauvinistic consequences. I agreed with Shapiro that if his interpretation of the embodied research programme was correct this would indeed be conflicting with the spirit of functionalism, however, not only with abstract and coarse-grained versions of functionalism, but with functionalism as such. The conclusion was that it was in conflict with the spirit of even fine-grained functional approaches. Not because of the chauvinistic character of the conclusion that it reached, but on account of it foreclosing the possibility of alternative implementations. In this connection, I argued that there seem to be good reason to think that such alternatives could be available. I argued against the Body-detail model, demonstrating that for each of the research projects presented by Shapiro the strong reading that one has to have a humanlike body to have a human mind was not justified. In this light it was possible to reject Shapiro’s claim that research in embodied and embedded mind undermines body neutrality and abstract functionalism. This rejection paved the way for a reading of the research of embodied, embedded mind as compatible with functionalism. With Clark I suggested a way in which this might be done. It rests on the observation that some arguments for viewing mind as embodied and embedded has an extended functionalistic approach. - 88 - These are the arguments in favour of viewing the body as part of a larger mechanism consisting of brain, body and environment and which overall operating profile determines our mental states. They have a functionalistic approach because in these arguments what defines a mental state is a coarsegrained functional role formulated in terms of commonsense psychology. The extension of the functionalist approach comes from realising that a functionalist definition of mental states does not in any way require that the realization of such states is limited to what goes on in the brain. Being part of a bigger system, the body under the extended functionalist interpretation has an important, but still more modest role than the one it has under the bodydetail model. This has, as we saw, the consequence that sameness of body is not required for sameness of mind. In fact, even the possibilities for multiple realizations are extended under this approach. In this way we escape the species chauvinism that would be a consequence of the body-detail model. Whether the extended functionalism should be accompanied by the further assumption that the body is functionally defined has been left as an open question to be determined by further research. The phenomenal feel Having the arguments fresh in memory, makes us vary of a difference between the two lines of arguments that we so far have left unnoted. The examples suggested as support the body-detail model focus on experience such as, for example, the experiences of hearing a sound, whereas the extended functionalists, in contrast, focus on cognition. The experience of the taste of chocolate has a different feel to it than the experience of the taste of whisky. An ache feels different from an itch. The arguments presented by the extended functionalist approach all focus on cognitive states and activities e.g. copying a set of coloured blocks or carrying - 89 - out advanced calculations. Such states and activities do not seem to have a phenomenal feel to them. Neither does the having of beliefs. It does not feel different to believe that “the museum is on 53rd street ” than to believe that “it rains a lot in Scotland”. Having such beliefs might be associated with experiences that has a certain feel to them, but there is in it self nothing it is like to have these beliefs. In contrast to the extended functionalist approach we saw that experiences and the phenomenal feel associated with these constituted an important issue for Noë.9 Could it be that experiences are determined by the characteristics of our body whereas cognitive states are not? Could it be that the body-detail model is true of experience only: that is, having humanlike experiences with a certain phenomenal quality takes a humanlike body? Does the extended functionalism ignore the way that having a certain body shape influence the phenomenal quality of experience? When considering this question we are of course not concerned with the cases where experiences are about the body shape. In such cases it is trivially true that body shape will make a difference to the experience. If type of phenomenological quality is tied to body type, where as, in contrast, cognitive processes can be similar even in creatures that have different bodies. This would imply that the role of the body is unique when it comes to experience, whilst when it comes to cognition the environment plays a significantly bigger role. Consider the difference in the experience of driving a Porsche and a tank. There is a difference in the way that it feels to shift gears and turn corners in these different vehicles. My immediate intuition is that there could be a similar difference between experience of having a snake body and a human body. 9 The issue of experiences and phenomenal feel does not have the same importance for Shapiro, who in his conclusions of The Mind Incarnate mentions leaves this an open question. - 90 - A test case for this intuition could be the example of tactile vision substitution in comparison with ordinary vision. Do people with TVSS really see? Do they their experiences have the same phenomenological quality as ordinary seeing? Noë takes the following approach to this question: “Tactile vision [is] vision-like because (or to the extent that) there is…. an isomorphism at the sensorimotor level between tactile vision and normal vision. In tactile vision, movements with respect to the environment produce changes in stimulation that are similar in pattern to those encountered during normal vision. The