Judith A. Dygdon, PhD PSYC 580 Part XV 1 More on Interactions between Classical and Operant Conditioning I foreshadowed this next topic last session when I talked about the feedback that I get from having matched the behavior of a model can come to function as a reinforcer for me if that matching is richly reinforced. In order to make that point, I had to take advantage of a fact we know to be true of interactions between classical and operant condition. Earlier in this session, I cautioned you to “not fight nature”. Well, here is another place in which you need to keep nature in mind. The processes of classical and operant conditioning are natural processes. These processes will work to make new behavior when stimuli are occurring in the prescribed ways. It is critically important for you to remember that these processes will result in new behaviors WHETHER OR NOT ANYBODY DELIBERATELY ARRANGES THESE STIMULI. Let me make this point through an example. Imagine that you are setting a table. You want the table to look nice and be functional. You are not interested in conducting a test of gravity. But gravity is a natural process. It is around all the time, even if you aren’t trying to manipulate or demonstrate it. Now, getting back to that table you are setting – should you accidentally nudge a fork over the edge, it’s going to fall to the floor. It doesn’t matter that you weren’t trying to “apply gravity”. Same thing with classical and operant conditioning. Any time a behavior regularly meets with a desirable consequence, it’s going to get strengthened. Also, any time a relatively neutral stimulus is paired with a desirable or undesirable stimulus, that neutral stimulus will come to take on some of that emotional meaning. Now, any time we have a three term contingency in place SD ----- R ----- S we have a likely neutral stimulus (the SD) paired with an emotionally meaningful consequence (the reinforcement or punishment) that follows. Also, the R (or the feedback I get from having performed the R) might start out as neutral, but it too is being paired with the emotionally meaningful consequence. Judith A. Dygdon, PhD PSYC 580 Part XV 2 It is through this process that “matching to sample” becomes a conditioned reinforcer. Also, a couple of weeks ago, I asked you if you can think of any problems, or problematic fall out from the use of punishment. Does the natural opportunity for classical conditioning trials in operant conditioning events help you see any possible problems resulting from the use of punishment? CHAINS Because the mission of our work so far this semester has been to help you establish a sound basis in the principles established within Learning Theory, most of my examples have been of relatively simple, discrete, behaviors. In other words, when looking for an example of a principle, I’ve focused the “scope” on one behavior. Sometimes, the R on which we’ve focused the scope was not truly a single behavior but a “bunch of little behaviors”. The scope will help us understand those kinds of behaviors; It just needs to focused, sequentially, on each component. Here enters the concept of a behavioral chain. At this point, I invite you to watch the first video at https://youtu.be/nWU0bfo-bSY After you’ve watched that video, please come back here. Focusing the scope on the squirrel, we could describe the squirrel’s behavior as …. (1) SD ----- Base of Pole R run ----- SPR peanuts Judith A. Dygdon, PhD PSYC 580 Part XV …. but that would not capture the richness of this behavior, nor would it capture what we have come to learn happens to the environmental components involved. This squirrel’s behavior in (1) give you a good picture of a behavioral chain (alternatively, chain of behavior). In a chain, we see many small behaviors, each one surrounded by antecedent and consequent stimuli. A chain is pictured as follows: (2) SD- R -SPR | SD- R -SPR | SD- R -SPR | SD- R -SPR | SD- R -SPR | SD- R -SPR | SD- R -SPR | SD- R -SPR | SD- R -SPR 3 Judith A. Dygdon, PhD PSYC 580 Part XV 4 Understand that diagram in (2) this way: The first SD- R -SPR captures the first behavior you see (in the squirrel caper, that would be base of pole --- run up pole --- arrive at “diving board”). The final SD- R -SPR (and there is no theoretical limit to the number of steps in a chain) captures the ultimate reinforcer (in the squirrel caper, that would be board at end of tube --- leap to bowl --- peanuts.) I’ve drawn a vertical bar between many SPRs and SDs to communicate “having attained this reinforcer, signals the opportunity for the next behavior in the chain.” For a behavioral chain to be effective, these SPR / SD relationships must become established. They can be deliberately