Tuesday, February 16, 2021

Six rounds with Wittgenstein (part 2)

Arthur Dove, Nature Symbolized (1911)

Picking up from where we left off...

 101. We want to say that there can't be any vagueness in logic. The idea now absorbs us, that the ideal 'must' be found in reality. Meanwhile we do not as yet see how it occurs there, nor do we understand the nature of this "must". We think it must be in reality; for we think we already see it there.

102. The strict and clear rules of the logical structure of propositions appear to us as something in the background——hidden in the medium of the understanding. I already see them (even though through a medium): for I understand the propositional sign, I use it to say something. 

105. When we believe that that we must find that order, must find the ideal, in our actual language, we become dissatisfied with what are ordinarily called "propositions", "words", "signs".

The proposition and the word that logic deals with are supposed to be something pure and clear-cut. And we rack our brains over the nature of the real sign.——Is it perhaps the idea of the sign? or the idea of the present moment?

All right. I'm going to level with you here.

I wrote four or five paragraphs about the reification of concepts, and then stopped because I wasn't sure how any of it answered what Wittgenstein seemed to be saying.

Then I started over. I composed seven paragraphs (and transcribed a two-paragraph block quote from Hayes, Barnes-Holmes, and Roche's Relational Frame Theory: A Post-Skinnerian Account of Human Language and Cognition (2001) and then scratched it again. Not only was I unsure how any of it addressed Wittgenstein's points, I realized I have no idea what Wittgenstein is trying to get across—and I'm increasingly confident that he wasn't really sure, either. 

You're undoubtedly familiar with the strain of obscurantist academic writing where an author lays out a dense fog of specialized or very idiosyncratic jargon in order to inflate a weak idea or to conceal from the reader that they're not being guided to any kind of conclusion. What Wittgenstein does here is take the minimalist version of that approach: making pithy, evocative statements that actually say much less than what the reader may be led to interpret of them.

159. But when we think the matter over we are tempted to say:  the one real criterion for anybody's reading is in the conscious act of reading, the act of reading the sounds off from the letters. "A man surely knows whether he is reading or only pretending to read!"——Suppose A wants to make B believe he can read Cyrillic script. He learns a Russian sentence by heart and says it while looking at the printed words as if he were reading them. Here we shall certainly say that A knows he is not reading, and has a sense of just this while pretending to read... 

160. ...Suppose that a man who is under the influence of a certain drug is presented with a series of characters (which need not belong to any existing alphabet). He utters words corresponding to the number of characters, as if they were letters, and does so with all the outward signs, and with the sensations, of reading....here we should certainly be inclined to say he was making up an alphabet for himself ad hoc and then reading accordingly.

Except as probes or prompts for discussion, both remarks are useless. There are very few occasions where the experience of reading matters more than the controlling relation set up between the person looking at the page and the visual stimuli presented to him.

159 is additionally useless because it treats an isolated episode. Suppose B hands A a sheet of Cyrillic text that A has never seen before. If B (or any other party) is interested in whether A is actually reading or not, they're not going to worry about what A's "sense" of his action might be. They're going to probe the efficacy of Cyrillic script as a controlling stimulus.

160 is useless for the same reasons. Let's say that "under the influence of a certain drug," a person looks at a series of random characters and "believes" he is reading them. By placing the act of reading in the mind, or characterizing it as a sensation instead of as a type of controlled response, Wittgenstein's argument isn't even wrong (to borrow a phrase from Dr. Woit).

Imagine we show the subject (still under the influence of A Certain Drug) a sheet of paper with the following characters:


And he reads it off to us. Whether or not he's reading in a real or an ad hoc language is irrelevant. For argument's sake, let's say he responds by "reading" in English, and while viewing the "phrase" above, he says:

Uma likes cod  

Well, maybe he got lucky. He produced from this jumble of nonrepeating characters an English sentence in which no letter occurs twice. Maybe he's able to point to each symbol, one by one, and tell us what phoneme it represents. But the determination of whether an actual controlling relationship exists between these random characters and our subject's verbal behavior cannot consist of a single observation of a single trial.

