Wittgenstein on Doubt

Ludwig Wittgenstein’s case against scepticism (or at least against global scepticism) is simple. We can't doubt without exempting some things from doubt. As Wittgenstein himself puts it in his On Certainty (##341-4.)

“The questions that we raise and our doubts depend on the fact that some propositions are exempt from doubt, are as it were like hinges on which those [doubts] turn.

“That is to say, it belongs to the logic of our scientific investigations that certain things are in deed not doubted…

“My life consists in my being content to accept many things.” 

To put this at its simplest. 

Say that you're doubting a person’s thesis in geology. You wouldn't, thereby, also doubt the meanings of your own words or the words of the person who's offering his geological thesis. That would be semantic doubt, not geological doubt.

Similarly, you wouldn't doubt that the geologist were a person rather than a zombie or a machine. That would be a doubt about other minds, not a doubt about (again) geology.

Even if the other doubts aren't philosophical, they still needn't be geological doubts. 

For example, you may doubt the geologist’s honesty or why he's saying what he's saying. (You may doubt that you put your underpants on. If you did, then perhaps you wouldn't pay attention.) Thus, these doubts must be (as the philosopher David Lewis once put it) "properly ignored". 

What's at the heart of these "exemptions" is the "context" in which the doubt (or the exemption of doubt) takes place. As Wittgenstein (again) puts it:

“Without that context, the doubt itself makes no sense: ‘The game of doubting itself presupposes certainty’; ‘A doubt without an end is not even a doubt.’” (On Certainty, #115; #625.)

If one doubts everything, then there's no sense in doubting anything. Doubt occurs in the context of non-doubt. 

According to Descartes, one thing one can't doubt is that one is doubting. (Or one can't doubt the meanings of one’s words or that one’s words mean the same today as they did yesterday.) Even psychologically speaking, one needs a context for one’s doubt.

The Things We Cannot Doubt

The important point to make about Wittgenstein’s position is not that, as Timothy Chappell puts it, 

“there is some special class of privileged propositions that we simply can’t doubt”. 

This isn’t a Cartesian or "foundationalist" position. The propositions we mustn't doubt could be of (just about) any kind. The general point is that there must be some propositions (of whatever kind) which we mustn't doubt in order to get the ball rolling. We can't start ex nihilo - as Descartes ostensibly did. We must bounce off certain propositions which we don't doubt. We can't doubt, then, literally everything - again, as Descartes supposedly did.

What we choose not to doubt (indeed what we also choose to doubt) will depend on our context. That will determine the nature of our doubts (or our lack of doubt vis-à-vis particular propositions or possibilities). 

Timothy Chappell gives some very basic, and non-philosophical, examples of this. He writes:

“[I]n each context, there is a very great deal that is not in doubt: the existence of the chessboard, the reliability of the atlas, the possibility of generally getting shopping sums right. This background makes it possible to have doubts, and possible (in principle) to resolve them. Where there is no such background, says Wittgenstein, the doubt itself makes no sense.” 

We can create a table of what we can't doubt, and what we can doubt:

1a) The existence of the chessboard. 
1b) The sincerity of our chess opponent’s naivety.

2a) The (general) reliability of the atlas. 
2b) Whether or not the atlas is up-to-date.

3a) The possibility of (generally) getting our shopping sums right. 
3b) That one’s hangover (today) is affecting one’s arithmetical judgement.

To put the above another way. One couldn't doubt the sincerity of our chess opponent’s naivety if before that we actually doubted the existence of the chessboard. We wouldn't doubt whether or not our atlas was up-to-date if we had already doubted its general reliability. We wouldn't doubt our arithmetical skills during a hangover if we had already doubted our skills in all contexts.

Not only that: we can only resolve our lesser doubts if we simply disregard the more global (or extreme) doubts which might have proceeded them. That is, I can go ahead and win my chess opponent only if I simply disregard the possibility of the chessboard simply not existing in the first place.