taught (e.g., by a squirrel trainer) or happened upon through trial and error (e.g., by an exploring squirrel.) You can build a chain of behavior by starting at either end. If you start at the initial “link”, you will need to artificially introduce a reinforcer (as “arriving at diving board” probably has no inherent value to most squirrels.) If you do this, we call the process “forward chaining”. Alternatively, you can build a chain by starting at the terminal link. Here, the important reinforcer “naturally” occurs, but you will have to “restrict” the environment. Expressed in terms of the squirrel example, that would mean: Place the squirrel on the board at the end of the tube (SD) (with the peanuts in view). When he leaps (R), he’ll get the peanuts (SPR). This will work because, as you learned last week, classical conditioning trials are naturally embedded in operant conditioning trials. Thus, the contingent appearance of “board at end of tube” (CS) and “peanut” (UCS) will eventually establish “board at end of tube” as a conditional reinforcer, ready to strengthen the penultimate behavior in the chain. Judith A. Dygdon, PhD PSYC 580 Part XV 5 I mention chains because we recognize them, and seek to build them, in many clinical contexts. Our mission at this moment is not to talk about therapy, but let me give you a coming attraction. Problem-solving skills training is something to which we will appeal in many clinical applications. When we teach problem-solving skills, we teach a client to follow a sequence of steps in order to move from the identification of an aversive situation, through the generation of possible response options, to the application of a response option with good promise of success. (You’ve probably seen this. Often, we’ll give these steps names, like: Identify the problem, Determine the goal, Generate alternatives, Evaluate alternatives and choose, Plan for Action; etc.) What I’d like you to see is that when we teach problem-solving skills, we are teaching a behavioral chain. Note that “problem-solving” is a private, cognitive behavior. Nevertheless, we believe it to be a private, cognitive, behavioral chain. Judith A. Dygdon, PhD PSYC 580 Part XIV 1 BEFORE WE TALK ABOUT MODES OF CONDITIONING, I NEED TO TELL YOU ABOUT TYPES OF THEORIES Our focus in this class is Learning Theory. As I said at the opening of our class, Learning Theory is not the only theoretical constructional we have in clinical psychology. Our various theories can be categorized into two types. Learning Theory is best classified (using labels introduced by Catania) as a functional theory. Functional theories lay out rules for the experiential events that will cause behaviors to be acquired or changed. Being a good example of a functional theory, learning theory lays out the rules for experiential events that will cause behaviors (both elicited and emitted) to be acquired or changed. In case you haven’t already noticed this, please see now that learning theory rules tell you how behaviors are acquired and change, but it does not tell what are the important behaviors to be acquired and changed. There are other theories with a “flip-side” focus. Catania would call these structural theories. Structural theories do not take on the task of laying out generally how behaviors are acquired and changed. Rather, they address the question, “What are the important behaviors for human beings to have?” Learning theory is a good example of a functional theory. There are many structural theories. One that comes to my mind right now as I am looking for an example to give you is Attachment Theory. Attachment theory outlines the importance, for human beings, of establishing secure attachments and the problems that follow from variants of secure attachments. What is important for you to recognize is that functional and structural theories are not at odds with each other. They simply address different missions. Many learning theory oriented therapists follow their clients’ leads (or information from the client’s family or cultural group) in determining what are the important behaviors for the client to acquire or change. Others will look for direction from any of a number of structural theories. Some will do a little bit of both. MODES OF CONDITIONING (This is a really good time to reread the Dygdon, Conger, and Strahan, 2004 article I assigned a couple of weeks ago.) Judith A. Dygdon, PhD PSYC 580 Part XIV 2 In the previous section of my notes, I started to talk about some concepts that help us use operant conditioning with complex human behavior. I want to continue along that path. Direct Mode So far, in my talk about classical conditioning and in my talk about operant