Monthly Archives: January 2018

The Quine/Skinner discussion.

In 1981 B.F. Skinner held a series of small seminars with his colleagues to discuss a variety of different philosophical and scientific topics. In one of the seminars behaviourist philosopher Willard Quine was a guest. Other members of the discussion that day were the linguist Pere Julia, as well as another unidentified male guest (perhaps Gerald Zuriff who attended most of these discussions), Margret Vaughan and Will Vaughan were present as was the Swedish linguist and poet Lars Gutafson.
The discussion begins with an analysis of motivation, reasons and causes. One of the central themes is Skinner’s distinction between rule-governed behaviour and contingency shaped behaviour. Skinner first made this distinction in his 1967 book ‘Contingencies of Reinforcement’:
“Society codifies its ethical, religious, and legal practices so that by following a code the individual may emit behaviour appropriate to social contingencies without having been directly exposed to them. Scientific laws play a similar role in guiding the life of scientists…Discriminative stimuli which improve the efficiency of behaviour under given contingencies of reinforcement are important, but they must not be confused with the contingencies themselves, nor with the effects of those contingencies…The behaviour of one who speaks correctly by applying the rules of grammar merely resembles the one who speaks correctly from long experience in his verbal community.” ( Contingencies of Reinforcement. p. 125)
Skinner discussed the matter further in his (1974) ‘About Behaviourism’:
“To say that “The child who learns a language has in some sense constructed a grammar for himself” is as misleading as to say that a dog which has learned to catch a ball has in some sense constructed the relevant part of the science of mechanics. Rules can be abstracted from the reinforcing contingencies in both cases, and once in existence may be used as guides. The direct effect of the contingencies is of a different nature…There are then two extremes: (1) Behaviour shaped only by the contingencies of reinforcement, in which case we respond “unconsciously”, and (2) rule governed behaviour in which the contingencies from which these rules are derived may not have affected us directly. Between these extremes lie a wide range of degrees of “awareness”. (ibid pp 126-128)
We can see from above the importance Skinner made of the distinction of behaviour governed by explicit rules codified in language and behaviour caused by the contingencies of reinforcement. From at least 1967 Skinner was making a clear distinction between a person’s behaviour fitting rules, and people’s behaviour being guided by rules. Quine made a similar distinction explicit 5 years later in his paper ‘Methodological Reflections on Current Linguistic Theory ’:
“My distinction between fitting and guiding is, you see, the obvious and flat-footed one. Fitting is a matter of true description; guiding is a matter of cause and effect. Behaviour fits a rule whenever it conforms to it; whenever the rule truly describes the behaviour. But the behaviour is not guided by the rule unless the behaver knows the rule and can state it. This behaver observes the rule” (Quine: Methodological Reflections on Current Linguistic Theory’ p. 386).
So at the time of the discussion both Skinner and Quine both were very concerned with accounting for the distinction between rule-governed behaviour and behaviour that was entirely causal. Below is the audio recording of the discussion

Below is a transcript from the first 15 minutes of the conversation ( which should be read while listening to the audio recording) where they discuss reasons, causation and motivation. The rest of the conversation hasn’t been archived I will archive it at some later stage and post it here.