Wittgenstein also seems to say that total (or global) doubt simply “makes no sense” because there needs to be a reason to doubt. If one doubts everything, then there can be no reason to doubt at all – unless the act of doubting (everything) is itself the reason to doubt! Perhaps the sceptic would concede that (according to Wittgenstein) senseless position!

Descartes’ Fallacy?

Chappell then offers us a logical argument against Descartes’ global or total doubt. He argues that it rests on a fallacious argument. He writes:

“Descartes – you could say – begins his philosophy by arguing that since any of our beliefs might be false, therefore all of our beliefs might be false. But this is a fallacious argument. (Compare: ‘Any of these strangers might be the Scarlet Pimpernel; therefore every one of these strangers might be the Scarlet Pimpernel.’) What is true of any belief is not necessarily true of every belief. So – the claim would be – Descartes’ system rests on a fallacy (the ‘any/all fallacy’, as it is sometimes called.)”

In fact Chappell's argument does seem to follow. That is, “if any of our beliefs might be false, therefore all of our beliefs might be false”. He isn't saying that all are false if one is false; but that all of them may be false if one is (found to be) false. On the other hand, perhaps that doesn’t logically follow. 

One belief (or “any” belief) being false doesn't entail every belief being false, or even their possibly being false. However, doesn’t it leave open that possibility? 

The analogy with the Scarlet Pimpernel doesn't work because, by definition, only one person can be this person. There's nothing strange about saying that every (or all) our beliefs may be false - or even that they are all false. Not all our beliefs need to be numerically identical. However, there can only be one other person who is numerically identical with the Scarlet Pimpernel. 

So saying that

“any of these strangers might be the Scarlet Pimpernel; therefore every one of these strangers might be the Scarlet Pimpernel” 

isn't the same as the Cartesian example at all. Two beliefs may both be false. However, they needn't be identical beliefs. On the other hand, if there were two people who were the Scarlet Pimpernel, then they'd need to be identical – indeed numerically identical.

The Language Game of Scepticism

Wittgenstein brings in his notion of language games to make sense of global doubt. Again, his argument against doubt is simple. That argument is that philosophical (or sceptical) doubts simply don't arise in any of our language games (outside philosophy!). Therefore we should ignore them! Chappell writes:

“The trouble with crazy sceptical hypotheses, according to Wittgenstein, is that they don’t crop up in any of the various language games that make up the texture of ordinary life in the world. That is why it doesn’t make sense to discuss them.” 

This means that “crazy sceptical hypotheses” don’t have any context. If they have no context (outside philosophy!), then “it doesn’t make sense to discuss them”. However, the septic (or philosopher) may reply:

So what! I don’t care if scepticism has "no context" or if there's no sceptical "language game". What I'm saying may still be legitimate and even true! In any case, why can’t scepticism (or philosophy generally) itself be a language game?

After all, philosophy is a language game (if we insist on using Wittgenstein's words) which has been played for over two thousand years. And scepticism itself has been an important and influential language game in our culture generally. What better example of a language game could you have?

Moreover, does scepticism only exist in the language game of philosophy? What about the many conspiracy theories that are so much a part of culture in the U.K and the U.S? These theories can be deemed to be sceptical in nature – after all, they distrust the truths of the “Establishment” or the “status quo”, just as the philosophical sceptics (in part) did.

In addition, shouldn’t a Wittgensteinian say that the very fact that that “crazy sceptical hypotheses” have been discussed at all means that they must have been discussed in one (or various) language games? Every discourse - crazy or sane - needs its own language game. Indeed, wasn’t that one of Wittgenstein’s points about language games?

Despite saying all that, Chappell states that 

“the sceptic isn’t playing any legitimate language game in his discourse, and so is talking nonsense”. 

Again, who says that the sceptic isn’t playing a language game? And who says that if the sceptic is playing a language game, then his language game isn't "legitimate"? Is it because it's not the language game (or language) of the ordinary man speaking "ordinary language"? The sceptic may again say:

So what! Why should I care about ordinary language or the ordinary man?

