Mands in non-human animals
(this post is in English because I’m hoping for some input also from my English-speaking friends 🙂 )
Right now I am visiting the University of North Texas and tonight I had the opportunity to attend to a lecture on teaching verbal behavior to autistic children – which sets the background for this blog post. It’s rediculously long so here’s the table of contents:
- What is a MAND?
- MANDS in animal training
- Learning from MAND work with children with autism
- The difference between MANDS and other behaviors
- Further elaborations
What is a MAND?
In the field of Behavior Analysis, verbal behavior is a large and important area. For those interested in how behavior analysis deals with verbal behavior I recommend for example Catania’s book “Learning” from 2007. In short, what’s important for my purpose here is:
- Verbal behavior is per definition is reinforced by or through the behavior of another individual.
- Verbal behavior can be vocal but it doesn’t have to be. For example, writing or sign language is also verbal behavior.
- Verbal behavior is categorized into various categories such as mands, tacts, intraverbals and some more. The categorization has to do with the A-B-C contingency, for example whether the A (the “cue”/discriminative stimulus) is verbal or non-verbal and whether the C (the reinforcer) is social or non-social.
Discussions about verbal behavior is for various reasons normally restricted to humans, but I do believe that some of the concepts can be very valuable for us working with non-human animals. One example of this is MANDS – both the concept as such, and how mands can be taught.
A mand is the most basic type of verbal behavior. When manding, the behavior specifies the reinforcer – in other word, what reinforces the mand is that you get what you ask for. (Think “demand”…) For example, if a toddler sees an apple and wants it, he can point at it and say “apple” to get his parent to give it to him. (As a contrast, if the child points at an apple and says “apple” not in order to get the apple but in order to get attention and social approval, that’s not manding since the reinforce isn’t specified by the behavior. Instead, it’s an example of a tact).
With autistic children, mand training helps the child getting his “wants and needs” met. Children who are not capable of manding often revert to “problem behaviors” such as tantrums in order to get what it is they need or want. Think of the newborn baby: Crying is manding! The baby wants or needs something that he’s not currently getting (such as food, a fresh diaper, or company). He cries, and his parents come to his assistant by more or less quickly figuring out what’s up. Eventually, in the interaction between the baby and his parents, the baby’s behavior and the parents’ responses will become differentiated so that the baby more clearly can communicate precisely what it is that he wants.
So, looking at the A-B-C contingency of a mand:
- A (antecedent, what happens before the behavior and sets the occations for it): a state of deprivation, or a state of aversive stimulation (in plain English: Wanting something, or wanting to get rid of something).
- B (behavior): the mand.
- C (consequence, what happens after the behavior and reinforces it): Whatever is specified by the mand (in plain English: Getting or getting rid of what you wanted to get or get rid of)
MANDS in animal training
While listening to the lecture today I started thinking about mand-like behavior contincencies in animals and in animal training. My first example is Tizla’s “can I have some food, please” behavior: A)I sit at the table with food and Tiz is hungry (“slightly food deprived”), B)Tiz lies down on the floor, C)I (eventually) give her food. This nicely fits the A-B-C contingency of a mand according to the bullet list above.
Other examples from our daily life, where the behavior specifies the reinforcer, is:
- Tiz needing to relieve herself (“aversive state” of full bladder) but we’re indoors, B) Tiz puts her chin on my leg – or on my face if I’m sleeping…, C) I let her out so that she can do her business.
- Tiz is thirsty and finds the water bowl empty, B) Tiz paws and digs in the water bowl, C) I fill it the water.
- Tiz is hungry and I’m eating at the table, B)Tiz lies down on the floor, C) I (eventually) give her food.
You get the picture 🙂 Everyone living with an animal probably has their own example of behaviors like this – behaviors that the animal uses to tell you what he wants so that you can help him get it.
With our animals, this type of behaviors are really good because they helps us know what the animal wants and needs, thus making it possible for us to attend to those needs. This same type of behavior can also be really annoying – especially if it’s inadvertently taught… Many “problem behaviors” can actually be seen as mands! We see this all the time when the animal wants something, does a certain behavior, and through this behavior gets the human to do what he wants. Barking or whining for attention is only one example; A) “Attention deprived dog”, B) dog barks or whines, C) dog gets attention from the human.
I suggest that there can be several advantages to recognizing MANDS as a specific type of behavior in the animal-human interaction:
1) it can be easier for us to teach desired behaviors and avoid teaching undesired ones,
2) our capability to meet the animals’ needs can increase
3) the animals’ control over their environment can increase
4) animal-human communication and relationship can expand.
(these are just the advantages I think of from the top of my head, I’m sure there can be more!)
Learning from MAND work with children with autism
With autistic children, teaching mands is really important. We want children to be able to ask for what they want, and to do so in appropriate ways. So if there is a “problem behavior” occurring because the child wants something but cannot ask for it properly – or wants something but cannot have it – it can be dealt with within the teaching of mands. As Einar Ingvarsson said in today’s lecture: “Problem behavior is a form of communication”. So with children, behavior analysts work with Functional Communication Training (FCT) which basically is teaching mands.
I have a feeling that with our animals, often the first choice to deal with this kind of “problem behavior” is to extinguishing it – in other words, to not give the animal whatever it is that he is “asking” for. But that way we actually take away the animal’s capability to use mands for communication 😦
By looking at how behavior analysts work with mands with children with autism, I think we can find creative ways to deal with problem behaviors and build new levels of communication with our animals. Of course many people (probably most people!) already do this – but I think it’s interesting to view it from the perspective of the mand 🙂
One example is when we teach the animal to “ask” for whatever it is that he wants by offering a more appropriate behavior. Example 1) When Tiz was a puppy she’d jump at the person preparing food. However, she quickly learned that jumping lead to nothing while backing off (eventually shaped to lying down, and for longer and longer periods) paid off by her getting pieces of the desired food. Example 2) Tiz has a tendency to whine when she needs to go out. I hate the whining, but I very much want her to let me know when she wants to go out. So what I did was to not react to her whining by the door, and just wait. Eventually she’d come to see where I was, and then I cued her “chin” and reinforced the “chin” behavior by immediately saying “Let’s go out” and hurry to the door with her. Gradually I could fade the “Chin” cue, and now she just comes and presses her chin to me.