Pere Julia discussion group:
First section: An analysis of Reasons, Causes and Motivation:
Quine: That’s what I meant by reference to drive. There will still be something in the way of motivation…maybe that’s true of all caused behaviour.
Julia: You did it. And you may verbalise the wrong reasons. I did it for such and such a reason but as the Freudian would say you did it for very different reasons.
Skinner: That was Freud’s little …he would indicate different reasons for doing something than the reasons you gave. But the reasons he would have given…If they are unconscious reasons, then they must be what I have been calling causes and not reasons because they are not verbal.
Julia: So you have causes on the one hand which would be a list of the variables that lead to the behaviour. Now you might describe these causes and then you would be giving reasons for having done it. But these reasons may not be the real causes. So we would have an intersection of two sets but they are not necessarily the same set.
Skinner: When you were young you were reinforced in many ways in the presence of your mothers face. Then you grow up and you fall in love with a woman who looks like your mother looked at that time. That’s all a matter of cause in the sense that it’s just a behavioural process. But you give reasons why you find her beautiful. Then Freud turns up and says aha you overlooked the fact that she looked like your mother when you were a child. So that the contingencies, the unanalysed, un-rationalized contingencies, were simply the fact that that person is attracted to her and that is where motivation comes in, whether the mother was feeding you caressing you and so on, and you go for that kind of person, and you go for this woman. But then you give a lot of other reasons, she is very intellectual, you enjoy talking to her, and so on. You give all sorts of reasons, which as Freud would point out are not the real reason you married her. You married for unconscious reasons, and the fact that it is unconscious means that you haven’t talked about it, and couldn’t talk about it, without converting it into reason governed behaviour. The real reason you married her was because she looked like your mother and as soon as you say that it is now a description of the contingencies.
Quine: Is this only difference? Would you say that reasons are causes verbalised by the subject himself? And when we say causes we mean real causes, and these aren’t kidding himself.
Skinner: These are behavioural processes unwinding, that is what we mean by cause.
Quine: So reasons would be separated, the way that the notion of cause separates reasons from false rationalisations. False rationalisations and reasons have in common that they are verbalised, but reasons are distinguished from the false rationalisations in that they really causes as well…so all reasons are causes.
Skinner: It’s very important that you don’t get people to do anything by giving a reason. That doesn’t have what you would call a motivation. That is why you advise someone in therapy. Let’s say he is a pathological gambler you can say “you should give up gambling and that is a good reason. If you give up gambling you won’t lose money”. But that doesn’t mean that he is going to take that knowledge and act on it and stop.
Quine: That’s saying that they are verbalised by the subject, not the advisor but by the subject himself.
Skinner: Well the reasons could be imparted. When you tell some one, when you give someone reasons to do things. You are hoping to change the behaviour by giving the reason. But unless there is some reason to follow the reason, unless there is a cause there, that would be “you do this or else” it could be that kind of cause. You describe the behaviour, if you can threaten a person will do it, if you can say and they have news for you, that would be a way of getting them to do it because in the past when people have said things you tend to do things and so on. But a mere statements of contingencies may not be enough.
Quine: Well now try this one. A reason is a verbalised cause (the verbalisation may be through someone else) such that the subject accepts the verbalisation and is aware that it is the cause. Of course the trouble here is that awareness comes in.
Skinner: Then knows what will happen if. But that does not mean that there is any disposition to do it. That is where the motivational side is missing.
Julia: I think that Professor Quine said the key phrase before and that is when subject is speaking about himself and is not kidding himself reasons and causes would appear to be the same thing even if he is listing and describing the causes whereby he is doing something. And if he is indeed not kidding himself then the two things would be the same. Then we have to talk about the case of somebody falling in love with someone who resembles his mother. He may not be aware of it, then the causes would be a broader set of things than the reasons he gives he would give for his having falling in love with; they may not coincide. So the question is whether he is kidding himself or not, whether he knows himself or not. On the one hand so far as the speaker is speaking about himself, and when we knew some of the ultimate, we can give good reasons, we can describe the contingencies he is following but that doesn’t mean we will be effective.
Quine: If they are effective then there reasons again.
Person B: Ordinarily when a person gives a reason there is a step missing to get from that reason to what we consider to be the cause, you still have take what the person says and do some kind of translation or something to get to the independent variables. Rarely when a person is giving reasons would he state what a scientist would accept as independent variables.