So I’m not sure why - or how - Wittgenstein excluded scepticism from all language games or denied that it's a legitimate language game. Chappell too appears to agree with this position against Wittgenstein’s chauvinism against the sceptical language game. He writes:

“[S]ince the sceptic’s discourse makes sense, it must be part of a Wittgensteinian language game – a particular form of human linguistic activity with its own rules – called the ‘scepticism game’.” 

Perhaps Wittgenstein might have replied:

But that’s where you're wrong! The sceptic’s discourse doesn't make sense. It's meaningless. It's meaningless precisely because it's not ordinary language. (It doesn't use accepted terms in the way that we use them in everyday life.) Therefore, the sceptic’s discourse doesn't make sense. It's nonsense.

It's certainly true that sceptical “linguistic activity” does indeed have “its own rules”. Indeed it can hardly not do. And because it does have its own rules, then it must also be a bona fide language game. However, it just happened to be a language game that Wittgenstein didn't like. (Just as William P. Alston – in his paper 'Yes, Virginia, There Is a Real World' - likes religious language games, though he doesn't like the language games of "relativism" or "scientism".) If we truly believe in Wittgensteinian language games (that is, in their existence and autonomy), then we simply can't pick and choose which ones we accept and which ones we reject. If it's a “human linguistic activity with its own rules”, then it's a language game. (That's whether or not we like it or agree with its beliefs or theories.) Indeed, according to the theory of language games, it's irrelevant if you or I (who belong to other language games) agree or disagree with other language games (to which we don’t belong). After all, all language games - almost by definition - are autonomous and thus beyond the criticisms of other language games. That is the truly relativistic aspect of Wittgensteinian languages games. And that's despite the fact that Wittgenstein himself - and many others - mightn't have liked the relativist language game itself.

Functionalism Applied to Life

Believers in Strong AI believe (to put it very simply) that if computers behave in certain ways, then they have intelligence and even minds.

Not only can you be a functionalist when it comes to the mind, you can also be a functionalist when it comes to life itself.

According to John Horgan, Christopher Langton, of the Santa Fe Institute,

“described himself as a functionalist, who believed life was characterised by what it did rather than by what it was made of” (200).

Horgan elaborates:

“If a programmer created molecule-like structures that, following certain laws, spontaneously organised themselves into entities that could seemingly eat, reproduce, and evolve, Langton would consider those entities to be alive – 'even if they're in a computer'.” (200)

One can ask here why Horgan uses the words “seemingly eat” instead of the simple “eat”. If artificial beings eat, then they eat. That is, they gain some kind of energy or nutrition from what it is they eat – even if what they eat isn't organic.

In addition, why would artificial life automatically need to evolve? Since it would be artificial, there's no automatic reason that evolution should also apply to artificial life. Then again, there's no automatic reasons why such artificial entities shouldn't evolve either. It depends on the nature of the artificial beast.

Of course these artificial entities could do all the things mentioned above and still not be conscious or have minds. They could eat, reproduce and evolve and not have minds or display conscious states. Such things as eating, reproducing and evolving don't entail mind or consciousness.

However, it seems that such things do entail consciousness – or at least the possibility of pain. Langton says:

“I like to think that if I saw somebody sitting next to me at a computer terminal who is torturing these creatures.... I would try to get this guy some psychological help!”

I presume that if these 'creatures' can feel pain, then they must also display that pain. How would they do so? Again, artificial eating, artificial evolution and artificial reproduction don't entail consciousness or mind and therefore they don't entail pain. How would he (or we) know that his artificial creatures felt pain? (How would they know that even if they displayed 'pain behaviour'?)

Horgan goes into more detail as regards Langton's life-functionalism. He writes that he

“wanted people to realise that life might be a process that could be implemented by any number of arrangements of matter, including the ebb and flow of electrons in a computer” (200).

Horgan then quotes Langton:

“At some level the actual physical realization is irrelevant to the functional properties. Of course there are differences. There are going to be differences if there's a different material base. But are the differences fundamental to the property of being alive or not?”