We'd show him more sheets of paper with assortments of non-alphabetic characters printed on them. If we're decent scientists, we'll be paying close attention to which phonemes he attributes to each symbol. We'll ask him to read each "sentence" repeatedly; we'll point to each symbol and ask him to "sound it out;" we'll make him revisit pages he's already "read" to us. If we find a consistent one-to-one correspondence between symbols and sounds between tests, then the "sensation" of reading is irrelevant: what he's doing is reading.

But the possibility (indeed, the entire scenario) is so ludicrous that we shouldn't even be indulging in hypotheticals.

If we're agreeing with Wittgenstein that should "be inclined to say he was making up an alphabet for himself ad hoc and then reading accordingly," we must also allow that somebody knowing absolutely nothing about webcode can open Notepad after taking a few gravity bong rips, start punching in something that seems to him like a coding language, and then announce that he's made a website—and we could not tell him he's being ridiculous.¹ The test, of course, will be if what he's composed becomes a web page when accessed through a browser—just as the test for basic reading comprehension is if specific letters or characters evoke a specific conditioned response.²

161. ...Try this experiment: say the numbers from 1 to 12. Now look at the dial of your watch and read them.——What was it that you called "reading" in the latter case? That is to say: what did you do, to make into reading?

Once again: the answer is in the controlling relation between stimulus and response.

Saying the numbers one through twelve is an example of intraverbal behavior. The self-generated stimulus of saying "one" strengthens the response of saying "two," just as hearing "I pledge allegiance" strengthens the response "to the flag."³

(An explanation as to why or how we should respond to a familiar [verbal] audio pattern by speaking or "hearing" the next sounds in the sequence is a matter of physiology and evolutionary biology.)

Intraverbal behavior is typically initiated by something in the environment—say, a printed message asking you to recite the numbers one through twelve. Saying "one" begins the chain, and intraverbal contingencies maintain it until a stopping point is reached. (In this case, the verbal instruction to stop at "twelve." What if no stopping point is specified? Well: start counting out loud. Pay attention to the events that precede your giving up and doing something else.) 

In reading, the controlling stimulus is visual. Your digital watch reads 11:36. (Wittgenstein would have been locking at a clockface, but the difference is insignificant here.) "One," you say. If you held your finger over the other three numbers and immediately withheld the watch from view, you might be inclined to say "two"—owing to the strength of "one" as a typical intraverbal stimulus. But if you're reading the watch, the relevant variables are the digits on its face.

The concomitant sensations are, once again, incidental.

184. I want to remember a tune and it escapes me; suddenly I say "Now I know it" and I sing it. What was it like to suddenly know it? Surely it can't have occurred to me in its entirety in that moment!——Perhaps you will say: "It's a particular feeling, as if it were there"——but is it there? Suppose I now begin to sing it and get stuck?——But may I not have been certain at that moment that I knew it? So in some sense or other it was there after all!——But in what sense? You would say that the tune was there if, say, someone sang it through, or heard it mentally from beginning to end. I am not, of course, denying that the statement that the tune is there can also be given a quite different meaning——for example, that I have a bit of paper on which it is written.——And what does his being 'certain', his knowing it, consist in?——Of course we can say: if someone says with conviction that now he knows the tune, then it is (somehow) present to his mind in its entirety at that moment——and this is a definition of the expression "the tune is present to his mind in its entirety".

Again: Wittgenstein asks "what was it like to" when he should be asking "what was happening when." 

But he is correct in surmising that a tune cannot be present to anyone's mind in its entirety; the temporal dimensions of sound patterns the possibility. A person might be capable of singing a tune "in his head" from beginning to end, but potentiality is not what is meant by "present to his mind in its entirety."  

For simplicity's sake, let's say we're talking about a person who wants to sing "Yesterday" by the Beatles, but reports having trouble "recalling" the lyrics. We put "recall" in quotes because the radical behaviorist considers it a somewhat problematic term, tainted by the implicit metaphor of the storage room in mentalistic descriptions—the notion of an inner agent, a small person working in a depot within the human brain, taking memos instructing him to retrieve pieces of information from the rows of shelves and bring them to the front desk. The behaviorist rejects the concept of "memory storage" as a facile, fictive description of human activity: a song our subject can sing from the first note to the last does not exist in her.