Julia: Well paraphrasing it into technical language.
Person B: Is that always possible?
Julia: If he is not kidding himself, he has given the reason, so long as it satisfies our translation.
Person B: Ok. Now do valid reasons when you are not kidding yourself always translate into a scientific analysis.
Julia: Do you have an example?
Person B: Well when people give purposes when people explain why they do it because of wants, desires, plans, thought and so on. We don’t know that all that can be translated, first of all because we don’t have a complete science, and secondly because nobody has done this.
Julia: Well you would have to review case by case. Maybe we would run into examples which would defy translation.
Person B: It would be odd if people just intuitively know what is going to turn out to be the scientific explanation for why they behave as they do.
Julia: No but very often people do describe the reasons why they do things simply because in the past they have been trained to observe their own behaviour in relation to causes and why did you do it, to whom did you speak. That is where self descriptive repertoire come in.
Person B: Are we saying that contingency shaped behaviour where no rules are involved. Now that behaviour can also have reasons.
Skinner: You can extract them from the contingencies.
Person B: And the person in fact himself after having done the behaviour can give a reason even though the behaviour may have been contingency shaped.
Skinner: Yes and he may continue to use a statement about the contingencies in order to keep himself going.
Person B: Then it becomes rule governed.
Skinner: I use an old example of a medieval blacksmith who discovers how to use the bellows. The bellows are near the fire and he himself discovers just by the contingencies that you may as well go up quickly as there is no air coming out as you are doing that and you don’t down to too fast in order to get a steady flow of air. Then he makes a little poem “Up high down low, up quick and down slow that’s the way to blow”. But then he tells the apprentice the poem. The apprentice is only following the rules; he is doing what he was told to do. The blacksmith is doing it first of all because the fire blows well when he does it this way, then he describes his own behaviour, and that is useful to him.
Quine: In fact this example brings out another complication in this concept. Namely, the apprentice has his reason for working the bellows in the ways that he does it. But it isn’t because he wants to steady the flame, it is because he is following the blacksmiths rule.
Skinner: No exactly. See now the rule has taken over entirely. The blacksmith does it both ways, he gives himself additional assurances. The redundant cause is to do it the right way and he may find himself getting careless and doing it the wrong way. But the apprentice’s behaviour is entirely governed by the description of the contingencies; the description of behaviour and the consequences. But the blacksmith, I suppose many blacksmiths before there was verbal behaviour, was doing something like that only because of the physical contingencies.
Quine: Now what would we say was the apprentices reason?
Skinner: I wouldn’t want to use reason. I would simply say that a certain kind of behaviour was reinforced by a steady fire.
Quine: I was speaking of the apprentice.
Skinner: Oh the apprentice. You have to give him a reason. The point is you can tell him that he now knows how to blow. But it isn’t going to do him any good. Knowing how is not enough you have got to give him a reason. You signed a contract in the old days and if you didn’t do it you got a beating.
Julia: I guess that’s Quine is getting at, why should he be doing it?
Quine: And that beating may never have been verbalised.
Skinner: That’s true. I don’t mean to say that there was anything that was not reinforced.
Quine: What I am worried about now is that here we have something that we would like to call a reason, namely the apprentice blows the bellows the way he does so he won’t be punished and that’s his reason. But that never did get into words. So verbalisation is not a necessary condition of something being a reason.
Skinner: No but what he is doing is; doing as directed with words, imitation would have been enough. With imitation you wouldn’t need words to demonstrate. But if you are writing it and you can’t demonstrate you have got to use words and then were getting into words. But you always have to take imitation as a special case where you induce someone to behave for your reasons, not for his, until his reasons take over. I am using ‘reasons’ wrong again there…this is very confusing…
Person B: There is another aspect to this every time he does it wrong you could whip him. So that again wouldn’t respond to the fire he would respond to the whip.
Skinner: Well you could of course do this by shaping up his behaviour. He is hungry and you have bits of food, he wonders around and when he puts his hand on the bellows you give him bits of food. And then you do it again… you could eventually shape this up, you could do this in a monkey for example without words at all. And that would be now just getting someone doing what you wanted him to do without resorting of the contingencies; fire, bellows etc.
Second Section the practical consequences of understanding motivation and reasons:
This section is a discussion of the consequences of Skinner and Quine’s take on rule-following and motivation to practical problems in environmentalism, ethics, etc.
Third Section a discussion of forms of philosophical discourse: the dialogue, the essay etc.
In this section they discuss philosophy, poetry, and the effect of form on philosophical reasoning.