It seems quite incredible that Langton should argue that the 'material base' isn't fundamental. Or at least he says that it may not be fundamental. Then again, it may well be fundamental. After all, it's a simple fact that all living things are organic, not artificial. The inductive evidence supports the position that physical constitution is important and fundamental. That just seems obvious.

Indeed isn't it the case that functionally speaking we've already replicated many of the things about life and mind that we wanted to replicate? So why haven't we actually got life or mind at this juncture? What's the missing ingredient? The functional or computational realities of computers and whatnot are already highly complex – so what's missing? Is the missing link biology - or the special qualities of the organic - after all?

Perhaps instead of replicating functions (such as computations, etc.), the scientists of artificial life and artificial mind should attempt to replicate biological matter (or the brain) instead. Though of course that would be fiendishly complex and it's not in sight at the moment. And that's partly why functions (rather than material bases) are emphasised so much in the AI and AL literature.

Searle on Mind, Computations & Computers

The Mind as a Computer: Syntax and Semantics

The first thing you can say (in accordance with John Searle) is that when a computer manipulates 0s and 1s, it doesn't know what they mean, symbolise, stand for, or what their referents are. Indeed the 0s and 1s don’t have any semantic features. They're purely syntactical. The only thing that matters to the computer is the shape of '0' and '1' – nothing more. That's why, as Searle says, that “any old symbol will do just as well”.

At its most basic, a computer simply scans a tape. Or, if not literally a tape (as in a Turing machine), then it scans something or other. This tape (or this something) will only contain 0s and 1s. What can the computer (or computer ‘head’) do to these 0s and 1s? It can perform four operations:

1. It can move the tape one square to the left.
2. It can move the tape one square to the right.
3. It can erase a 0 and print a 1.
4. It can erase a 1 and print a 0.

Here’s where the analogy with logic comes in. Instead of logic’s rules of inference, we have a set of rules of the form “under condition C perform act A”. Rules such as "under condition C perform act A" are called the computer programme. And the purpose of the programme is to encode information. This information is encoded in the binary code of zeroes and ones.

The computer translates the encoded information (which is in the form of 0s and 1s) into electrical impulses. It then processes these electrical impulses (which are now bits of information) according to the rules of the programme. We can say that the computer programme is a set of rules for processing information (or for processing electrical impulses).

In a sense, if the computations or symbols have no meaning (or they don't actually symbolise anything), then they aren't actually symbols at all. Of course they're symbols for us; though not for the computer itself. The only thing that matters for the computer are the formal and syntactical features of the symbols; whether these symbols are 0s, 1s or whatever.

According to Searle, the human mind doesn’t just manipulate symbols (whatever those symbols are taken to be). What more is there to minds? Well, “minds have contents”. What does content mean? It means that if we're thinking in English (or even manipulating English symbols such as ‘y’ and ‘s’, ‘cat’ and ‘tail’ or ‘The cat has a tail’), it's not just a question of the forms, shapes or syntax of these symbols: we also need to know what they actually mean. Thus in the sentence “The cat has a tail” the words ‘cat’ and ‘tail’ have references, and “has a tail” is predicated of the subject (which is a cat). And so on.

Not only that: some of the words have a sense. The whole sentence has a sense (or meaning) and a truth-value. We have a semantics which includes meaning, reference and predication; none of which matter to a computer because this is a question of content not syntax. That is, formal symbols alone doesn't guarantee or provide semantic content. And without semantic content we have no mind. Thus computers (or their programmes) aren't minds.

Searle sums up his argument thus:

1. Programs are entirely syntactical.
2. Minds have a semantics.
3. Syntax is not the same as, nor by itself sufficient for, semantics.

     4. Therefore programs are not minds. Q.E.D. 

It follows that for minds, semantics is important. Or, more commonly, for minds meaning is important. Because computers (or their programmes) don't have meanings (or know what their symbols mean), then they can't be minds.

