|
|
|
||||||||
|
||||||||||
|
Outline the single strongest objection to functionalism, and determine whether this objection is sufficient to cancel out functionalism's advantagesFunctionalism identifies mental states with their functional role, in an analogous way to mind-brain identity theory identifying them with physiological states, or behaviourism identifying them with dispositions towards certain behaviour. However, it has advantages over both. It allows for the possibility of a bodiless brain. In theory, a brain that is removed from its body and sustained by technology would still have thoughts. However, as the brain is incapable of behaving in any meaningful way, this is a problem for behaviourism, which relies on direct observation of behaviour in order to identify mental states. Unlike behaviourism, functionalism allows for indirect observation of mental states. This is a perfectly acceptable practice in many branches of science where direct observation is impossible (e.g. palaeontology, forensics or particle physics). In this case just because the bodiless brain displays no observable behaviour, does not mean it is not experiencing mental states. However, the danger of indirect observation is that it can trigger false positives. A spike in the readings of a particle accelerator could be an undiscovered quark or a bug in the software. In the case of functionalism something can be declared to possess mental states when in fact it doesn't. This is our first indication of a major objection to functionalism: liberalism. Another advantage of functionalism is that it is multiply realisable. Whereas the mind-brain identity theory is locked into the necessity of a human type organic brain, functionalism makes no such assumptions. Just as a postbox could still be a postbox if it was yellow and plastic, so a mental state is a mental state as long as it fulfils the requisite functional criteria, whether it is caused by neurones, photons or transistors. However, this again leads to accusations of liberalism. Because it makes no assumptions on the composition of mental states, there is a danger it can be applied to systems that patently do not have them. A postbox could still theoretically function as a postbox if it was made of cardboard, but not if it was a solid lump of metal. What are the theoretical functional limits of mental states? Is there any implicit reason why a pebble, a painting or an electrical discharge could not all possess mental states? If functionalism is too liberal, and encompasses systems that cannot be experiencing mental states, then it loses its worth as a philosophical model. It's clear that if my definition of a rabbit is "something small and furry that hops" it is too liberal, as it will encompass things that are obviously not rabbits. Consider once more our bodiless brain, but extend the idea to that of a disembodied mind with no physical existence whatsoever. Functionalism will accommodate it in exactly the same way as the bodiless brain, even though we have now crossed into dualism. Indeed, functionalism is compatible with both materialism and dualism. This could be considered a positive point - highlighting functionalism's multiple realisibility. It could also be a negative. If functionalism can so easily accommodate such fundamentally different views, what else can it accommodate? Can a postbox have mental states? Functionalism defines a mental state in terms of the functional interactions of its inputs, its outputs, and other mental states. At first, defining mental states in terms of other mental states may seem like begging the question, but in a closed system it is quite acceptable if you imagine you are defining all states simultaneously [URL1]. This is acceptable practice in mathematics, for example in number theory. Just as all numbers are defined in terms of their relation to other numbers, so are all mental states defined in relation to each other. Now, while one plus one is two, and one orange plus one orange is two oranges, this does not imply that oranges are numbers. Thus while a postbox is defined by its functional states and so is a mind, this does not imply that a mind is a postbox. Therefore we can abandon the most extreme (and uncharitable) accusations of liberalism, and must instead try and show that two systems - one sentient, the other not - are functionally equivalent. Our first example is the philosophical zombie. A philosophical zombie is simply a human (or humanoid) without consciousness, but with an otherwise complete mental capacity [URL2]. If you entertain the idea that you are the only real person who exists, then all other people must either be your delusions, or these philosophical zombies. The zombie accumulates experiences, as we do. Experiences are compared to situations in order to consider possible courses of action. Behaviour is initiated with the aim of bring about a desired outcome. We have all the requirements for a functional definition of a mental state, without any recourse to consciousness or experience of qualia. Yet these zombies are not experiencing subjects of a life. This is an example of the problem of "absent qualia". Qualia are the personal experiences we have of perceptions. The quale of "redness" is the experience I have of a red object that differentiates it from an otherwise identical green object. The fact that I experience redness differentiates me from a camera that merely sees redness. Intuitively we expect that in order for a system to be sentient it must experience qualia, yet systems that do not can be functionally equivalent to those that do. Consider also the China Brain argument, put forward by Ned Block to illustrate this problem [p73]. Imagine that China decides to set up an experiment whereby its billion-odd citizens all behave as neurones in a simulated brain, connected by