“If this is rationalism, and incompatible with Locke’s empiricism, then so much for the rationalism, and so much the worse for Locke…For, whatever we make of Locke, the behaviourist is knowingly and cheerfully up to his neck in innate mechanisms of learning-readiness. (Quine: Linguistics and Philosophy p. 57)
In this blog I will discuss some basic concepts of Applied Behavioural Analysis: reinforcement, punishment and extinction. While summarising these concepts I will consider whether any of them are committed to the idea of a blank slate. Pinker in his (2002) ‘The Blank Slate’ targeted behavioural science as a blank slate position which denied human nature. I will argue here that Pinker was incorrect on this issue, behaviourism isn’t necessarily a blank slate position in fact it is largely consistent with both evolutionary theory and the idea of human nature. I will not consider whether Pinker is correct or incorrect in his attacks on other scholars and disciplines, my focus here will be entirely on behaviourism. When discussing behaviourism I will outline of how some concepts of behaviourism are used in the field so I will discuss a lot of Applied Behavioural Analysis.
Reinforcement is when a stimulus immediately follows the behaviour and increases the likelihood of that behaviour reoccurring in similar circumstances. It is a three term relation consisting of: Responding, Consequence, and Change in Behaviour. There are two types of reinforcement: positive reinforcement and negative reinforcement. Positive reinforcement occurs when a response is followed immediately by the presentation of a stimulus and this presentation increases the future occurrences of said behaviour. Negative reinforcement occurs when a response is immediately followed by the removal of a stimulus and this removal increases the likelihood of the behaviour increasing in the future.
An example of positive reinforcement is as follows: Every time subject A blinks subject B who he is talking to smiles, this results in an increase in the occurrence of blinking behaviour by subject A. An example of negative reinforcement is: Every time subject A feels intense hunger he eats some food, this behaviour removes the feeling of hunger and means that the response of eating when hungry is more likely to occur in the future.
When considering reinforcement (both positive and negative) it is important to take note of the distinction between primary and secondary reinforcers. A primary reinforcer is a basic physiological need such as hunger, thirst, sexual desire, etc. Now it should by admitting unconditioned reinforcers into their theory behaviourists are not blank-slaters, they openly admit that our evolutionary history will have shaped our instincts so that certain behaviours will occur independent of learning from our current environment. Skinner is quiet explicit on this point
“Just as we point out the contingencies of survival to explain an unconditioned reflex, so we point out to ‘contingencies of reinforcement to explain a conditioned reflex” ( Skinner 1974 p. 43)
“The task of a scientific analysis is to explain how the behaviour of the person is a physical system is related to the conditions under which the human species evolved, and the conditions under which the individual lives” (Skinner 1974 p. 14)
Here we can see that Skinner far from advancing some silly blank slate position holds a theory and whose methodology relies on the notion of instincts from the start. Secondary reinforcers can be use to modify some behaviour. A secondary reinforcer is something that becomes reinforcing because of being associated with the primary reinforcer. A good example of a primary reinforcer is hunger; if through using classical conditioning I associate relief from hunger (negative reinforcement) with a particular form of behaviour (pressing a lever), pressing a lever will become a secondary reinforcer because of its association with the primary reinforcer.
Careful observation and recording of what behaviours are reinforcing and what behaviours are not is vital. So, for example, while food may be reinforcing for a child who has not eaten for a while it is not likely to act as reinforcer for the child who has just eaten a big dinner. So when choosing reinforcers take into account both satiation and deprivation. Satiation occurs when a desire is satisfied fully, while deprivation occurs when you are deprived of a particular reinforcer for a sustained period. Obviously ethically when one is working with people we should not implement a schedule of deprivation to make reinforcement more likely. Rather we should observe the behaviour and schedule treatments at appropriate times. If, for example, one wanted to use food as a treatment, then it would be best to implement the treatment a couple of hours after dinner or lunch to ensure that the child will be slightly hungry.
Both positive and negative reinforcement are extremely useful ways of dealing with behavioural issues associated with autism. One area where it is particularly helpful would be in the implementation of Discrete Trial Training. Discrete Trial Training is a method of teaching in simplified and structured steps. Instead of teaching in one go, the skill is broken down and built up using discrete trials that teach each step one at a time (Smith 2001). There are five steps ( Malott and Trojan Suarez 2006, Smith 2001) in a discrete trial:
- Inter-trial interval
Step three is the key one as it relies heavily on the notion of reinforcement. Let us consider a very simple trial where a child is being taught to discriminate between a picture of a Dog and a picture of a Cat. In the antecedent the teacher has a picture of a Cat and a picture of a Dog. The teacher points to the picture of the Cat, and says “this is a picture of a cat”. With the prompt phase the teacher may point to the picture of the Cat as he asks the child to “Point to a Picture of a Cat”. The response phase involves the child trying to answer the question. If the child answers correctly he receives positive reinforcement (for example a sweet or a high five depending on what has been determined to be the best reinforcer for the child). If the child answers incorrectly the instructor will bring the child’s hand to the correct picture and ask the child again. When the child replies correctly based on the second prompt reinforcement is withheld, as you don’t want to reinforce the child for getting the question wrong in the first place. So in this particular training schedule the child is only reinforced for getting the correct answer.