Strong Artificial Intelligence

It's not thought that the physical aspects of a computer can bring about or cause mind or consciousness: the programme itself is a mind. So this isn’t the case of emergence from the programme’s implementation in hardware. Mind is the programme. Mind is the software.

So if software (or the program) is enough in itself, then of course the hardware won’t matter when it comes to a computer being a mind or it having mental states. Anything could implement the programme. It doesn't really matter what does so because the programme itself constitutes mind or mental states. In computers it just happens to be silicon chips and electrical circuits. In human beings it just so happens to be biological brains. Of course the programme will need some kind of hardware; though it doesn't really matter which kind of hardware. (In the case of the brain it's ‘wetware’.)

Despite all that, many things can be said to be computers. So to say that the mind is like a computer (or even is a computer) may not amount to much. Searle writes:

“For example, the window in front of me is a very simple computer. Window open = 1, window closed = 0. That is, if we accept Turing’s definition according to which anything to which you can assign a 0 and a 1 is a computer, then the window is a simple and trivial computer.” 

Is it really just a question of anything we can assign 1s and 0s to being a computer (or should I say, a digital computer)? In any case, why is it simply just a case of 0s and 1s, why not 3s and 4s as well? Why not the letter ‘S’ or the words ‘hat’ or ‘Jack’? Indeed why not the symbols ‘/’ and ‘*’ instead? From what Searle has said, these shapes or syntactic marks could work just as well. After all, it’s all about syntax and not about what ‘*’ means or what it symbolises or signifies.

Brain Processes and Computations

Searle has said that the brain is a machine. And if the brain is a machine, it must have machine processes. So what are the brain’s machine processes? One example would be a neuron firing; which is like "internal combustion". However, neuron firing, internal combustion and other machine processes aren't like computation. Why is that? Searle writes:

“… computation is an abstract mathematical process that exists only relative to conscious observers and interpreters. Observers such as ourselves have found ways to implement computation on silicon-based electrical machines, but that does not make computation into something electrical or chemical.” 

This means that neuron firing and internal combustion don't “exist only relative to conscious observers and interpreters”: computations do. Computations need to be observed and interpreted because they're abstract mathematical processes. We can make a distinction between computations (or abstract mathematical processes) and the physical things which implement such things. However, we can't make a distinction between neurons firing (or internal combustion) and the physical things that implement them. Neuron firings just are their physical implementations. They aren't abstract and they're not intrinsically mathematical or intrinsically anything other than physical and biochemical.

The Computer’s Simulation of Mind

If one were a behaviourist or a functionalist, then the behaviour of computers alone would tell us if they have minds. Though, according to Searle, this would be a simulation of minds. That's why we can simulate minds (or the workings of minds) more precisely in computers. But the simulation of mind is not mind. Searle writes:

“Computers are immensely useful devices for simulating brain processes. But the simulation of mental states is no more a mental state than the simulation of an explosion is itself an explosion.” 

That's why the zombie scenario is so popular in the philosophy of mind. In a sense, a zombie simulates a human person by behaving or acting like a human person. Though behaving or acting like a human person isn't the same as being a human person. Does the parrot which says "Hello John" act or behave like a human person simply because it simulates a greeting every time its owner arrives home from work? Does this verbal response make the parrot a person? Does it even have a mind simply because it can articulate the words "Hello John"? Does it understand these words? Does it know what they mean? Indeed does a computer know what the words "Hello John" mean? If a turd said "Hello John", would that turd have a mind? If, by accident, the pebbles on a sea shore spelled the words "Hello John Searle", would the sea shore or the beach have a mind?

Lamarck & Darwin Compared

 

The important distinction that must be made between Jean-Baptiste Lamarck's position and Charles Darwin's is that the former believed that animals acquired characteristics. In other words, organisms or animals can change while alive. Darwin, on the other hand, stressed the inheritance of characteristics, not their acquisition during the existence of animals.

