Today, in the early 21st century, many people are already familiar with the idea that we have no proof that the world external to our mind actually exists. There is no way of proving that what I think of as “myself” is not actually just a disembodied brain in a vat full of chemicals with signals fed in by some aliens’ computer making me think that I have a body that is taking part in an inert material world. Alternatively, perhaps I am just a piece of software running in a giant supercomputer executing a simulation game. We can’t prove that such an outlandish alternatives to common sense reality are false.
However, that should not mean that there is nothing we can be certain about; phenomenological knowledge should be safe. When Descartes in his Meditations (1641) famously provided that “I think therefore I am”, this can be understood as asserting that we may not know that the tree before me exists or not, but I do know that I am undergoing an experience of seeing the tree or believing that I see the tree or being presented to in a tree-like manner. Phenomenological knowledge is awareness of the sense-data without actually any commitment to any external reality as the source of that data and certainly not to the nature of that reality. So I know that “I am seeing a red ball” even if I don’t know that it is a red ball that I am seeing. The outlandish alternatives that opened this section do not undermine the sense-data statements; they are compatible with these statements.
However, this solution too, has proven not to stand the test of time. There are numerous problems with a phenomenological foundation. Sellars (1956) is one of the highlights of the arguments that undermined the sense-data foundation. Even if some facts are “given”, such as the experience of red, these facts are sub-lingual and useless as a basis for logical inference. Other philosophers (for example, Putnam 1990) focus on the fact that any sentence that you assert about the experience will include words that you would retract if an “expert” corrected you. In fact, language as a whole is a culture-created entity that plays a very large part in the creation of the sentence that you finally assert about what you see. So, one can neither assert that it is true that there is a red ball before me, nor can a person assert that it is true that “I am having an experience such that it seems to me that I am seeing a red ball”.
All we are left with is (what seems to be) the psychological fact that a sense-data assertion such as “I am seeing a red ball” is an assertion that one typically would not want to retract. However, there is little point in putting too much importance in the psychological fact itself. Rather, it would be best to present the whole thing as a game about sentences in the English language. The rules of the game are such that sentences such as “I am seeing a red ball” or “the dial I see is pointing roughly to 3” should only be changed if one really has no choice. On the other hand, there are other sentences in the game, that are prime candidates for change.
Quine (1951) presents the tantalizing image of a web of belief. Rather than get involved in the complexity of the “belief” concept, the concept used here will be a web of textual assertions. The sum of the knowledge of a system is represented only in terms of sentences and some basic rules of inference that can act on the sentences. The knowledge system in question, could be, for example, the totality of an individual’s knowledge. Of course, an individual also has commitments that may be sub-verbal which cannot be expressed as sentences. However, while not denying the existence of these non-verbal assertions, they are not included in the system for now. At most, they may be seen as the cause of the addition of specific assertions into the web of sentences.
Returning to the game, there is a player of the game and there is a database or web of sentences or wfs’s (see the last post on pre-logic). The rules of the game mark some sentences as sense-data sentences. These will having a high change cost. However, the player may add other sentences at will. These will have initially a very low cost to add or change or delete. The new sentences may be simple fact-like assertions or they may be procedures, like inference rules, for changing other sentences in the database. As before, it is important not to confuse the intrinsic inference rules of the system which by the rules of the game may not be changed and these non-intrinsic rules which are added by the player. The sense data sentences change over time by some other, seemingly external, factors. The aim of the game is for the player to create calculations using the low change-cost sentences such that new sentences are created that are identical to the sense-data sentences.
Once a high degree of success is found in predicting the sense-data sentences, it will be found that changing some sentences has an effect on a large number of other sentences of the web. Therefore, if such a highly connected sentence is changed many other sentences will have to be adjusted if the sense data sentences are going to continue to be predicted successfully. In the light of this, a change that requires so much work is also to be considered high cost. Therefore sense-data sentences are not the only ones that have high change-cost.
The game is entirely syntactic. If strings such as “true” or “false” are useful in creating successful predictive calculations, then they will be used but they will have no special significance. One may speak of strings having meaning but this will be no more than an indication of the fact that these words participate in the reference/retrieval mechanism of the intrinsic inference rule of the system. (The intrinsic inference rule is the rule that defines how to play the game itself. It was explained in a different context in the post on Pre-Logic.)