To "know" a song, in the context of the person who's trying to sing "Yesterday," means that the subject is capable of singing it from beginning to end. The difference between "not knowing" and "unable to remember" might be trivial if we're dealing with a subject whom we know nothing about (has she heard "Yesterday" before? has she sang it on a prior occasion?), and if we only intend to observe her during the time when she's apparently struggling to to sing the next verse. But a behavior analysis undertaken with no information about the subject's history, and that restricts itself to a single observation, would be worthless.

To the behaviorist, the subject's "not knowing" the song means she cannot be made to sing it; that there exists a zero percent (or otherwise infinitesimal) probability that an event might occur that elicits from her the verbal operant sequence we specify as "Yesterday," sung with pitch and rhythm which closely resemble Paul McCartney's performance on the 1965 "Yesterday" single.

As Hume might point out, there is no way of knowing this for certain unless we perform trials unto perpetuity. But for all practical purposes, settling for a pragmatic confidence in light of the unattainability of absolute certainty has adequately served the physical and natural sciences in the centuries since Newton.⁴

If our subject has only "forgotten" the song, we understand that she has emitted the above operant in the past. Like any other form of operant behavior, singing a particular song is dependent on environmental variables. Our subject "remembers" the song when a state of affairs arises such that the behavior of singing "Yesterday" occurs.

There is no circularity in this description, provided we take as given that the subject has been capable of singing "Yesterday" in the past. She has at least once performed an elaborate verbal operant; she "forgets" and cannot perform it. She "remembers"—now she performs it again.

What we're interested in is what happened to facilitate this. A complete account is probably impossible; it would entail observation of every occasion in which she responded to "Yesterday" being played, as well as the times she hummed the melody, sang one or two verses, and possibly even those times when she "heard" the song during instances of involuntary musical imagery. It may seem like a devilish lot of trouble to go through just to determine the events that set up the variables on which the possibility of a person singing a pop song depend—but if we're truly concerned with what happens when somebody "recalls" a song they'd been unable to sing a moment earlier, we gain more useful insights though a methodologically sound construction of events than a facile, reductionist description of "suddenly remembering," restricted to a particular instant in the subject's life.⁵ 

We might also note that any observation of a tune "escaping" somebody presupposes an attempt to sing it—a halted execution of behavior that has been previously carried out at least once before.

Contra Wittgenstein, recalling a song does not occur spontaneously: it is brought about by a change in the controlling variables. Some of these changes are probably owed to the behavior precipitated by the frustrated attempt: say, humming the melody or singing a known verse to strengthen an intraverbal response. (Such actions may be performed subvocally.) Maybe there was a cue in the environment: the aggravated singer hears somebody speaking a phrase, or encounters a particular object or event that strengthens a verbal response that functions as a formal of thematic prompt (a variable supplementing existing sources of strength) that initiates an intraverbal chain that concludes with singing the song in its entirety.⁶ (Some of these explanations are obscure, true: but if we wish for "mental" events to coexist in the universe with physical events, they must adhere to the same physical laws of causality, even if our acknowledgement of the fact makes it difficult or impossible to precisely determine the controlling variables during a given occasion.)

A report of "what is was like to suddenly know it" tells us nothing about Wittgenstein's success in remembering his tune unless he gives us some indication of what changed to permit his singing of it. Moreover, he apparently takes for granted that successfully singing the tune must follow the statement "now I know it:" it is entirely possible that a subsequent attempt to sing the tune will fail despite the strengthening effect of a successful probe. 

198. "But how can a rule shew me what I have to do at this point? Whatever I do is, on some interpretation, in accord with the rule."——That is not what we ought to say, but rather: any interpretation still hangs in the air along with what it interprets, and cannot give it any support. Interpretations by themselves do not determine meaning. 
"Then can whatever I do be brought into accord with the rule?" Let me ask this: what has the expression of a rule——say a sign-post——got to do with my actions?" What sort of connexion is there here? I have been trained to react to his sign in a particular way, and now I do so react to it. 
But that is only to give a causal connexion; to tell how it has come about that we now go by the sign-post; not what this going-by-the-sign really consists in. On the contrary; I have further indicated that a person goes by a sign-post insofar as there exists a regular use of sign-posts, a custom.

I read this six times. I began responding to it twice. Once again, I'm going to have to abandon the effort because I don't know what the hell Wittgenstein is getting at with "what going-by-the-sign really consists in." Everything else is Behaviorism 101.