All behavioural therapies rely heavily on using reinforcement whether positive or negative. However, some therapies use the concept of extinction to remove reinforces that lead to various types of challenging behaviour. Extinction can be used to remove different types of behaviour whether positive or negative. So, for example, if a rat in a lab is trained to press a lever through receiving the reinforcement that every time he presses the lever a pellet of food appears, then he will learn through reinforcement to press the lever when hungry. If a behavioural scientist wanted to get rid of the lever pressing behaviour he would remove the reinforcing stimulus that is maintaining the behaviour i.e. the food.
Using extinction is not however without its dangers. One danger is the occurrence of an extinction burst. An extinction burst typically occurs when extinction is used to remove a particular behaviour. This burst is typically short term and after a period the targeted behaviour typically disappears altogether. There is always the possibility of spontaneous recovery after a period of time but the main thing is to not reinforce the recovered behaviour, and it will typically disappear after again.
There are three basic steps to applying extinction to decrease a response.
- Observe what happens and identify the reinforcer.
- Determine how often a particular behaviour occurs within a particular period.
- Remove the reinforcer. (Miltenberger ‘Behaviour Modification’ p. 105)
Step 1 above is not a trivial matter and if one is trying to determine what the reinforcer is there are six key steps that should be taken:
- Determine the baseline rate of the response.
- Identify the potential positive reinforcer for the response
- Withhold the stimulus that you have identified as the potential positive reinforcer for the response.
- Measure the rate of the response. If it decreases (even after an initial increase), it is likely that the stimulus you identified, is the positive reinforcer for the response.
- If it is important to demonstrate that the stimulus is the actual positive reinforcer, reinstate the stimulus following the response.
- Measure the rate of the response. If it increases again, the stimulus has served as the positive reinforcer for the response. (Chapter 4 extinction p.37.)
Extinction is an effective procedure for decreasing challenging behaviour but it typically needs to be implemented with other procedures. In Lerman et al. (1999) they studied the effects of implementing extinction programmes they noted that typically treatment with Operant Extinction leads to two main problems:
- Extinction Burst:
- Extinction Induced Aggression:
In their analysis of the 41 different cases of people who engaged in SIB they discovered that 40% of the people being treated had one of the side effects (almost 20% of the cases showed both side effects). Bursting was slightly higher than extinction induced aggression. It is important to note that both side effects were less likely when the extinction was combined with other therapies. These findings show that side effects may be minimised when extinction is combined with other treatments (e.g. differential reinforcement). So their study indicates that while extinction treatment can be very effective it is best to be combined with other forms of therapy.
In connection with extinction it is important to note how the use of it is clearly not some hippy-dippy blank slate position. Behavioural Analysts don’t just assume that applying some technique in a route manner will automatically modify the behaviour of the subject understudy. The subjects under study can react in a variety of different ways, as seen above it may happen that prior to stopping the behaviour there is an extinction burst, or the extinction induced aggression. What should be clear that in practice analysts don’t just assume they are dealing with a blank slate, rather they take meticulous data to indicate what behaviour modification is likely to produce what behaviour and they modify their theories accordingly. So, for example, Fisher et al (2004) have done a experimental study whose results indicate that dual treatments result in greater reductions in target behaviour than did extinction alone and suggested that competing stimulus assessment may be helpful in predicting stimuli that can enhance the effects of extinction when non-contingent attention is not available.
They begin by talking about the limitations of the extinction method. That it takes away reinforcements from the patient. That it sometimes leads to Extinction Bursts, and Extinction Related Aggression. They speak of how using Non-Contingent Reinforcement can help overcome some of these problems. Non-Contingent Reinforcement involves discovering what the reinforcer is that the child is seeking with his aggressive behaviour. Once this is discovered the reinforcer is presented to the child at fixed time intervals independent of the child’s behaviour.
It was been discovered that when Extinction is done alone Extinction Bursts sometimes occur, while when extinction is done along with Non-Contingent Reinforcement treatment the Extinction Bursts and Extinction Related Aggression don’t occur. A difficulty with the NCR approach is that the parent may not be able to deliver the particular reinforcer at times when it is needed to offset potentially destructive behaviour. As a response to this difficulty an approach has been developed which aims to develop arbitrary or competing reinforcers that may be useful in these circumstances.
They compared the effects of:
- Extinction Implemented Alone.
- Extinction Implemented with non-contingent delivery of the reinforcer that maintained destructive behaviour.