conditioning, I have used examples that fit a direct mode of conditioning. What I mean by that is that I’ve given examples of things people experience (e.g., Bob was bitten by a dog) or things people do (“I write a lecture) in real life, with real contact with those experiences or with the consequences of their behavior. We can certainly learn new elicited and new emitted behaviors through direct experience. However, direct experience is not the only way in which we can experience classical and operant conditioning trials. There are two other modes through which we can learn. Perhaps you are already guessing what they are. We will use all the same rules of classical and operant conditioning to organize and predict learning through these other modes, but they won’t involve directly experiencing things in the environment. Verbal Mode Humans (and some other organisms) are capable of using symbols to represent the real world. What is our biggest set of symbols? Language So, in addition to learning through the DIRECT mode, one other mode through which we learn is: VERBAL Judith A. Dygdon, PhD PSYC 580 Part XIV 3 OK. Some comments on how to think about language in human behavior: Language is a set of symbols, used to represent things in the environment, behaviors of people, relationships among things, relationships among behaviors, relationships between things and behaviors. Remember, learning respects biology -Humans have an innate capacity for learning to use language • We don’t learn the capacity (that’s innate) • We do learn how to use the capacity The way in which we learn to use language, or more generally, to communicate symbolically, is called Stimulus Equivalence Learning We say that a person has developed a set of equivalent stimuli if, for a set of independent stimuli, the subject can use the stimuli in 1. reflexivity (A = A) 2. symmetry (if A = B, then B = A) 3. and transitivity (if A = B, and B = C, then A = C) relations and • after some of those relations are explicitly trained, some others naturally emerge. (When you think about it, this is another one of those “Wow” moments.) You see stimulus equivalence learning all around you. Imagine language learning for a child: • teach the child to match your spoken word “dog” (A) with his spoken word “dog” (A) (This is teaching a reflexivity relation; that is, this Judith A. Dygdon, PhD PSYC 580 Part XIV 4 spoken word “dog” means the same things as that spoken word “dog”, etc.) • teach the child to match your spoken word “dog” (A), with a photo of a “dog” (B) (This is teaching a symmetry relation.) • teach the child to match a photo of a dog (B) with the written word “dog” (C) (This is teaching a symmetry relation.) Following this: Other reflexivity uses are likely to emerge. • Child will understand “dog” (A) from another, novel, speaker to mean the same word “dog” (A) he has heard before Transitivity uses are likely to emerge. • Child will be able to match spoken word “dog” (A) with written word “dog” (C) even if that relation was not explicitly trained !!!!!!! NOW … why did I ask you to read that 1991 article by Hayes? First thing: I realize Hayes’ article is dense and could justify weeks of focus on its own. I will not ask you (on an exam) anything about Hayes’ article that I do not cover here. I will not ask you to define Hayes’ “mutual entailment” or “combinatorial entailment”. At this stage of your development in learning theory, you first must master the terms of “reflexivity”, “symmetry” and “transitivity” used by the early workers (like Sidman, as cited by Hayes, 1991) in stimulus equivalence. (I am hoping that you can see “reflexivity”, “symmetry” and “transitivity” embedded in Hayes’ terms, but I won’t test you on that.) What is important for you to know is that this article by Hayes opened an important new path in learning theory based clinical work, thus I feel it is important for you to “taste” it. Specifically, note that Hayes 1991 is elaborating on stimulus equivalence learning and acknowledges its application to how human beings learn language. His big point in this paper is the introduction of “Crel”. That is, he argues that, in order to use language effectively, the stimulus configuration of a particular context signals whether an acquired stimulus equivalency is in effect. What exactly does this mean? Judith A. Dygdon, PhD PSYC 580 Part XIV 5 It means that human beings who have the ability (to use context to signal whether an equivalency is in effect) can determine when a verbal utterance is veridical versus when that same utterance is offered as a joke, solely for affiliative purposes, or offered by a speaker who is not trustworthy. Hayes will go on to call this perspective “Relational