But what about if I already know what it is my animal wants – I just don’t intend to give it to him!? Well, that’s the same as the child asking for candy for breakfast. Thus, this is an issue that is addressed by everybody teaching mands to children. The answer is: Stimulus control. The learner (child or animal) just needs to learn under which conditions the mand will work. Child example: Asking for candy only works on Saturdays (or, only when the “candy sign” is visible on the kitchen table, or whatever…). Dog example: I could decide that Tiz’ asking me to go out by putting her chin on me only works if at least an hour has passed since the last time she asked. Or, as exemplified by Karen Pryor in Don’t Shoot The Dog: Barking outside the door in order to be let in only works if the white sign is hanging on the door.
Here’s a brief description of the mand training procedure described in today’s lecture:
START AT: What are the behaviors currently functioning as mands? In other words, what is the individual doing in order to get what it is that he wants?
THE TEACHER’S JOB is to teach more appropriate forms and/or more elaborate forms of mands.
FIRST ASSESS the current mands: *the reinforcers, *the form/topography of the mand, *the match of form/topography x motivational state x reinforcement contingency
SELECT mand forms to teach. Consider the audience (who is going to respond to the mand?) and the portability (can the mand work in different contexts?).
LOOK AT RESPONSE CLASSES AND RESPONSE COMPETITION, contsidering *response effort, *schedule of reinfordcement, *latency, *quality and magnitude of reinforcement, *probability of reinforcement. (Especially note the last one! If the mand behavior you risks being punished, then not much point in teaching it…)
HOW TO TEACH: Either just capture (or shape) the mand when the behavior (or something that can be shaped into it) happens to occur, or set up “contriving EOs” (establishing operations, things that’ll make the reinforce more reinforcing) by setting up a situation where the animal is likely to want whatever it is you plan to teach a mand for (For example, empty the water bowl if you want to teach your dog to ask for water. Put some food on the table if you want to teach your dog to ask for food. Play with a stick if you want to teach your dog to ask you to throw it. Et c.).
Note that you can strategically teach manding to be resilient (so that the dog keeps asking until he gets what he want) just like you teach resilience in any behavior. For example, variably but gradually increase the number of mands required – that builds resistance! You can also bring mands under stimulus control, so that he mand only “works” in certain situations. You can teach multiple mand forms if you like, so that the animal has several different ways to ask for the same thing. With the child, this might be “I want candy”, “Can I have candy please”, et c. Tizla asks to be let out either by offering her chin behavior or by standing by the door looking at it. When multiple mands are accessible, the individual will switch to another one if the first one isn’t “working”. This is an example of resurgence; a previously learnd behavior occurs when a certain behavior no longer leads to reinforcement.
The difference between MANDS and other behaviors
I’d say Tiz sitting when I say “sit” is not a mand. Why? Because her sitting doesn’t specify the reinforcer. Sitting sometimes leads to treats, sometimes to toys, sometimes to cues to other behaviors – and all these reinforcers function to strengthen her sitting behavior. Also, the behavior should be occasioned by a state of deprivation or aversion. As animal trainers we usually try to teach behaviors so that they are more or less independent of “motivation” – I want Tiz to sit when I ask her to even if she isn’t hungry… So I really don’t want her sitting to be a mand for food.
However, what is and isn’t a mand is definitely a fine line! It’s quite possible to argue that the rat pressing the lever in the laboratory is manding food. This is where the behavior analysis definition of verbal language sets boundaries, requiring that the verbal behavior is established in a verbal community (which more or less excludes non-human animals). So I’m really not that interested in where to draw the line for the definition of mands; these definitions have been elaborated in detail by behavior analysts ever since Skinner published his book Verbal Behavior.
The meaningfulness I see in the concept of mands when it comes to non-human animals has to do with the very core of the mand concept: an antecedent “state” correlated with a specific reinforcer, and a behavior that the animal can use to get another individual (the human) to provide that reinforcer.
I think we can have lots of fun with teaching mands. A friend of mine, Nina Mortensen, once mentioned teaching a target behavior directed at various symbols as mands (even though we didn’t have that term in our vocabulary then, it was six years ago…) for various events – one symbol for “go out”, another for “give me food”, et cetera.
Another type of mand that could be interesting is what’s called a rejection mand. That is, the possibility to say “No thank you”.
Also, something I definitely plan to look into further some other time (since it’s over midnight and since this post is getting more than painfully long) is how dogs use mands in their natural language. I’d say that’s what animal communication is about, most of the time… Growl = “leave me/my stuff alone”, reinforced by the other guy backing off. Play bow = “wanna play?”, reinforced by play invite from the other guy. Sniffing, gasping and other calming signals = “slow down/leave me alone”, reinforced by the other guy doing just that. Et cetera. Mind, though, that if it’s a mand the function is to get the specified reinforce – so animal behavior that is reflexive (driven by antecedent stimuli, not by consequences) do not fit the definition. Hmmm much to think about here 🙂
So…. I absolutely intend to think further about this subject, because I find it totally fascinating. And if anyone has had the stamina to read all of this, I look forward to any input!
130920: Please take a look at this wonderful video by Stephanie Edlund – great example of manding 🙂 www.youtube.com/watch?v=pZ945XyMpO4&feature=youtu.be