- Extinction Implemented with non-contingent delivery of competing stimuli (those identified with competing stimulus assessment)
They found that non-contingent presentation of the reinforcer that maintained problem behaviour or a competing reinforcer in combination with extinction can produce rapid reductions in destructive behaviour.
Again the above study shows that multiple techniques and empirical tests are used when trying to modify behaviour. People are not treated as blank slates. If a technique does not reduce the behaviour, various different experimental procedures are used to improve the technique. Behaviour analysts who spend years working with people with severe intellectual disabilities and/or autism, don’t expect miracles. It is a slow procedure that while it can and does improve the lives people, doesn’t provide cures for severe developmental disorders.
Another way of modifying unwanted behaviour is through punishment. In her blog ‘The Difference between Positive and Negative Reinforcement and Positive and Negative Punishment’ Kelley Prince defined punishment as follows:
“Punishment is also a useful way of modifying undesirable behaviour. A key difference between negative reinforcement and punishment is with negative reinforcement you are increasing the behaviour whereas with punishment you are decreasing the behaviour. Punishment is a process by which a consequence immediately follows a behaviour which decreases the future frequency of the behaviour. Like punishment a stimulus can be added (positive punishment) or removed (negative punishment)”.
The important point to note is that the concept of punishment is not a concept that has to do with the intrinsic negativity of a consequence. Rather punishment is simply a contingency between an antecedent, a behaviour and a consequence. If we are to count something as positive punishment then the probability of a given behaviour occurring must decrease after a particular stimulus is presented as a response to the behaviour. So, for example, if every time a child talks out of turn in class we sent them the back of the room, this will count as a punishment if this consequence results in a decrease in the behaviour of talking out of turn. If however this consequence does not decrease the child’s behaviour then we cannot call the consequence a punishment even if it intuitively feels that it is a punishment. The same is true of negative punishment. In negative punishment something that is typically reinforcing is removed as a result of behaviour that we do not want. So if a child is talking out of turn in class and we know that the child finds attention reinforcing we could remove the attention from the child. If the removing of the attention results in the child decreasing the behaviour of then this is a form of negative punishment.
In his ‘On The Status of Knowledge for using punishment: A Commentary.’ Horner (2002) correctly notes that punishment is an effective treatment for problem behaviours. However it has some down sides. It can lead to escape and avoidance procedures. It can also sometimes result in negative emotional and aggressive actions. For that reason punishment is only used when it positive or negative reinforcement have been shown to not work. Furthermore when punishment is used it is combined positive reinforcement it usually works better because you are not just punishing the subject for bad behaviour but are also rewarding the subject for good behaviour.
His major thesis is that development of a comprehensive technology of behaviour change requires more detailed research on punishment with special attention to (a) conditioned, intermittent and delayed punishment, (b) Interaction effects of reinforcement and punishment, (c) the value of functional analysis in the design of punishment interventions, (D) greater attention to treatment failures to identify how punishment may be used more effectively in clinical interventions.
His central point is that punishment is a fact of life. It occurs in all areas of organism interaction with each other and the world. Yet we have not researched into the nature of punishment nearly enough. He argues that this is a big gap in the literature that needs to be addressed. So punishment like extinction is a form of treatment that is effective but can be have negative consequences when used on its own. It is vital that more research is used in this tool in order to achieve a better understanding of how to modify things like self injurious behaviour.
Again with punishment we can see with Horner (2002) above that the use of punishment is linked to the evolutionary history of our and other species. So there is no denial of instincts being passed on because of the different evolutionary history of different species. Furthermore different species may find different things as punishing, and so may different people. Whether something is considered a punisher is to be determined by the effects on the subject being studied. Again there is no denial of human nature no claim that humans are blank slates, just the uncontroversial method of deciding what a subject finds punishing by looking at their behaviour.
In this short blog I have outlined a few key concepts in behaviourism and showed that these concepts do not deny human nature nor and are consistent with evolutionary theories. Thinkers like Steven Pinker have attacked and caricatured behavioural science and has convinced many people that it is an absurd blank slate position. In this brief blog I have tried to show that behaviourists are anything but blank-slate thinkers. I think that if Pinker and his ilk want to criticise behavioural approaches they will need to deal with actual theories of behaviourists and critique them as opposed to attacking straw men.