For the sake of saying something about this, the question with which this item begins might be answered in a couple of different

I did it again. I typed three paragraphs in response to the first question (which were just a basic paraphrasing of how conditioning works), and then another four expatiating on the question "then can whatever I do be brought into accord with the rule?" and had to stop because, at best, the question is so nebulous as to merit any number of answers to any number of sorts of rule-governed situations, and at worst it's just nonsensical.

The short answer is: depends on the nature of the rule, the agencies enforcing it, and the frequency and severity of consequences for violating it.

If the rule is "don't spit on the sidewalk," and the question is "then can whatever I do be brought into accord with the rule?" the answer is: you can do whatever you want with regard to the rule, provided you don't spit on the sidewalk. But if Wittgenstein's idiot is asking "can I spit on the sidewalk anyway?" the answer might be "go ahead" if the community has set up faulty contingences of reinforcement (i.e., seldom punishing people who spit on the sidewalk).

If the rule is, say, arithmetical, it depends on who's grading your worksheet or auditing your tax forms. If you're an architect who wants to play fast and loose with arithmetical rules, you probably won't be an architect for very long.

To go back to the question "how can a rule show me what to do at this point?" consider this example. 

You're given a make-work job sitting at a conveyor belt with a ball-peen hammer in your hand. A procession of colored beads travels past you when the belt activates. Your job is to smash the black beads and let the white beads roll on untouched. Wittgenstein's question would be relevant when a purple bead passes you.

If you smash it, it may be because of stimulus induction: the stimulus of the purple bead between two white beads is similar enough to the stimulus of a black bead between two white beads to elicit the response. In a case where the purple bead appears between two black beads, whether you smash it or stay your hand depends on the extent to which the relata have more or less control over your response than absolute characteristics of the stimuli: in other words, whether the conditioned stimulus property is (1) the color black (or a particular shade of it) in and of itself, (2) its distinction from the color white, or (3) both. 

Let's assume the only verbal instruction you are ever given is "smash the black beads." Unless your training also consisted of a differential reinforcement schedule—a session or two where, after smashing black beads and sparing white beads for fifteen minutes, you were explicitly reinforced for sparing dark gray or midnight blue beads when they were presented, or corrected for smashing them—the extraneous properties of "not white," "darker than white," "almost as dark as black," (however difficult they might be to specify without actually taking optical wavelengths into account) may have acquired some control over your responses in this context. (Think again of the pigeon tapping the orange button.)

If you do not smash the purple bead, only further observation can determine whether you refrained because you "understood the rule" and held your hammer in place, or because of what Skinner called "algebraic summation:" two competing responses to an event ("smash the bead" and "hold the hammer in place") resulted in a weakened response that topographically resembled "hold the hammer in place," but perhaps with different contingencies than in your typical response to the white bead. (One probable effect might be the emotional response called "anxiety.").

Absent an intervention where the rules are clarified one way or another, your self-generated response to sparing the first purple bead will become a reinforcer which contributes to the determination of your response when the next purple bead passes you.

This has all been to say that when, in the context of behavior carried out according to a rule specified and enforced by some mediating social agency, a stimulus occurs for which no response has been specified or trained in advance, other variables in the actor's history determine the prevailing response. Maybe Wittgenstein was hinting at this? Who knows?

1A. What would seem to the novice like a coding language? A more relevant question would be: if he's never examined or typed out code before, what are the variables that control his output as he works?

1B. We might also ask him to describe in advance what the page will look like.

2. For this case, it doesn't make a difference if the subject's "audience" is a web browser. We're still testing fluency in a written language. 

3. Presuming, of course, that you attended primary school in the United States. Chances are, the response still won't be strong enough to pass beyond subvocal emission.

4. The social sciences are another matter entirely.

5. "Momentary behavior," writes William Baum, "is an oxymoron." The moments comprising an organism's lifespan, however we measure them, exist in continuity.

6. More jargon. Skinner on prompts:

Why the behavior is not strong enough to be emitted without supplementation does not matter. A response may be simply poorly conditioned, or controlled by stimuli which are currently weak, or related to states of deprivation or aversive stimulation which are moderate or weak, or displaced by other behavior as a result of earlier punishment, or confused by other current variables. Sometimes the problem is to make previously subvocal behavior vocal, but usually it is to evoke behavior which will not otherwise be emitted...