Skinner: Autoclitics and expository description

Some philosophical influences on Skinner’s analysis autoclitics in ‘Verbal Behaviour’: Carnap ‘The Logical Syntax of Language’, Russell ‘Inquiry into Meaning and Truth’, Tarski ‘The Concept of Truth in Formalized Languages’, Quine ‘Elementary Logic’, Mathematical Logic’, John Horne Tooke ‘The Diversions of Purley’.
Some critics of Skinners analysis of autoclitics in ‘Verbal Behaviour’: Chomsky ‘Review of Verbal Behaviour’, Steven Hayes ‘Advances in Relational Frame Theory’.
“Such “propositional attitudes” as assertion, negation and quantification, the design achieved through reviewing and rejecting or emitting responses, the generation of quantities of verbal behaviour merely as such, and the highly complex manipulations of verbal thinking can all, as we can see, be analysed interms of behaviour which is evoked by or acts upon other behaviour of the speaker” (Skinner ‘Verbal Behaviour’ p.313)

(1) Descriptive Autoclitics:
“The speaker may acquire verbal behaviour descriptive of his own behaviour. Although the community can establish such a repertoire only by basing its reinforcing contingencies on observable behaviour, the speaker eventually exhibits it under the control of private events…We shall refer to such responses, when associated with other verbal behaviour effective upon the same listener at the same time, as “descriptive Autoclitics” The term autoclitic is intended to suggest a behaviour which is based upon or depends upon other verbal behaviour.” (ibid p. 313-315)