Control Theory” and argues that this behavior (the ability to use contextual cues to determine if words should be trusted as “truthful” ) is a learned behavior. “Relational Control Theory” went on to become the basis of Acceptance and Commitment Therapy (ACT). ACT is designed to help people who do not possess this critical skill; that is, those who have not learned to restrict the application of language based self-statements. Because ACT developed from within the learning theory tradition, and uses learning theory as its functional base, you will learn about it in PSYC 641B. Note that, though it uses learning theory as its functional base, Relational Frame Theory gets a little structural in that it says that the ability to use contextual cues to determine if words should be trusted as “truthful” is an important behavior to learn. --------------------------------------------------Back to the Verbal Mode of Conditioning: Our ability to use words as symbols for things in the physical world allows us to learn new elicited behaviors through verbally represented CS/UCS pairings and new emitted behaviors through verbal S-R-S trials. Judith A. Dygdon, PhD PSYC 580 Part XIV 6 Rules and Rule Governed Behavior We have a special name for emitted behaviors taught through verbal representations of the consequence they will attain. We call those verbal representations “rules” and behaviors that follow such rules, “rule governed behaviors”. • We can deliver SD – R – SPR, SD – R – SNR, SD – R – SPP, SD – R – SNP, trials through words ad this can save some learning time. Isn’t communicating with rules this just giving advice? • Yes and No. • We’ve learned that some rules are better than others • Or some advice is better than others at building new behaviors. Good rules do a better job of capturing specific SD ----- R ------ SPR information. Some Comments on Skinner’s Talk about Language Since we are talking about language, let me interject some important terminology. Skinner, though many don’t realize this, was very interested in how learning, specfically operant conditioning, was manifest in human behavior. He did a lot of writing about human verbal behavior and how it is acquired. Learning believes that we don’t learn the capacity for language, though our “verbal community” does shape how we use that capacity (the fact that there are several languages in the human verbal community is simple evidence of that). Because Skinner was all about function (i.e., Why is the behavior there? What reinforcers does it access?) he came up with categories for verbal behavior Judith A. Dygdon, PhD PSYC 580 Part XIV 7 based on their functions. You will see labels used occasionally now. I expect you will see them used more often as we more further forward in studying human behavior. Here are his labels: TACT – identifies a verbal behavior that names a behavior, a stimulus in the environment or inside an organism, or the interactions among behaviors, among stimuli, and among behaviors and stimuli. A “rule” as discussed above, is a tact. Though it may sound to you like Skinner has just developed another name for “noun” that is not the case. Skinner is not labeling parts of speach. Note that the definition reads “... a verbal behavior...”, not “a word”. When I say things like “I am typing”, “table”, “she is reading”, “they are standing on the corner and waiting for a bus”, I am tacting. MAND – identifies a verbal behavior that “tacts it own reinforcer”. In everyday language, when “I ask/tell you to do something” we often call this a “command”. The label “mand” comes from “command”. When I say things like “Bring me a glass of water”, “Come here”, “Stay away”, “Read these notes in preparation for the exam”, I am manding. AUTOCLITIC – these are behaviors that emphasize or qualify tacts or mands. No, they are not exactly adjectives and adverbs. Those labels apply to single words. “Autoclitic” applies to behaviors, which might be single words, or might be much more, that qualify, direct, emphasize, etc. verbal behaviors. For example, when I say “stay away” I am issuing a simple mand. When I say “You, stay away”. I am also manding, but I am restricting the focus of the mand. Used in this way, “you” counts as an autoclitic. Note that I could accomplish the same thing by pointing to you, while I say “stay away”. In this case, the pointing would also count as an autoclitic (even though it, itself is not verbal) because it qualifies my verbal behavior. Judith A. Dygdon, PhD PSYC 580 Part XIV Skinner says that, independent of specific language, we learn to use verbal behavior to tact and mand and learning autoclitics help those work better. Skinner argued that, despite a capacity to develop and use language, humans need to have verbal communities to demonstrate language use and to reinforce its appropriate production and application. 