In (2011) Jerry Fodor co-authored a book with Piattelli Palarmi called ‘What Darwin Got Wrong’. The book was universally slated and its critics noted that the book was attacking a straw man. Evolutionary psychology was one of the fields that Fodor and Piattelli attacked, in my view they were attacking a straw man. I am sure that Pinker and his supporters agree. I find it interesting that Fodor and Piattelli use the construct a similar straw man when attacking behaviourism:
“In fact, the two theories are virtually identical: they propose essentially the same mechanisms, to compute essentially similar functions under essentially identical constraints. This raises a question about which prior discussions of NS have been, it seems to us, remarkably reticent: it is pretty generally agreed, these days that the Skinnerian account of learning is dead beyond resuscitation. So if it is true that Skinners theory and Darwin’s theory are variations on the same team, why aren’t the objections that are routinely raised against the former likewise raised against the latter? If nobody believes Skinner anymore, why does everyone believe Darwin? We are going to argue that the position that maintains the second but not the first is not stable.” (Fodor and Piattelli ‘What Darwin got Wrong’ (2011) p. 3)
I think that in both cases Fodor is simply caricaturing theories without engaging in them seriously. I am sure that Pinker agrees with me on the evolution aspect, but would support Fodor’s caricature of behaviourism. I would urge Pinker to look closer at behaviourism before making unjustified attacks on a promising discipline. I should note before finishing that I am not a behaviourist, but I think that if people like Pinker want to critique a discipline they should do so in a fair and objective way.
 Quotes taken from Ludvig (2002) ‘Why Pinker needs behaviourism’.
PECS AND VERBAL BEHAVIOUR
In this blog I want to briefly discuss PECS (Picture Exchange Communication System), which is used as a communication device to help people with various different intellectual disabilities and/or Autism communicate better. The PECS programme is heavily influenced by B.F Skinner’s (1957) book ‘Verbal Behaviour’. Now in virtually all cognitive science courses students are informed that Chomsky has conclusively refuted Verbal Behaviour. Later scholars like Stephen Pinker and Fodor (amongst many) have repeated the claim that Skinner was refuted by Chomsky, and Pinker has even upped the ante by claiming that Skinner was a blank slate adherent. Because of this consensus in cognitive science about the falsehood of Skinner’s Verbal Behaviour it is surprising that PECS is used so prevalently in the hospital I work in as well as in the majority of western hospitals to help children to learn to communicate better. The fact that Skinner’s theory of language learning is a useful tool in helping people with intellectual disabilities to communicate is surprising considering that his theory is generally considered to be conclusively falsified in the Cognitive Sciences.
For my PhD I did research into the debate between Chomsky and Quine on the nature of language. Like with Skinner, it is generally assumed that Chomsky has conclusively refuted Quine’s behaviourist conception of language. When I began my PhD it was with the assumption that Chomsky’s theory of language acquisition was superior to Quine’s, and that since Quine had such a huge influence on philosophical naturalists; exploring what aspects of Quine’s theory of language acquisition was false would be useful. But to my surprise I discovered that some of Chomsky’s arguments were absolutely terrible and didn’t prove anything like what him and his adherents believed. Furthermore I discovered that Chomsky’s criticisms of Quine were based almost entirely on misreading of Quine’s texts. A close analysis of Quine’s texts shows that most of Chomsky’s attacks on Quine were simple caricatures. Furthermore the empirical evidence amassing over the last 50 years has repeatedly refuted key claims of Chomsky’s about the inability of connectionist models to learn grammatical constructions, the supposed poverty of stimulus facing the child, and the amount of correction the child receives from their peers. I concluded that contrary to what is typically believed Chomsky has not refuted any aspect of Quine’s project and in fact the empirical evidence indicates that Quine’s approach is a live option which needs to be tested in much more detail than it has in the past.
Quine and Skinner were friends and in ‘Word and Object’ Quine claimed that his theory of language acquisition followed closely Skinner’s theory. However I am not an expert on Skinner so I do not know how far Quine and Skinner’s views on language coincide. Bearing that in mind I will not make the assumption that because Chomsky caricatured Quine he therefore caricatured Skinner. As far as I am concerned it is at this point an open question whether Chomsky is correct in his analysis of Skinner. So I will begin this blog by outlining the nature of PECS how it follows Skinner’s ‘Verbal Behaviour’ and how successful the approach can be in teaching some children language. I will then discuss Chomsky’s ‘Review of Verbal Behaviour’ and see whether the points he raised in that paper apply to the PECS programme. I will consider a variety of possible responses to Chomsky’s criticisms. In the last section I will deal with the objection that since PECS is typically used with people who have developmental difficulties its success or failure is irrelevant to Chomsky’s criticisms of Skinner.
PECS AND VERBAL BEHAVIOUR
Skinner argued that it is more useful to attend to the functional control of verbal behaviour than its form. He focused on four types of verbal operants which were defined in terms of consequences in certain kinds of stimulus conditions:
- Mands: Some examples of Mands are: Requesting specific Items, requesting assistance, and rejecting offered items or activities.
- Tacts: Tact is evoked by a particular object or event or property of an object of event. Some examples of Tacts are naming objects, events and activities, including the relationship between such items or events.
- Intraverbals: Intraverbal is used to define behaviour that is under the stimulus control of other verbal behaviour, initially from other people but increasingly from oneself as one’s verbal repertoire expands. These include echoic such as when I say ‘Hello’ and the other person says ‘Hello’, and non echoic such as when I say ‘How are you?’ and the other person says ‘Not too bad’.