radio to a brainless body. If the experiment is set up carefully enough, then the animated body is functionally identical to a real person. However, it seems inconceivable that China (rather than its constituent citizens/neurones) is consciously experiencing mental states. Recall our philosophical zombies. Notice that a computer with any sort of learning capacity will meet the same basic criteria of accumulating experiences that impact the relationships between inputs and outputs. Does this mean that they too have mental states? If so, at what point does a computer stop being an adding machine, and start having mental states? Functionalism will embrace artificial intelligence as easily as it encompasses materialism and dualism. The Turing test is a test for artificial intelligence. If a machine's responses to questioning cannot be distinguished from those of a human, even by an expert observer, then that machine can be said to have intelligence. As inputs and outputs match for machine and person, and the functional role for each is identical (in the given setup), the machine and the human are functionally equivalent. However, the machine is an artificial, non-conscious system. Thus, any positive results will be a false positive for functionalism. The Chinese Room is one such hypothetical Turing Test devised by John Searle [p103]. A person who knows no Chinese is sealed in a room with a number of Chinese symbols and a detailed instruction manual on how to formulate correct responses (in Chinese) to given questions. If the instruction manual is detailed enough, and the experiment is set up with enough care, then this is functionally identical to a Chinese-speaking person sealed in a room forming responses from their own knowledge. Searle proposed this thought experiment with the intention of disproving strong AI and questioning the Turing Test, by showing that at least one system that passes the test cannot have intelligence. The same argument applies to functionalism. Given inputs during similar states produce similar outputs. The functional role - that of conversing in Chinese - is identical. It is difficult to see how the two systems are not functionally equivalent. Yet it seems clear that the room - rather than the person sealed within - cannot be experiencing mental states. The room is not conscious. It is not experiencing qualia. To counter these objections, functionalism needs to draw a distinction between systems with consciousness and that experience qualia, and those without and that do not. One tactic is to deny that there are qualia at all, or to reduce their importance to a trivial detail. This obviously jars with our common-sense view of our own mentality, but also it does not help to defend against the objection. There is no way that eliminating a requirement of a mental state is going to reduce the number of false positives. If a police enquiry was seeking male caucasian smokers, but were getting too many false leads because they couldn't tell a smoker from a non-smoker, they are not likely to eliminate the requirement of "smoker". A variation of this approach is to claim that qualia are naturally occurring by-products of mental states [AC6]. This is no better. Footprints are naturally occurring by-products of furry things that hop, but the footprints of a rabbit and a kangaroo are different enough to show that they are different creatures. A better alternative is to try and explain qualia in functional terms. This way, a system with qualia becomes functionally different from one without. This at last might focus the range of systems that functionalism can be applied to. So are qualia functional? Virtually all our responses to colour, for example, are based not on our actual experience of the colour, but our associations with it. Red means "danger", or "stop". Green means "go" or "environmental". If, when I look at a red traffic light, I experience what you would consider the experience of "green", I am still just as likely to stop. This is illustrated by considering a monochromatic world where colours are identified using labels. We know something is red, because there is a label attached saying "Red". In such a world, our actions and interactions are functionally identical to those we would have in a world of colour, yet without the experiences. This implies that qualia cannot be functionally described. However there may be situations in which qualia are functionally relevant, for example in the cases of style, dress sense or colour co-ordination. Our favourite colour is chosen for our experience of it. Whether one colour clashes with another is a matter of the experience of the combination. One person may look at a bright pink shirt and love it. Another may look at it and be appalled. Assuming the difference in reaction is a matter of taste rather than associations, then we have an example where our inner experience of a colour affects our behaviour. This implies that qualia can be functionally described. So it is clear that qualia present a problem. Sometimes they can be functionally described, and sometimes they cannot. It does suggest that some of the false positives can be explained away by looking at the functional role of the qualia - considering them as a kind of sensory input perhaps. However, there will be others in which qualia still present a major qualitative difference, but not a functional difference. Liberalism presents a strong objection to functionalism, and is almost sufficient to cancel its advantages. A functional definition of qualia can go some way to repair this deficiency, but it still leaves us with functionalism ascribing mental states to systems that have none. Bibliography
|
|
Except where stated, this site and its contents are copyright ©2009 Alec Harkness and licensed under a Creative Commons Attribution-Share Alike 2.5 License.