same reservoir of sensorimotor skill is drawn on in both instances.” (Noë 2004 p. 27). From the last two sentences in the above quote one get the impression that Noë equals vision and tactile vision, but here we have to be careful. Note that the first sentence only states that tactile vision is vision-like. According to Noë there is an extent to which there are similarities between tactile vision and normal vision also on a sensorimotor level, however on this level it is also possible to spot where the isomorphism breaks down. According to Noë the sensorimotor view places constraints on the degree of similarity that is required to achieve similarity of experience. Where vision and tactile vision come apart, according to Noë and the sensorimotor account is the ability enabled by the systems to subserve patterns of sensorimotor dependencies. This difference, according to Noë, is due to the differences in the systems: The vibrator array placed on the skin in the tactile vision system is crude and - 91 - simple compared to the retina involved in normal vision. This, according to the sensorimotor account, differences in the qualitative aspect of the experience. I think Noë is correct in noting that the information flow of these systems does not match the speed and flexibility that normal vision provides. However, if we refined the TVSS systems with higher level of sensor input. What then? The question is could it ever be like real vision or would it always lack that certain kind of feel to it? The question of phenomenal quality has been a joker in philosophy of mind for decades and especially in relations to discussions of functionalism. I believe that further research both philosophical and empirical is needed to answer this question fully. In these investigations, however, it is important to remember that just establishing that the body does play an important and distinct role in determining our experiences does not suffice to establish that no alternative implementations are possible. All about levels? When looking away from the question of phenomenal quality, the quote from Noë above illustrates a point that we came across during the exposition of Bechtel and Mundale’s views. This point being that a central question here is about levels of abstraction. Noë rejects the label of chauvinism for the sensorimotor approach on account of it being able to recognize tactile vision as a form of vision at a certain level of abstraction e.g. it can recognize that the same reservoir of sensorimotor skills is drawn on in both instances. According to the analysis of Bechtel and Mundale, the traditional debates concerning mental states and functionalism were conducted without much attention - 92 - being paid to the context of inquiry. The arguments presented in the previous chapters both for the Body-detail model and Extended Functionalism are highly informed by the most recent developments within science in general and cognitive science in particular. However, I still believe that what according to Bechtel and Mundale applies to the traditional debate, could be said to apply to the current debate as well. Paying more attention to context of inquiry might reconcile intuitions and positions that at first seem to contradict each other. On my take the coarse-grained analysis of the extended functionalist approach is not necessarily exclusive. A fine-grained approach can also be appropriate at times depending on what you are interested in. Recognizing this creates room for recognizing and explaining our intuition that differences in our bodily construction makes a differences to mind. Does this mean that there is no fact of the matter about whether some alien or alternative organization ‘really’ implements the same mental states as our own? This question lies outside the scope of this thesis, but would be an interesting focus for further research. - 93 - ______________________________________________ CONCLUSION F ______________________________________________ My goal in this project has been to explore the logical space between functionalism and embodied, embedded approaches to mind. In this connection my focus has been to evaluate the claim made by Shapiro that embodied, embedded approaches undermine a functionalist program. What this project has shown is that we should reject the strong reading of the embedded, embodied approaches that would yield such a conflict with functionalism. Although, it seems reasonable on the basis of the research presented by Shapiro to conclude that the body plays a very important role in cognition, this does not entail that the characteristics of body type are indispensable. As stated, wearing a pair of glasses can make up for weak eyesight and research on sensory substitution and brain plasticity gives us reason to believe that alternative implementations could be made available. Our embeddedness in a language using community is a further reason as to why characteristics of body do not determine type of mind. The rejection of the body-detail model paved the way for a reading of the embodied, embedded approach that recognizes that gross bodily states and processes do play a role in determining mental states, but at the same time reject that they play a unique and ineliminable role. Under this reading type of mind is not tied to type of body and embodied, embedded approaches and abstract commonsense functionalism goes hand in hand. The question of whether experiences pose special cases has been left to be determined by further research, but as stated for this asymmetry to be justified it would have to be - 94 - shown not only that the body plays a distinct and important role in determining our experiences but that it plays a unique and ineliminable role. - 95 - ______________________________________________ BIBLIOGRAPHY F ______________________________________________ Anderson, A (July 2007), "Am