Skinner notes, sensibly enough, that the contingencies necessary for self-descriptive behaviour are arranged by the social community. The community asks us a variety of different questions “Did you say that?”, “Why did you say that?” etc. Because getting us to answer such questions are useful in many ways in understanding and predicting behaviour. As a result of being implicitly trained by the community to answer these questions, we eventually begin to ask such questions about our own Verbal Behaviour. Autoclitic behaviour is behaviour that is based on or depends on other verbal behaviour. While descriptive autoclitics inform us of the type of verbal behaviour we are omitting (whether Mands, Tacts, etc). There are many different types of autoclitics. (A) A descriptive autoclitic that informs the listener of the type of verbal operant that it accompanies. Some examples: A reader is reading a news paper and says I see it is going to rain, the I see informs the listener that it is going to rain is omitted as a textual response (ibid p. 315). Skinner notes that behaviour that is acquired as textual or echoic behaviour but emitted as Intraverbal behaviour is often prefaced by ‘I see’, ‘I recall’, ‘I am reminded’, ‘I hear’. He notes other examples of descriptive autoclitics such as ‘I demand’, ‘I ask you’ which when prefacing mand make it more effective. He gives examples such as ‘I tell you’, ‘I observe’ which when they preface tacts make them more effective. In these examples the autoclitics which preface the tacts, mands are not necessary but they do make the verbal behaviour more effective. (B) This type of descriptive autoclitic describes the strength of a response. Examples include ‘I guess’, ‘I estimate’, ‘I believe’, ‘I think’, ‘I hesitate to say’ etc. The preceding descriptive autoclitics indicate that what follows is based on insufficient information. Skinner calls the preceding examples autoclitics of weakness. Examples of autoclitics of strength include ‘I insist’, ‘I swear’, ‘I promise’ etc. (C) This type of descriptive autoclitic describes the relation between a response and other verbal behaviour of the speaker or listener, or other circumstances where behaviour is emitted (ibid p. 316). Some key examples are ‘I agree’, ‘I confess’, ‘I infer’, ‘I predict’ ‘I dare say’ ‘I wish’, etc. These responses are helpful for the listener. They help the listener to situate the response which follows to other aspects of the current situation. (D) Another type of descriptive autoclictic indicates the emotional state of the speaker, these type of autoclitics indicate the personal relation between the speaker and the listener. Examples are ‘I regret to inform you’, ‘I hate to say’, ‘I must tell you’ etc. (E) Negative autoclitics qualify or cancel the response which they accompany. Some examples are ‘I don’t think that he has gone’ ‘I would not go as far as to say’ ‘I doubt’ ‘I deny’ etc. (F) Another autoclitic indicates that what is to follow stands in a subordinate position. Examples include ‘for example’, ‘for instance’ etc.
(2) Qualifying Autoclitics:
An important class of responses serve the autoclitic function of qualifying the tact in such a way that the intensity of direction of the listener’s behaviour is modified. There are two main types of qualifying autoclitics. (A) Negation: As Skinner notes philosophers have long tried to analyse the notion of ‘negation’, (this practice goes back at least as far as Parmenides). One of the difficulties is in trying to understand what sentences which include ‘not’, ‘no’ etc refer to. Skinner discussed the sentence ‘It is not raining’. A difficulty occurs when we try to pick out what the sentence is referring to. A possible solution is that the sentence is evoked by a lack of rain in the environment. However this leads to the obvious difficulty of explaining how the infinite amount of things we don’t encounter in our environment don’t likewise compel us to mouth ‘It is not xing’, ‘there is no x’ etc. Skinner, following Russell (whom he cites) notes that an obvious solution to this problem is to argue that sentences like ‘It is not raining’ are evoked by not by environmental conditions but by other verbal behaviour such as ‘Is it raining today?.’ While Skinner thinks that this solution is to some degree correct he also notes that some non verbal stimulus can evoke the response ‘no’. A clear instance of this would be ‘no’ used as a mand to stop a person from engaging in some kind of non-verbal behaviour. This response is also extended to verbal responses such as when a child says ‘2 + 2= 5’. On the response ‘no’ Skinner notes:
“The response is acquired from the reinforcing practices of the verbal community. The child first hears ‘no!’ as the occasion upon which some current activity must be stopped if positive reinforcement is to be received or aversive stimulation avoided” (ibid p. 223)
The child may find himself doing something that typically elicits a ‘no’ from others, the child says ‘no’ himself and this results in him not engaging it the behaviour. This activity will result in the child receiving less punishment and will therefore be reinforcing. Skinner speculates that this practice of saying ‘No’ will eventually be used along with other verbal behaviour such as saying things like ‘Red’ etc. This will lead to the child saying ‘Not Red’ in appropriate circumstances, because of the standardisation in linguistic practices, and punishment and negative reinforcement for those who don’t follow such practices. (B) Assertion: The assertive autoclitic enjoins the listener to accept a given state of affairs (ibid p. 327). Skinner argues that since the assertive autoclitic enjoins someone to do something then it must be considered a special kind of mand. However he qualifies this by noting:
“An autoclictic will sharpen the effect by indicating some of the source of strength, as well as the degree of strength. The assertive autoclitic has the specific function of indicating that the response is emitted as a tact or, under certain circumstances, as in Intraverbal. Other verbal operants are characteristically not asserted. The mand does not need to be, because of the reinforcing contingencies which are responsible for it, and in echoic and textual behaviour the important conditions for the listener are those which prevailed when the echoic or textual stimulus was produced by someone else.” (ibid p. 327)
(3) Quantifying Autoclitics:
Skinner gives as examples of Quantifying autoclitics ‘All’, ‘Some’, ‘A’, ‘The’ etc. In the case of ‘All’ he considers the example ‘All Swans are White’. He notes that when doing logic we would be justified in arguing that ‘All’ modifies ‘Swans’, however things are different when we are concerned with Verbal Behaviour. His reason for arguing thusly is that because in a scientific account of verbal behaviour we cannot assume that a person ever responds to ‘ALL’ Swans. It is more reasonable to say that a man responds to all of the swans in his own personal history. Skinner argues that in this case we are better off interpreting the ‘All’ as always it is possible to say. Thus the ‘All’ will modify the whole sentence ‘Swans are White’ not just the ‘Swan’. Likewise you can translate ‘Some’ as ‘sometimes it is possible to say’. Again the ‘Some’ modifies the whole sentence not just the ‘Swan’. And ‘No’ can be translated as ‘it is never possible to say’, where the ‘No’ modifies the whole sentence not the ‘Swan’ part. The other common quantifying autoclitics are ‘The’ or ‘a’ which serve to narrow the reaction of the listener by indicating a response and the controlling stimulus (ibid p. 329). Quantifying autoclitics typically serve the purpose of modifying the reaction of the listener to the responses they follow.
(4) Relational Autoclitics:
Skinner argues that the “agreement” in number, gender and case between the noun and adjective a language like Latin is a paradigm of a relational autoclitic (ibid p.333) (A) Predication: Skinner notes that a predication occurs when a relational autoclitic is added to an autoclitic of assertion (ibid p. 334) Thus, for example, he notes that the statement ‘The chocolate is good’ shows a relational autocliitic of grouping and ordering and it also contains an autoclitic of assertion. He argues that when you take these together you get a predication (ibid p. 335). (B) Relational Autoclitic Behaviour: This leads Skinner to ask an important question. What are the processes that lead to the emission of a relational autoclitic behaviour?
“Something less than full-fledged relational autoclitic behaviour is involved when partially conditioned autoclitic “frames” combine with responses appropriate to a specific situation. Having responded to many pairs of objects with behaviour such as ‘the hat and the shoe’ and ‘the gun and the hat’ the speaker may make the response ‘the boy and the bicycle’ on a novel occasion. If he has acquired a series of responses such as ‘the boys gun’, ‘the boy’s shoe’, and ‘the boy’s hat’, we may suppose that the partial frame ‘the boy’s_’ is available for recombination with other responses. The first time the boy acquires a bicycle, the speaker can compose a new unit ‘the boy’s bicycle’. This is not simply the emission of two responses separately acquired. The process resembles the multiple causation of Chapter 9. The relational aspects of the situation strengthen a frame, and specific features of the situation strengthen the responses fitted into it.” (Verbal Behaviour p. 336)
(5) Manipulative Autoclitics:
“It is only upon genuinely novel occasions that the listener is specifically manded to modify his behaviour. But these occasions do occur, and the explicit autoclitic activity of the speaker in manipulating his behaviour must be taken into account as an important verbal function” (ibid p. 343)

Some examples are ‘for’, ‘but’, ‘if-then’, etc. Quine deals with these examples early in his ‘Elementary Logic’.