Since there is no semantics involved, there is no differentiation between valid rules of (non-intrinsic) inference and invalid rules. Of course, it could be that inventing one rule (call it MP, for Modus Ponens) will eventually result in a high degree of sense-data predictability whereas some other rule (call it AC, for Affirming the Consequent) will be found to be entirely useless. In that case MP will play a central role in almost all prediction calculations and therefore any attempt to replace it will require massive changes throughout the web; changes that may or may not succeed in achieving the same levels of prediction. If that is so, MP will have a very high change cost, higher even than sense-data sentences. MP may even be useful for calculating results in thought experiments of what might be possible as opposed to what actually is observed. People might even speak of MP as being “necessarily true”. It might be. However, MP’s utility is a sufficient explanation of its high change-cost. Therefore any attempt to assign any greater significance to MP cannot benefit from data actual or even (imagined) possible.
What is required in order to succeed at predicting sense-data is the creation of complexity. The intrinsic inference rule must allow for creation of calculations, processes or laws that produce complex, intricate and repeatable results. Just as Conway’s Game of Life can create extremely complex patterns developing over time with just a handful of very simple rules, so Wittgen can (as well as many other systems). Once there is a way of generating complexity, it should not be surprising that it is possible to create procedures that track the transformations that sense-data sentences make over time.
If one views the human mind past and present, individual and collective, and common-sense as well as advanced science, as participating in a game as just described, an interesting result emerges. The rules of the game did not allow the player of the game to change in the sense-data sentences (except under extreme circumstances). Nevertheless, they are not only seen to change but, it turns out relatively simple and unified systems of calculation can predict and track those changes1. These sentences are the given2. Just as there are processes that track these givens, there are processes that fail to track these givens. Thus, even if there is no truth or semantics to the system, there is no sense of “anything goes”. Only fanatic attention to the details of the given and herculean efforts at finding successful predictive procedures will result in a moderate degree of success at tracking the given. If there is a “reality” then it is the fact that the given, the sentences not under the control of the game, do change and they change in regular, predictable ways. Perhaps there is another meta-“reality” behind this “reality”, but nothing said so far seems to require postulating this meta.
Sellars may be right about the fact that there is no “truth” to the given or that the sub-verbal truth cannot participate in inference. However, it does not seem to matter. Sense-data need not be true. There is no need for truth, validity in inference or semantics in order to explain the success of common sense or advanced science. It is sufficient that we play a game that commits to make the goal that of predicting some sentences, taken as given, rather than ignoring or modifying these sentences3.
References:
Sellars, W., 1956, Empiricism and the Philosophy of Mind, Herbert Feigl and Michael Scriven, eds., Minnesota Studies in the Philosophy of Science, Volume I: The Foundations of Science and the Concepts of Psychology and Psychoanalysis. University of Minnesota Press, pp. 253-329
Quine, W. V., 1951, ‘Two Dogmas of Empiricism’, Philosophical Review, 60:20-43
1 There is little point in arguing that the player may be generating the sense-data sentences unconsciously. The question of what or who is generating the sense-data sentences does not concern the discussion. The only issue is the law-like behavior of the permutations of the sense-data sentences. To see this, imagine that your individual subconscious or perhaps the collective subconscious is some super-mathematician capable of calculating all the values experimental data should have and putting it in our way for discovery. Then it would turn out that this mathematician is the matrix-master supercomputer. There is no reason to call it a sub-faculty of the player.
2 Note, the use of the word given here is different from that of Sellars’. Sellars uses the word in an absolute, externally defined sense and argues that it cannot be applied to sentences. Here the word is used only as a label for the subset of sentences, which, in accordance with the rules of the game, have an extremely high change cost by virtue of being understood as sense-data descriptions of internal experience.
3 There are both similarities and differences between the ideas presented here and Kant’s phenomenal-nomological divide, Putnam’s Internal Realism, Rorty’s views, coherence theories of truth and many more. Mapping the location of these various views is beyond the intention of this essay, which is to present a philosophical underpinning for Wittgen.
No comments:
Post a Comment