- Autoclitics: These Verbal Operants are under the control of the speaker own verbal behaviour. (Bondy p. 313)
When thinking of the operants becareful to divide them into your 3 term relation Antecedent, Behaviour and Consequence (ABC). It is important to note that Skinner also discussed impure verbal operants and those under mixed or multiple controls are also postulated by Skinner (Bondy p. 311).
With Autism the behaviourist doesn’t assume that that people are blank slates, when people behave differently this fact is noted. Thus when it is noted that for young children with autism social reinforcement is less effective than other type of reinforcement the behaviourist will modify what type of reinforcement is used. So the aim is to use the most effective tools to modify behaviour. The Mand is usually the first verbal operant taught within PECS.
PECS Training and Verbal Operants.
Training on PECS begins with a reinforcer assessment where a behaviour analyst/teacher tries to discuss what objects are reinforcing for the particular child under assessment. The training typically involves three people. There is the subject being trained, the teacher/behavioural analyst who communicates with the subject who is being trained, also a third person (behavioural analyst/teacher) whose role is to physically prompt the subject being trained. The training begins with the communicator trying to entice the child with the item we know they want. When the child reaches for the item the prompter (from behind the student) physically assists the child to give a picture of that item to the communicative partner (Bondy p.320). When the child is being physically prompted his communicative partner keeps his hand open palm up to the child. When he receives the picture he says ‘Oh, Cookie’ and immediately reinforces the child by giving him the object. There is no verbal prompting prior to the exchange therefore this is a MAND not an INTRAVERBAL. Strictly speaking it is a MAND-TACT because the object is present when the child requests it. Bondy notes “The aim of Phase 1 is to shape the non-verbal reach into verbal behaviour directed to the listener rather than the reinforcer” (ibid p.321)
The physical prompts of the prompter are gradually faded (using backward chaining). Typically after a few trials the child is requesting objects using pictures with people who are within arms reach. When this occurs the behavioural analyst begins to move out of reach of the child so the child learns to seek out people to give the pictures to. At this early stage it is important that different teachers/behavioural analysts act as the communicative partner so the child doesn’t associate only one person with the act of giving pictures to request objects. It is also important to reinforce the child for moving towards the picture so he will learn to search for it if it is not immediately present.
The next part of the training introduces discriminating between pictures. One way of doing this is to show the child two pictures one of a highly desired object and beside a highly undesirable object. If the child really likes ice-cream and has been trained to use a picture of ice-cream as a MAND, then a picture of something irrelevant like a picture of a piece of wood. If the child picks up the picture of the piece of wood this indicates that he is not distinguishing between the actual content of the pictures and is rather using the plastic the picture is printed on as a general way of getting what he wants.
Mand/Autoclictic frame: The child is manding with the picture and a Velcro picture is added to beside the picture this new picture says “I want” (this is called a sentence strip) and the child is taught through backward chaining to use the “I want” picture in combination with this particular picture and then with a variety of other pictures. This is taught using the usual three person process where one of the analysts guides the child’s to add the picture (of say Ice-cream) to the ‘I want’ picture. When the child does this and hands the sentence strip to the teacher, the teacher reads out the sentence and gets the child what he wants. In this process the child is taught to place the ‘I want strip’ on the card and add pictures as he wants something.
The child is also taught to use interverbal tacts where “I see”, “I hear” “I have” is added to the sentence strip. Using these various techniques children with no language have been trained to become quiet competent users of verbal behaviour (some never speak because of problems with their vocal cord). This PECS programme has been very effective in teaching language to people with intellectual disabilities and autism see Adkins, T and Axelrod, S. (2002), Albis, J, and Reed (2012), MacFarland, S, and Umbreit, J (2012). So it is interesting to consider why this programme has been so successful given the supposed fact that Verbal Behaviour has been refuted by Chomsky.
Mac Corquodale Reviews the Review.
Kenneth Mac Corquodale analysed Chomsky’s arguments against Verbal Behaviour and showed that the arguments resulted from a misunderstanding of Skinner’s project. Skinner viewed ‘Verbal Behaviour’ as a research project which was to be tested experimentally. Chomsky though treated ‘Verbal Behaviour’ as a completed project and ridiculed the lack of experimental proof put forth by Skinner, this misunderstanding lead Chomsky to misread the book from the start.
Mac Corquodale noted that Chomsky’s Review has two parts:
- A criticism of the Analytical Behaviour that Skinner brought to Verbal Behaviour.
- A criticism of the application of Skinners theory.