8 Judith A. Dygdon, PhD PSYC 580 Part XIV 9 Observational Mode So, we can learn behaviors through DIRECT and VERBAL classical and operant conditioning trials. Is that it? No, one more We learn new behaviors through • Through the DIRECT mode • Through the VERBAL mode and • Through the OBSERVATIONAL mode We say that learning through observation is the imitation of the response/behavior of a model without previous reinforcement of that specific response. The idea of learning through observation sounds simple. But in fact, it is a bit complicated. Bandura, in his social learning theory, wrote a lot about observational learning. His writing is outside the realm of learning theory because of his heavy use of hypothetical cognitive processes, but his beliefs about the importance of learning through observation are shared by the learning theory community. Bandura’s social learning theory point is that humans learn to engage in behaviors that they see others do and be reinforced for. There are bunches of data to support that this takes place and that other tenets of Bandura’s perspective are true (e.g., these behaviors are more likely to be imitated the more like the learner the model is). Judith A. Dygdon, PhD PSYC 580 Part XIV 10 I hope that you can see that what Bandura is saying is that humans can learn when they see another person experience SD ----- R ----- SR Learning acknowledges that this kind of observational learning takes places and that it extends to classical conditioning where I might learn an elicited behavior (like fear) because I have seen someone else experience UCS ----- UCR | | CS In addition to the observational form of operant learning that Bandura discussed, learning theory research has identified another mechanism through which observational learning can occur – and this form doesn’t require that the person doing the modeling be reinforced. This line of research is generally referred to as work on “matching to sample”. You see things like the following all the time: The Somename family is eating dinner together. They’ve just finished and dad leans forward, puts his arms on the table, and crosses them in front of his chest. Three year old Billy leans forward, puts his arms on the table, and crosses them in front of his chest. Mom smiles at Billy, Dad ruffles Billy’s hair, and both parents speak in pleasant tones about “Look how Billy wants to be like Daddy!” What do you think Billy learned in that interchange? Perhaps Billy accumulated a little bit of information that will help him to lean forward and cross his arms after dinner, but we don’t think that that’s the Judith A. Dygdon, PhD PSYC 580 Part XIV 11 most important thing he learned. We believe that Billy learned something more (and that humans are innately ready to teach their young this next thing.) In interchanges like this, we believe that Billy learned “Match what daddy does (people get excited)”. We believe that through lots of experiences like that dinner caper, Billy (and all other human children) learns to “match to sample”. Here is how we think learning to match to sample works. Experiences like the dinner caper build a new reinforcer. We believe that humans are capable of reflecting upon or “seeing” their own behavior to the extent that I can more or less tell if my behavior matches a standard. Given this, if I am richly reinforced for “copying” others, then we should assume that I have learned a conditioned reinforcer that naturally occurs when I produce a match. We believe that our ability to learn through matching to sample helps us acquire behaviors at a very fast rate and helps us acquire more behaviors than we ever could through direct experience alone. However, being taught to “go out there and match” can have some problematic fall out. Can you think of any? By the way, we think that we probably need to learn through observation before we can learn language. When a child imitates words, this action produces “match-to-sample” reinforcers just as many other imitative behaviors in the past produced “match-to-sample” reinforcers. One last comment about “matching to sample” -- I have just described the argument that says for humans, through experience, “matching a sample” becomes established as a conditioned reinforcer. There is another side to this argument. Some believe that “matching” is an innate human behavior. Either way, it helps us learn lots, and fast. The jury is still out on this one, so I can ask you your opinion. What do you think: Is imitation a learned or innate behavior?
Purchase answer to see full attachment