A thematic prompt is a supplemental source of strength in the form of a tact or intraverbal response. It is better known as a "hint."* Thus we may stimulate our hostess to ask More tea? either by inspecting our cup, or conspicuously draining the last drop, or by supplying an intraverbal stimulus containing such forms as drink, beverage, coffee, and so on. It is assumed the response More tea? exists in some strength; if the supplementary stimulus is so strong as to generate the required behavior entirely on its own, the hint is too broad to be called a prompt.

...Accidental thematic prompts [the kind we are considering] also occur, as when we are "reminded of a topic about which we had intended to speak." We may translate an elliptical expression of this sort by saying that "behavior which existed in some strength receives an accidental supplement from related thematic materials."**

* Skinner writes about verbal tests in the laboratory; in the case of a prompting a song in everyday life, "hints" will occur in the environment rather than be provided by an operator.

** We might characterize a "thematic" stimulus as one which increases the probability of more than one verbal response. If my friend Madalyn is talking about Florida, several different verbal responses in me are strengthened: I might say something about lizards, about the manatee that was ill-used by a Trump supporter, about tropical weather, or so on. The other variables that determine which of these "competing" responses prevails are present in the environment during our conversation (and may be exceedingly subtle), and their effectiveness in producing the eventual response is a function of my history of reinforcement. (My behavior isn't random or self-determined; it is complex.)

1 comment:

  1. I feel like I want to defend Wittgenstein here on one point in particular. You accuse him of repeatedly looking at the experience of what happened rather than just looking at what happened. You also say that he seems to assume that when a person says they know how to go on and attempt to sing a tune they will always succeed. This seems extremely out of step with Wittgenstein as I understand him, and it reminded me of a passage (really quite a few passages, but I'll only quote the one) from the second half of Philosophical Investigations:

    'The words "It's on the tip of my tongue" are no more the expression of an experience than "Now I know how to go on!"—We use them in certain situations, and they are surrounded by behaviour of a special kind, and also by some characteristic experiences. In particular they are frequently followed by finding the word. (Ask yourself: "What would it be like if human beings never found the word that was on the tip of their tongue?")'

    It seems to me that in this passage in particular, but really at every step, Wittgenstein wants to insist on exactly what you want to insist on: that the important thing is not the experiences (certainly not the thoughts) which may or may not accompany the words, but ALL the myriad things that do accompany the words or have accompanied them in the past, as well as the context or situation in which this is all embedded in. (I think to this he would add that the process is not the important thing either, I gather Behaviorism may disagree somewhat here?)

    Neither to my understanding does Wittgenstein seem to believe that behavior (such as recalling a song) happens spontaneously in the sense that it follows from nothing. You seem to be saying that nothing can be talked of as happening suddenly or spontaneously (at least no human behavior) because to a behaviorist there is a certain process that always goes on, and this process is not spontaneous, but tightly ordered (though complex.) But Wittgenstein here is I think using the word "suddenly" as it is usually used and not assigning a claim of metaphysical spontaneous generation of tune memory to it. In part 2 (I apologize that I keep quoting from part 2 since I suggested you just read part 1) he says the following:

    'Would this situation be conceivable: someone remembers for the first time in his life and says "Yes, now I know what 'remembering' is, what it feels like to remember".—How does he know that this feeling is 'remembering'? Compare: "Yes, now I know what 'tingling' is". (He has perhaps had an electric shock for the first time.)—Does he know that it is memory because it is caused by something past? And how does he know what the past is? Man learns the concept of the past by remembering.
    And how will he know again in the future what remembering feels like?'

    He says this in the course of arguing that "Remembering has no experiential content." If he does not even think that somebody can remember without prior experience with remembering then I don't see how he could be said to think that rembering happens truly spontaneously.

    Wittgenstein's point in 184 isn't that knowing what it would be like to suddenly recall a song would tell us anything-- it's that the song isn't what remembering consists in, neither is the experience or feeling of remembering it, and neither is the psychological memory of it. Some of these things are not even there. What is there are certain behaviors in particular contexts, with various accompaniments (such as the behavior of singing the song, or a feeling of satisfaction at remembering it.) Wittgenstein is not concerned with what causes the song to be recalled, but with how the phrase "Now I know it" can have meaning, and what our concepts of "knowing" and "remembering" consist in.