Chomsky offered methodological arguments against Skinner and not empirical evidence. So in this sense the debate on the issue is still wide open. Mac Corquodale claims that Chomsky makes three methodological points against Chomsky:
- Verbal Behaviour is an untested hypothesis one which therefore has no claim upon our credibility. Mac Corquodale replies that Skinner knew well that his theory was a hypothesis, but he believed that it was a hypothesis that is plausible given the behavioural data he had amassed. Skinner believed that his hypothesis would stand up to experimental scrutiny but obviously we wouldn’t know until the hypothesis was tested. Chomsky on the other hand seems to think that because Verbal Behaviour is a hypothesis this a priori proves that it is false. This is a very bad argument. Chomsky also makes the claim that Skinner is technical terms with all the favourable connotations of sounding scientific; however the terms are really just metaphorical extensions vague ordinary language concepts. Chomsky criticises Skinner’s use of stimulus, he argues that the supposed relation between stimulus and response doesn’t exist. He claims that we often say words in circumstances where an object is not present etc. Mc Corquodale correctly notes that Chomsky is wrongly assuming that one stimulus is associated with one response, this is not what Skinner had in mind, as can be seen by a closer reading Verbal Behaviour. More points are also made firstly none of Chomsky’s arguments tell against the use of the term stimulus. The stimulus is real. The question is whether a particular word is brought under stimulus control. If Chomsky thinks Skinner is wrong to offer the hypothesis that a particular a particular word is under a particular stimulus control then he needs to test the issue. At this point in their careers both thinkers were forming hypothesis. Chomsky uses heavy rhetoric to undermine the possibility of any empirical testing. This again is an unscientific approach. Chomsky also attacked Skinner’s notion of ‘Reinforcement’. He argued that Skinner believed that slow and deliberate reinforcement of a child is necessary for a child to learn verbal behaviour. Skinner said nothing of the kind. He also argued that Skinner ignored the role that imitation played in a person learning language. But Skinner entirely agreed with this point (though he didn’t share Chomsky’s view that verbal behaviour was innate). Chomsky attacks Skinner’s use of probability saying that it is false. However Chomsky actually argues against a notion of probability that Chomsky doesn’t use (Hull’s version). Chomsky also doesn’t distinguish between the momentary probability that a word will be spoken with the overall probability that a word would be spoken. This leads to him badly misinterpreting Skinner’s theory.
- Skinner’s Technical terms are no more than simple paraphrases for more traditional treatments of Verbal Behaviour. Mac Corquodale argues that when one analyses the traditional terms and the functional relationships that Skinner talks about one sees that they are far far from isomorphic. So Chomsky’s argument fails badly.
- Speech is a complex behaviour whose understanding and explanation require, a complex meditational, neurological-genetic theory. Chomsky criticises the simplicity of the model offered by Skinner. Mac Corquodale argued that simplicity is a typical goal in science and that we cannot do know independent of the experimental research whether a particular theory is too simple. Chomsky seems to want to stipulate a priori that what is too simple. And his rhetoric seems to imply that there is no point in trying to test Skinner’s hypothesis empirically. Chomsky also criticises Skinner for not going into the neurological facts that underlies the behaviour. However Skinner argues speculating on internal mediators does not necessarily help with predicting and controlling verbal behaviour. He thinks that there are obviously lawful internal neurological laws but that they may not help predict and control verbal behaviour. I think that Skinner is probably in correct on this but it is and he would have benefited from using connectionist and Bayesian models along with his analysis of verbal behaviour and circumstance of talking. However his approach is still better than a lot of cognitive scientists who almost entirely ignore verbal behaviour and corpus data.
Chomsky also criticised Skinner’s notions MAND TACT INTRAVERBAL and AUTOCLICTIC.
- MAND: Chomsky criticises this notion because he argues that it is impossible to have information concerning the speakers motivational circumstances so the behaviour analyst cannot make the correct diagnosis of whether a response is a mand or not. He also argues that the hearer as reinforcement mediator could not know how or whether to reinforce “relevantly”. To these points Mac Corquodale replies that with careful observation behaviour analyst will be able to determine whatever behaviours determine the behaviour. Just because Chomsky cannot do so when speculating from the armchair it does not follow that it cannot be done in a scientific manner. The behaviour analyst doesn’t need to know the subjects motivation in order to be an effective mand conditioner. Suppose the subject is hungry and says “more bread” if the behaviour analyst provides the bread then he has reinforced manding behaviour of the subject. This occurs independent of whether behavioural analyst knows that the subject is really hungry.
- TACT: Chomsky argues that teachers of first born children could not teach them to tact because they do not have a history of reinforcements. But Mac Corquodale argues that this is not the case because parents have a long history of hearing TACTS from other speakers. One important criticism that Chomsky made that it was obscure how to explain the Tacting of private events like headaches. Skinner replied to this problem as follows:
“that only those internal stimuli that have obvious external correlates which are observable by the reinforcement mediator can become discriminated, so that, as he says, it is the community that teaches one to “know himself”.(ibid p.32)
Obviously those familiar with the philosophical literature on this topic (see Ryle, Wittgenstein, Bennett and Hacker) will know that Skinner’s brief comment isn’t enough to prove he is correct (he discusses the topic in detail in his 1959 pp 272-286). But it does show that Skinner is sensitive to these issues while Chomsky seems to imply that Skinner is simply ignoring the issues.