Nonetheless, surely an animal has to acquire a characteristic before it can be passed onto - or be inherited by - future generations. Yes, that's true – though only over time. That is, individuals don't acquire characteristics over life-times. Though over time species may acquire characteristics. Those characteristics, though, will be too small to be noticed by one generation and will certainly not be noticeable over the lifetime of an individual animal.

Is this true of all species? What about the microscopic ones which have very short lifespans? Is is literally impossible for one such species to acquire a characteristic during its own lifetime?

The Lamarkian position is that “the constant craning of a giraffe to reach leaves high in a tree would alter its sperm or egg that its offspring would be born with longer necks” (114). This seems like a ridiculous idea – though only in retrospect! That is, only in the retrospect provided by knowledge of Darwin's theories. Nonetheless, the argument is still that repeated behaviours or habits of animals has a literal affect on sperms or eggs. Thus if the sperm or eggs are affected by this behaviour, then they will automatically produce offspring that will be different in some small or even large way.

Darwin's position, on the other hand, is that there is no direct relation between animal behaviour and changes in that animal's sperm or eggs. What actually is argued about behaviour X (say reaching the higher leaves) is that it's more likely to survive and thus pass on its genes because of behaviour X. The eggs or sperm aren't changed due to behaviour. Though the behaviour leads to a situation in which that animal, and animals like it, are more likely to survive. Thus giraffes with longer necks are more likely to survive. And, because of that, those giraffes which have longer necks are more likely to pass on the long-necked gene than those giraffes with shorter necks. Thus, over time, short-necked giraffes die out because less of them survive. And the less of them that survive (due to having short necks) means that they can't pass on their genes. Short-necked genes aren't passed on; though long-necked genes are.

Thus behaviour doesn't affect genes. What does affect genes, in fact, is entirely random. Though if a random change in the structure of genes produces giraffes with long necks, and long necks are more likely to secure survival, then the genes for long necks are more likely to be passed on simply because giraffes with longer necks are more likely to survive than giraffes with shorter necks.

Kurt Gödel's Theorems & Physics

It's often asked whether or not Kurt Gödel's theorems can be applied outside mathematics. John Horgan certainly applies them to the theories of physics. Or, more accurately, he writes that

 

“Kurt Gödel's incompleteness theorem denies us the possibility of constructing a complete, consistent mathematical description of reality” (6).

Clearly there's a jump here from Gödel's mathematical incompleteness theorems to physical reality. Or, more accurately, from Gödel's theorems to a “consistent mathematical description of reality”. Is that jump justified?

Well, for a start, physics is utterly dependent on mathematics. Thus if all descriptions of reality in physics involve mathematics, and mathematics is subject to Gödel's theorems, then that must pass over to the descriptions of reality which are offered by physicists. In other words, if a mathematical system must be incomplete (or not entirely provable), then that description of reality must be incomplete (or not entirely provable). The two must fall and rise together.

More meat is put on this idea of whether or not Gödel's theorems are applicable to theories about reality when John Horgan says that the “British physicist John Barrow argued that Gödel's incompleteness theorem undermines the very notion of a complete theory of nature” (69). We move again from mathematical systems to the incompleteness of a “complete theory of nature”. It can be said here that Barrow is simply transferring the incompleteness of mathematics to the incompleteness of a “complete theory of nature”. Again, does the former necessarily pass over to the latter?

In more detail: Godel established that "any moderately complex system of axioms inevitably raises questions that cannot be answered by the axioms". Then Horgan moves onto to say that the “implication is that any theory will always have loose ends”.

Many scientists accept this application of Godel's theorems to physics, including Moravec, Roger Penrose and Freeman Dyson. The latter says:

“Since we know the laws of physics are mathematical, and we know that mathematics is an inconsistent system, it's sort of plausible that physics will also be inconsistent.” (254)

Thus what we have here is a logical argument:

i) Physics is mathematical.

ii) Mathematics is an inconsistent system.

iii) Therefore physics must be an inconsistent system (or simply incomplete).

The only problem here is seeing the entirety of mathematics as a single system (which itself incorporates systems). Perhaps it is.