If understanding admits of degrees, then so must knowledge, since understanding is a condition of knowledge, and this bodes ill for things known, for facts, or propositions or whatever. A child with the rudiments of arithmetic under his belt knows what the number four is, or knows a little bit about the number four. The effect of this small knowledge is that the child can reel off reports of his knowledge ad nauseam: ‘four is half of eight …, four is 1/250,000 of a million’. None of these reports, and no finite collection of them, exhausts his knowledge of the number four; and since he is not an advanced student of mathematics we cannot expect him to offer, assent to or even understand statements about the real-number system, the infinite multiples of four – in short the sort of statements that might be held to generalize and exhaust his knowledge. What fact or facts can we say the child knows? Does he know an infinite number of facts or just one or two rather general facts? If the latter, he knows facts the ‘expression’ of which in his native tongue he probably does not even understand. And what of the child whose arithmetical knowledge is shaky, who knows that two times four is eight, three times four is twelve, but who is unsure about whether four times two is eight, four times three is twelve, or who denies the latter equations? If he does not have the symmetry of ‘times’ down pat, can we really say he knows that two times four is eight, and if this is not what he knows, what does he know?
Ryle has drawn his well-known distinction between knowing how and knowing that. When one claims to know how to swim, one supports the claim by swimming, not by an essay on swimming methods; knowing how is a matter of a talent or knack or ability, not a matter of having some propositions or facts in one’s head. The distinction is certainly illuminating at some levels of discussion, but does it break down when it comes time to determine just what is known when one knows that something is the case? The case of the child’s arithmetical knowledge is revealing because it seems to straddle the line between know-how and know-that. The child’s knowledge is very much like a knack or trick he has learned, and yet it is more; what the child does in demonstrating his knowledge is more than merely a rote parroting, the utterance of a string of phonemes. Yet what the child demonstrates is apparently more like a knack than knowing that yesterday was Friday. Where does one draw the line? Said of an adult that for any (true) p either he knows that p or he does not is initially plausible until we examine the penumbral cases.1 Do I know that table salt is sodium chloride? Yes, of course. But do I? What does this mean to me, or what can I do with this information beyond just reporting it? Educators are prone to distinguish the mere learning of facts from understanding or learning with comprehension, but do we want to say a person actually knows a fact if all he can do is utter some sentence in a few limited contexts (in response to an examination question, for example)? Here we seem to have just the ‘opposite’ of knowing that – a clear case of rote know-how. Surely what the trained chemist knows when he knows that table salt is sodium chloride is more than what I know. Imagine stationing a man who does not understand German on a street corner and training him to respond to ‘Wo ist der Bahnhof?’ with ‘Der Bahnhof ist links um die Ecke’. Is not my knowledge that salt is sodium chloride merely on a sliding scale from his ‘knowledge’ that der Bahnhof ist links um die Ecke? Understanding the statement that salt is sodium chloride involves more, of course, than just understanding the words, or ‘knowing their uses’; understanding the statement involves knowing about sodium and chlorine, but also about potassium and oxygen and valences and so forth. We cannot draw a limit so that understanding a statement involves understanding just so much.
If one can talk of a fact known, it must vary from speaker to speaker for any given sentence which might be held to express a fact. Alternatively, if one can settle on some way of anchoring facts to sentences, then these will not serve well as things known. An encyclopaedia – a very small one – might be held to store just one fact if it consisted of one non-compound printed sentence, but a person could not be held to know just one fact or understand just one statement. The knowledge of one fact could not exist by itself because the fact could not be used, and hence could not be understood. However facts may be anchored for encyclopaedias, the metaphor of the walking encyclopaedia is not to be trusted unless what we mean is just that the ‘walking encyclopaedia’ really knows nothing, is quite literally no more than a sort of recorder. What a person can use his stored information for depends on what other stored information he has, what else he knows. The things we do with our knowledge are quite discrete, but our knowledge itself does not divide into neat, independent parts, and hence cannot be ‘listed’.
XXV LANGUAGE AND INFORMATION
Philosophers of language in the past have attempted to pin down the information ‘contained’ in particular sentences by appealing to particular placements of things in the universe (the cat on the mat), particular time-slices of spatio-temporal reality, particular concatenations of qualities present to the senses, but these attempts to tie information to states of affairs of one sort or another fail because they do not take into consideration the intermediaries between sentences and states of affairs, namely the sentence utterers and hearers, the makers of verbal messages. A message picks out some feature of a state of affairs that is functionally important to some ‘receiving system’; something is a message or a signal only when it goes on to effect functions in some self-contained Intentional system. The freezing of a pond is not in itself a signal to the effect that the temperature of the water is below the freezing point.