From my own reading of ‘Verbal Behaviour’ I agree entirely with MacCorquodale’s analysis and criticisms of Chomsky’s review. Chomsky’s review contrary to popular opinion has not refuted Verbal Behaviour. To this is could be replied that in the 55 years since Skinner wrote ‘Verbal Behaviour’ there has been mountains of data collected which refutes any hope of a behaviourist analysis of language. I will discuss this data in the next section.
More recent “Refutations” of the Possibility of Verbal Behaviour explanations
There have been some good arguments for an innate domain specific language faculty however in light of recent empirical evidence I believe they nolonger stand up to critical scrutiny.
Twelve lines of evidence are typically offered in favour of a language faculty and which theorists assume refute any possible explanation of language acquisition:
(1) Poverty of Stimulus Arguments: We have knowledge of grammatical structure x but we have not been exposed to x in our primary linguistic data. Therefore the knowledge must be innate. The primary example they give of a construction that is learned in the absence of experience is auxiliary inversion. Pullum and Scholz show that this example is used over and over in the generative literature. They cite eight different occasions that Chomsky uses the example (Chomsky 1965, 55-56; 1968, 51-52; 1971, 29-33; 1972, 30-33; 1975, 153-154; 1986, 7-8; 1988, 41-47). They also cite other Chomskian thinkers (including linguists such as Lightfoot, 1991, 2-4; Uriagereka, 1998, 9-10; Carstairs-McCarthy, 1999, 4-5; Smith, 1999, 53-54; Lasnik, 2000, 6-9; and psychologists such as Crain, 1991, 602; Macrus, 1993, 80; Pinker, 1994, 40-42, 233-234) who here endorsed the claim. Corpus analysis by Pullum and Scholz (1996, 2002), and by Sampson(2002) show that children are exposed to the relevant construction at least once every 10 days. While the mathematical models of language learning. So, for example, Clark and Eyraud (2007), Perfors et al (2006), Reali and Christiansen (2005) have all developed programmes which can learn from less data than discovered by Pullum, Scholz and Sampson. So I think that this particular instance of a poverty of stimulus argument doesn’t work
(2) The KE family: A family with Specific Language Impairment. Their general intelligence is supposedly unimpaired but they have specific impairments in their grammatical competencies indicating that grammar is not learned by general intelligence but by an innate grammar module.
Recent research by people like Farraneh Vargaha-Khandem have shown that the KE family do not suffer from a language specific impairment at all. They have general intellectual disabilities, and some of their supposed grammatical mistakes resulted from them having difficulties in motor control. (Studies of this case are un going but the evidence so far doesn’t support Pinker et al)
(3) The speed language is acquired: Children acquire language much faster than we would expect them to based on inductive and trial and error learning. Children also acquire their first language much faster than an learns his second language.
This is basically an intuition pump. Chomsky never specifies how fast is so fast that we need to postulate innate competence (Sampson 2004)
(4) There are universal rules of natural language.
There are indeed some linguistic universals, but this doesn’t tell us whether these universals result from people trying to describe similar features of the external world or are innate. Some supposed universals (Recursion) are A not unique to language so don’t support the postulation of a language faculty, and B are not necessarily universal see Dan Everett’s work with the Piraha.
(5) We are the only species who learn natural language, an animal exposed to the same linguistic data as a human doesn’t acquire a language.
This is true but it doesn’t necessarily support the postulation of an innate domain specific language faculty. Our memory and ability to track statistical features in our environment is superior to other animals. Pinker/Chomsky’s argument doesn’t prove whether a innate domain specific language faculty is necessary it just shows that we need some innate architecture which other animals don’t have.
(6) There are critical periods for language learning if a child doesn’t learn language within a certain period then he will find it difficult to acquire language when older.
This argument derived from Lenneberg does not work because. Lenneberg’s data shows all learning pre-puberty is done faster than post puberty so this doesn’t support a domain specific language faculty. (Sampson 2004)
(7) There is supposedly a language gene Foxp2.
Foxp2 despite advertisements is not a grammar gene. There is a lot of work being done on this. But there is good evidence that it is more important for motor control than language. (It is still an open question though a lot of work needs to be done before conclusions can be drawn.) For motor skills and Foxp2 see White et al (2006), Groszer et al (2008), Fisher (2008)
(8) People are not corrected for incorrect grammatical constructs.
This was argued for in a variety of different experimental works. see for example, Marcus 1993, 53-85; Gropen, J., S. Pinker, et al., 1989, 203-57; Crain, S. and M.Nakayama 1987, 113-25. However more recent evidence by Choinard and Clark (2002) shows that children are implicitly corrected for bad grammar and these implicit corrections do indeed make a difference to linguistic performance.