I My Android?" New Scientist 195 ( 2614):46-47. Armstrong, D. (1968), A Materialistic Theory of the Mind. London: RKP. Bach-y-Rita, Paul, and Stephen Kercel (2003), "Sensory substitution and the humanmachine interface", Trends in Cognitive Science 7 (12):541-546. Ballard, D.H., M.M. Hayhoe, P.K. Pook, and R.P.N. Rao (1997), "Deictic codes for the embodiment of cognition", Behavioral and Brain Sciences 20 ( 4):723– 767. Bechtel, William, and Jennifer Mundale (1999), "Multiple Realizability Revisited: Linking Cognitive and Neural States", Philosophy of Science 66 (2):pp. 175207. Block, N., and J. Fodor (1972), "What Psychological States Are Not", Philosophical Review 81:159-181. Block, Ned (1980), "Troubles with Functionalism", in Ned Block (ed.), Readings in the Philosophy of Psychology Cambridge, MA: Harvard University Press. Braddon-Mitchell, David, and Frank Jackson (2007), Philosophy of mind and cognition. 2nd ed. Malden, MA: Blackwell Pub. Churchland, P.M., Ramachandran, V., and Sejnowski, T. (1994), "A Critique of Pure Vision", in C. and Davis Koch, J. (ed.), Large-Scale Neuronal Theories of the Brain, Cambridge: MIT Press, 23-60. Clark, A. , and D. Chalmers (1998), "The Extended Mind", Analysis 58:10-23. Clark, Andy (2001), Mindware : an introduction to the philosophy of cognitive science. New York ; Oxford: Oxford University Press. ——— (2003), Natural-born cyborgs : minds, technologies, and the future of human intelligence. Oxford: Oxford University Press. ——— (In Press), "Pressing the Flesh: A Tension in the Study of the Embodied, Embedded Mind Philosophy and Phenomenological Research", Philosophy and Phenomenological Research. Collins, H. M., Robert Evans, R. Ribeiro, and M. Hall (2006), "Experiments with Interactional Expertise", Studies in History and Philosophy of Science 37 A/4 ([December]):656-674. Collins, H., J. Shrager, and A Clark ((Forthcoming)), " Keeping the Collectivity in Mind?" in, Special issue of Phenomenology and the Cognitive Sciences, , Evan Selinger (ed), Janet Levin Edward N. Zalta. Damasio, A (1994), Descartes' Error. NY: Grosset/Putnam,. Dennett, Daniel C. (1987), The intentional stance. Cambridge, Mass. ; London: MIT Press. Dretske, F. (1981), Knowledge and the Flow of Information. Cambridge MA: MIT Press. - 96 - Ferrier, David (1886), The Functions of the Brain. New York: G.P Putnam’s Sons. Fodor, Jerry (1974), "Special Sciences, or the Disunity of Science as a Working Hypothesis", Synthese 28:97-115. Glenberg, A. , and M. Kaschak (2000), "Constructing Meaning: The role of Affordances and Grammatical Constructions in Sentence Comprehension ", Journal of Memory and Language 43:508-529. Glenberg, A. M., and D. A. Robertson (2000), "Symbol grounding and meaning: A comparison of high-dimensional and embodied theories of meaning. " Journal of Memory and Language 43:379-401. Grush, R. (2003), "In defence of some ‘Cartesian’ assumptions concerning the brain and its operation", Biology and Philosophy (18):53-93. Haugeland, J. (1995), "Mind Embodied and Embedded", Acta Philosophical Fennica Mind and Cognition: Philosophical Perspectives on Cognitive Science and Artificial Intelligence (58):233-267. Henderson, John M. (Forthcoming), "Eye Movements and Scene Memory", To appear in S. J. Luck & A. Hollingworth (Eds.), Visual Memory. New York: Oxford University Press. Lakoff, G. , and M. Johnson (1999), Philosophy in the Flesh: The Embodied Mind and Its Challenge to Western Thought. New York: Basic Books. Lakoff, G., and M. Johnson (1980), Metaphors We Live By. Chicago: University of Chicago Press. Levin, Janet (2004), "Functionalism", in, Edward N. Zalta (ed): Stanford Encyclopedia of Philosophy, URL = <http://plato.stanford.edu/archives/fall2004/entries/functionalism/>. Lewis, David (1966), "An Argument for the identity theory", Australian Journal of Philosophy (63):17-25 (Reprented) John Heil,2004 Philosophy of mind : a guide and anthology Oxford University Press. ——— (1972), "Pychophysical and theoretical identifications", Australasian Journal of Philosophy 50 (3):249-258. Mandik, P. What's the logical relation between functionalism and multiple realizability? 2007 [cited 15.05. Available from http://philosophy.uwaterloo.ca/MindDict/functionalism.html. Noë, Alva (2004), Action in perception, Representation and mind. Cambridge, Mass.: MIT Press. O'Regan, J.K. , and A. Noë (2001), "A sensorimotor approach to vision and visual consciousness", Behavioral and Brain Sciences 24 (5):939-973. Putnam, Hillary (1960), "Minds and Machines", in S. Hook (ed.), Dimentions of Mind New York: New York University press. , 148-179 ——— (1963), "Brains and Behavior ", in, Reprinted in Mind, language and reality: Philosophical Papers, (Vol. 2). London: Cambridge University Press, 1975. ——— (1967), "The nature of mental states", (originally published under the title Psychological Predicates) Reprinted in Mind, language and reality: Philosophical Papers, (Vol. 2). London: Cambridge University Press, 1975. Shapiro, Lawrence (2004), The Mind Incarnate. Edited by Kim Sterelny and Rob Wilson, Life and Mind: Philosophical Issues in Biology and Psychology: MIT Press. ——— (Forthcoming), "Reductionism, Embodiment, and the Generality of Psychology", in, To appear in a forthcoming volume for Blackwell, H. Looren de Jong & M. Schouten (eds.). - 97 - von Melchner, L. , S. Pallas, and M. Sur (2000), "Visual Behaviour Mediated by Ausitory Coprtex Directed to the Auditory Pathway", Nature 404:871-876. Weiskopf, D. (Unpublished ms), Embodiment and Linguistic Understanding. Wilson, R. A., and A. Clark (Forthcoming), "How to Situate Cognition: Letting Nature Take its Course", in, To appear in M. Aydede and P. Robbins (eds.), The Cambridge Handbook of Situated Cognition. Wilson, Robert A. (2004), Boundaries of the mind : the individual in the fragile sciences : cognition. Cambridge: Cambridge University Press. - 98 -