In information theory it is often important to gauge the reliability of an information transmission channel or system, and for this a method has been developed for measuring amounts of information. Significantly, the amount of information in any signal is not directly a function of stimulus conditions or causes of the signal, or of any internal syntactic structure of the signal (which is usually, in any case, treated holistically as an ‘off’ or ‘on’ in a binary system). Rather the amount of information is determined by the degree of uncertainty diminished in the receiver. The receiver is given the task of singling out some individual or individuals from a limited ensemble or class of possibilities, e.g., finding out what day of the week it is. The signals received serve to exclude possibilities (e.g., the signal ‘It is not a weekday’), thus reducing the ensemble, or one signal can single out the individual; solving the problem in one step.
How much information is in the statement ‘This is Friday’? We now know that we must first determine the context. Suppose our ensemble was ‘The days falling between Thursday and Saturday’. Such an ensemble has one member, so that
I = log2 1 bits [the unit of information]
= 0 bits
The statement, then, contains no information.
In another context the result could be different. Suppose we know that since we are working it is neither Saturday nor Sunday. In this case, our ensemble has five equiprobable members, and
I = log2 5 bits
= 2.32 bits
Finally, let us suppose a man awakens from a coma. He has no idea how long he has been unconscious, and asks ‘What day is it?’ The seven possible outcomes are equiprobable, and
I = log2 7 bits
= 2.81 bits.2
A signal or message, then, like ‘This is Friday’, informs only relative to its function in ordering an ensemble, and the ensemble is determined by the receiver. To the man who knows that yesterday was Thursday and tomorrow is Saturday, ‘This is Friday’ is no news. This way of determining the amount of information works only for ensembles with a known number of equiprobable members. Thus it is of no use in determining the information content of ‘Your uncle just died’ or most of the sentence tokens occurring in everyday life. In the case of ‘Your uncle just died’ the ensemble might be held to consist of two members, uncle dead or alive, and if a person were waiting for news on the state of his uncle, then to say that the sentence carries log22 or 1 bit of information would make some meagre sense.
But in human beings, as opposed to devices with one limited job to do, the receipt of information allows a great many different ensembles to be partially ordered, depending on the knowledge already held by the receiver. Thus one of our intuitions about information in people is provided with a quasi-mathematical model: the information received by people when they are spoken to depends on what they already know and is not amenable to precise quantification.
An intuitive account of what happens when I tell someone something is that I try to transmit or impart or share with my listener something somehow held in me (as known or believed); I try to produce in another person something (knowledge of something) that I have. What is perfectly clear is that this something I am trying to transmit is information, and the something I am trying to produce in the other person is the storage of information, of information having the same or similar content to information stored in me. However, as soon as one supposes that sentences uttered are straightforward vehicles of particular and determinate morsels of information, that out of the building blocks of language we can construct vehicles of just so much information on a particular topic, paradoxes arise. If I do not know that Tully is Cicero and announce ‘Cicero denounced Catiline’, my listener, if he knows Tully is Cicero, will in effect come to know more than I was endeavouring to tell him. He will receive and store not only the information that Cicero denounced Catiline, but also the information that Tully denounced Catiline. Should one say that he received more information than I sent? Such dividends of information do not always hinge on synonymy or identity of reference of terms as in the Tully-Cicero case. On hearing ‘Your uncle just died’ Jones may be informed that he will soon be a rich man, that a certain Mrs. Smith is now a widow, etc., and however information is to be construed these dividends are not the same information the speaker intended to impart.
If, in telling someone what I know, the effect that signals success is the production of information storage in him similar or the same in content to information stored in me, the chances of success seem remote. If, as I argued earlier, content is in part a function of capacity to direct further efferent activity, it would be very rare for the listener to acquire the same content as that stored in me, since there will always be differences in these capacities except in the unlikely case where the listener has an information store that already duplicates mine in every relevant respect save just what I am communicating to him. This is brought out in a definition of meaning provided by MacKay. He defines ‘the meaning of an utterance (intended, standard, received) as its selective function (intended, standard, actual) on the range of possible states of the appropriate system’.3 As MacKay points out, this makes meaning relative, not an absolute property. If intended meaning is to be at all an approximation of standard meaning or actual meaning, there must be some similarity in selective function of the utterance from person to person, and this will only be the case if there is considerable similarity from person to person of the information storage systems. This requirement is easily overlooked, but it is evident in some of our everyday observations about our successes and failures at communication. Both speaker and hearer must share relevant knowledge for communication to occur. There is more to this than the fact that if I attempt to communicate in English with a person who speaks no English, I will not succeed in producing in him information similar to mine at all. Even if the hearer is English, he must also have much the same background of information on the subject of discussion as I have. The sentence ‘I’ve found a solution to the problem of other minds’, which contains no words that the average adult English speaker would not know, is still unlikely to be informative to a person who does not share with the speaker a background of knowledge of this traditional philosophical problem, the speaker’s activities, and what might be considered to be a solution to the problem. As MacKay says, ‘topic understanding multiplies the informational impact of a proposition’.4 The similarity of background information stored need not be complete, and where it is partial, communication is partial, as we observe when we say ‘I understand you, but what you say doesn’t tell me very much’.