(9) In some cultures children are not spoken to by their parents at all but they still acquire a language. This shows that the language was not taught but grew.
This case is a bit of a myth. Firstly in the case at hand the children are spoken to by older children, they can hear their family speak to each other, and if the children utter something ungrammatical the parent corrects the child.
(10) Children who are brought up by parents who only speak pidgin English manage to construct a grammatically complex language of their own.
This argument is interesting. And does offer some evidence to support Pinker/Chomsky.
(11) Impossibility arguments: It is impossible to learn certain grammatical constructions from the Primary Linguistic Data. The Gold Theorem.
Lappin and Clark (2001) have shown that these impossibility arguments can be over come mathematically.
12) Williams Syndrome: The opposite of SLI. Children with Low IQ typically are linguistically competent. Brock (2000) has done an extensive study of the linguistic competence of people with Williams Syndrome and has found that it is in fact abnormal. So we cannot say that language is necessarily unimpaired despite low IQ in people with this disorder. (Here any claims need to be carefully analysed all the data is not in). People with Williams Syndrome to seem to have linguistic abilities which go beyond their general IQ)
Most of these arguments which are supposed to support a genetically programmed language faculty and by extension show that behavioural or other explanations cannot work in principle. However we have seen that these arguments do not stand up to critical scrutiny. This does not show that Skinner’s Verbal Behaviour is the correct account of language acquisition but it does show that further research is necessary to discover the truth or falsity of the research programme and that Chomsky’s a priori arguments should not be used to argue against this doing this research.
EMPIRICAL RESEARCH INSPIRED BY VERBAL BEHAVIOUR
Michael (1982, 1988), Hall and Sundberg (1987) (, Caroll and Hesse (1987), Yamamoto and Mochizuki (1988) all used a behaviour chain procedure to teach mands to children with intellectual disabilities. Rogers, Warren and Warren (1980) studied manding without using the chain procedure instead they got the children to play with preferred objects and asked the children to mand for the ones they wanted. Simic and Butcher (1980) used two different kinds of foods and trained the subject to say I want a when the analyst entered the room with a tray of food. Savage-Rumbaugh (1984) and Sundeberg (1985) trained non-human subjects to mand.
Sautter and LeBlanc’s (2007) paper showed that between 1992 and 2007 the majority of Verbal Behaviour research focused on two areas (1) Mands (2) Tacts. Furthermore the majority of research in applied verbal behaviour has been with people with intellectual disabilities and/or autism. So Dixon et al, argue that more research needs to be done on people who are developmentally typical, while more research also needs to be done on more complex forms of language.
Their results showed that of the 99 articles they analysed 77percent were done atypical Members of the population. Of that number, 63 of the articles focused on children and 23 used adults. Only 27 percent of the articles investigated typically developing members of the population, and 19 of those examined children and 10 were with just adults. Only four studies (4%) examined both AP and TP in one article. (ibid p. 202).
They conclude that the vast majority of research in this area has been with children with developmental disabilities and/or autism. And while this is important and welcomed data the scope of the research needs to be widened to include a much bigger section of typically developing people if Skinner’s ‘Verbal Behaviour’ is to be championed as an adequate theory of language acquisition. Furthermore they correctly note that we need to go beyond Mands and tact’s and do experimental and empirical research into things like Autoclitics. They note that with recording devices and mountains of internet conversations taking place we are swimming in data and have the technology to record it and analyse the functions of various speech patterns. So if Verbal Behaviour research needs to takes these limitations in the research done and overcome them if the research programme is to remain alive.
I noted above that PECS is an effective way of treating people with Autism and Intellectual disabilities to communicate. However I cautioned that this fact may offer Skinner’s analysis of how developmentally typical people acquire language. Dickson et al (2007) have shown that the majority of research in Verbal Behaviour is done on Children with Intellectual disabilities and children with autism. Further empirical research is required before we should lend our support to Skinner’s Theory of ‘Verbal Behaviour’ but the book should not be dismissed a-priori as Chomsky tried to do.
 In this section I am following Frost and Bondy’s (2002) ‘The Picture Exchange Communication System: Training Manual’.
An interesting but way too short blog on PECS and Chomsky
I went to a conference a while back and one of the guest speakers was a behaviour analyst Dr Vincent Carbone. Excellent speaker, although his description of language may have raised the eyebrows of Noam Chomsky students. As a behaviour analyst he is obviously strongly influenced by B.F. Skinners 1957 book ‘Verbal Behavior’. (Chomsky reviews the book at cogprints.org)
When talking about augmenting communication in ASD students he had some really interesting arguments with regard to using signing rather than PECS (picture exchange communication system) – the basic argument was that signing is must closer to natural communication and quoted the deaf community as an example -i.e. why is there a whole community of signers out there but not picture exchangers? – signing is much more efficient system, more flexible and much closer to spoken language.
Any thoughts or experiences in using PECS and/or signing with ASD students?