The fact that the information-bearing capacity of language is thus finally dependent on the effects of language on a person, on what I have called an Intentional system, is the same fact on the personal level as the fact on the sub-personal level that the centralist is unable to ascribe precise contents to the events and states of such a system.
XXVI CONCLUSIONS
The problem of mind is not to be divorced from the problem of a person. Looking at the ‘phenomena of mind’ can only be looking at what a person does, feels, thinks, experiences; minds cannot be examined as separable entities without leading inevitably to Cartesian spirits, and an examination of bodies and their workings will never bring us to the subject matter of mind at all. The first step in finding solutions to the problems of mind is to set aside ontological predilections and consider instead the relation between the mode of discourse in which we speak of persons and the mode of discourse in which we speak of bodies and other physical objects. This studious avoidance of ontological commitments allows us to relax the requirements of a rapprochement between the language of mind and the language of science, and, as we have seen, none of the freedom provided us by this stance is gratuitous. Thoughts, for example, are not only not to be identified with physical processes in the brain, but also not to be identified with logical or functional states or events in an Intentional system (physically realized in the nervous system of a body). The story we tell when we tell the ordinary story of a person’s mental activities cannot be mapped with precision on to the extensional story of events in the person’s body, nor has the ordinary story any real precision of its own. It has no precision, for when we say a person knows or believes this or that, for example, we ascribe to him no determinable, circumscribed, invariant, generalizable states, capacities or dispositions. The personal story, moreover, has a relatively vulnerable and impermanent place in our conceptual scheme, and could in principle be rendered ‘obsolete’ if some day we ceased to treat anything (any mobile body or system or device) as an Intentional system – by reasoning with it, communicating with it, etc. That day is not to be expected – and certainly not hoped for – in spite of the inroads that are now being made in ‘impersonal’ ways of controlling people.
The feature that is central (if not quite universal) in the personal mode of discourse is Intentionality, and it is this feature that persistently tempts the theory-builder into positing mananalogues as elements in his analysis, thus obviating the analysis entirely. In his purest form the little man in the brain takes on the guise of brain-writing reader, an intelligent, communicating system capable of understanding messages. Positing the brain-writing reader is almost irresistible, for if we cannot understand central states and events of the nervous system as bearing content, as being messages of some sort, it is not clear how we can understand them at all. The temptation must be resisted, however, by recognizing the disanalogies between verbal communication and non-verbal intra-cerebral communication and indeed the primacy of non-verbal communication. Other roles played by the little man in the brain are merely specialized roles projected inwards from the details of our initial analysanda, the variety of affairs of a person. The solitary audience in the theatre of consciousness, the internal decision-maker and source of volitions or directives, the reasoner, if taken as parts of a person, serve only to postpone analysis. The banishment of these concepts from our analysis forces the banishment as well of a variety of other self-defeating props, such as the brain-writing to be read, the mental images to be seen, the volitions to be ordered, and the facts to be known. These props are self-defeating because they could only serve the functions for which they were designed in conjunction with interior person-analogues, and hence as elements in an analysis they reproduce the problems like images in a hall of mirrors.
NOTES
1
THE ONTOLOGICAL PROBLEM OF MIND
1 In the imm
ense literature of identity theory, several items stand out as germinal: U. T. Place, ‘Is Consciousness a Brain Process?’, British Journal of Psychology, XLVII, 1956, pp. 44–50. H. Feigl, ‘The “Mental” and the “Physical” ’, Minnesota Studies in the Philosophy of Science, Vol. 3, eds. H. Feigl et al., Minneapolis, 1958, pp. 370–457. J. J. C. Smart, ‘Sensations and Brain Processes’, Philosophical Review, LXVIII, 1959, pp. 141–56. In the spate of papers following these the greatest advance of outlook is to be found in T. Nagel, ‘Physicalism’, Philosophical Review, LXXIV, 1965, pp. 339–56. Place’s and Smart’s papers are reprinted with revisions in V. C. Chappell, ed., The Philosophy of Mind, Englewood Cliffs, 1962.
2 G. Ryle, The Concept of Mind, London, 1949.
3 I must postpone briefly an important and difficult question: are sentences containing these contextual disharmonies such as ‘I can sit on an opportunity’ syntactically ill-formed (and hence neither true nor false) or are they false by meaning (and hence have true negations)? For an excellent discussion of this and other questions raised in this section, see F. Sommers, ‘Types and Ontology’, Philosophical Review, LXXII, 1963, pp. 327–63, reprinted in P. F. Strawson, ed., Philosophical Logic, Oxford, 1967.
4 W. V. O. Quine, Word and Object, Cambridge, Mass., 1960, p. 244.
5 Ibid., p. 244.
6 Theorists of space and measurement have not always been alive to this category mistake. Descartes, for one, advanced the argument against the existence of vacuums that if there was no matter between A and B there could be no distance between them (Principles of Philosophy, Part II, secs. 16–18).
Content and Consciousness Page 23
