Francesca degli Espinosa
Why I want Darwin to counter surf
Navigating the world of dog training is as overwhelming as navigating the world of human educational intervention. I soon discovered that not unlike its human counterpart, there are different schools of thought, a myriad of techniques, and differing ethical viewpoints. Unlike human intervention, however, the theoretical paradigm from which canine interventions are derived is behavioural. In other words, altering, establishing or reducing a dog’s behaviour and its emotional correlates, requires the understanding of behavioural principles. The same could be said with respect to human intervention, but that’s a discussion for a different context.
A few days before Darwin’s arrival, all the dog paraphernalia arrived (crate, bed, toys, food…). For something so small, his equipment seemed to be occupying a vast amount of space. By the time I was ready to pick him up I had binge-watched as many puppy-training videos on youtube as I could. I had enrolled in a popular online puppy course, signed him up for puppy socialisation classes and enlisted the help of a trainer to teach me the basics. I was ready, or so I thought.
Having raised two of my three babies (now young adults) on a schedule of manipulating motivational antecedents for food, sleep and attention to prevent common infant problem behaviour (night waking, food refusal, excessive vocalizations, demand crying) it seemed logical to apply the same principles to my non-human learner. After all, behaviour is behaviour, and given that I was the agent in control of most of my learner’s reinforcers, it seemed reasonable to employ the same approach. Darwin slept through the night in his crate by the end of week 1 without the application of extinction procedures and by week 6 was house trained with no more than 10 total accidents. My problems lied elsewhere.
Labradors are known for their docile temperament and for being highly trainable. It is not a coincidence that they are regularly employed as service and therapy dogs. They are often described as being very food motivated. In other words, food is of extremely high value and consequently can be used as an effective reinforcer to establish target behaviours. I would watch basic training videos of these little puppies excitedly wagging their tail at the sight of treats or their bowl, eager to respond to verbal or gestural requests, making eye contact, sitting and salivating at the prospect of the incoming treat. When I went out, dogs would come charging across the field having caught a sniff of the delicious roast chicken, cheese, steak, and salmon I carried in my treat bag. But not Darwin. Darwin never jumped up at strangers who carried a treat pouch; in fact he never jumped at me for food or counter surfed.
At puppy classes he would do what the instructor showed once, maybe twice, then run off to meet the other puppies or walk away in pursuit of interesting smells. I struggled to get a head turn, a bit of eye-contact. He perked up a bit more with toys; he would tug for a few seconds only to drop the toy and walk off. When I tried doing short 5-minute sessions in less distracting environments, it was no different. When he couldn’t walk away or sniff the ground, he would simply scratch his ear or accompany the target behaviour with a little whine.
As a clinician, this problem was not novel to me, I also knew that it was one of the hardest to solve. The child who screams and shouts, tries to grab things off you, or scratches you to get what you are holding is letting you know those items are valuable to him. The child who, despite what you bring out of your Mary Poppins’ bag, walks away or engages in self-reinforcing behaviours, will test your behaviour analytic skills at all levels of the 4-term contingency. Darwin’s behaviour reminded me of the child who, the minute a whiff of an instructional contingency was in the air, for example if the “reinforcer” was placed on the “work” table, would walk away, only to grab the item when my back was turned.
By 6 months of age, Darwin would give me occasional moments of participation and I would lavishly deliver the best treats I had. By the time he hit adolescence, everything outside the walls of our home was far more interesting than anything I had to offer. I kept upping the ante. If he turned away from me, I would run like a maniac making silly noises. I was trying to be more interesting than anything else in the environment, as recommended by many experts. I was trying to be “sexier than the distraction”. I read all I could, purchased monthly memberships to various dog training websites, listened to podcasts. Most of the techniques and strategies seemed reasonable and they all shared the same underlying commitment to positive reinforcement. I tried everything and was getting more and more demotivated by Darwin’s lack of excitement and participation.
I realised I was doing Darwin a disservice by ignoring my own training and attributing all the responsibility to him, by justifying my teaching deficits with “he’s just not food motivated”. I did not need more techniques, I did not need to purchase more exotic cuts of meat, amazing toys, online courses or one to one consultations. I needed a functional comprehensive behavioural approach. I needed a behaviour analysis that could not be reduced to “ditch the bowl”, be more interesting than the female bitch in heat running off-lead across the field or use stinky (and frankly disgusting) raw tripe. I wasn’t a dog trainer, I wasn’t a dog behaviourist, my mechanical skills were terrible, but the principles of behaviour are universal, and those I knew reasonably well. I also knew that anytime two individuals are involved (in this case Darwin and me) learning is occurring on both parts. The first behaviour I had to look at was not his, but mine.
As a clinician I was taught to look at all four parts of the contingency when presented with a problem. Sometimes it will be sufficient to alter just one aspect, but in complex cases a comprehensive analysis and manipulations of all parts may be necessary. In this post I will focus on the antecedent manipulations.
A motivating operation is an antecedent event that alters the value of a stimulus as a reinforcer or punisher. For example, sitting at my desk for 4 consecutive hours establishes the value of movement as a reinforcer. Conversely, excessive movement (e.g., a 10k run) abolishes the value of movement as a reinforcer and establishes immobility (e.g., collapsing on the sofa) as a reinforcer. So, an MO is a momentary environmental change, either internal or external, that is linked to how valuable something is, there and then, and therefore to its effectiveness as a reinforcer. Skinner talked initially about “drive” to then replace it with the concept of deprivation and satiation. Keller and Shoenfeld first introduced the term establishing operation (EO). It was Jack Michael, however, who provided a comprehensive description of the phenomenon, and the generic term Motivational Operation (MO) to cover both EO and Aboloshing Operation (AO) effects.
Motivating operations can be conditioned or unconditioned. Unconditioned MOs (UMOs) are essentially linked to primary or unconditioned reinforcers. Conditioned MOs (CMOs) refer to events that alter the value of stimuli through learning. For example, If I want to go outside but the door is locked, the locked door alters the value of the key as a reinforcer. The key is not always of value – it is only important when I want to go outside and the door is locked. In other words, if you asked me to do some work and paid me with a key, I would be less inclined to do it than if you asked me to do the same work to gain access to the key when I wanted to go outside and the door was locked. The locked door is an example of a transitive CMO. An MO does not make a reinforcer more probable; it simply makes it more valuable. The Sd is the signal that under MO conditions, Sr is available.
In Darwin’s case, it would seem logical that to increase the value of food and turn it into an effective reinforcer, I would simply need to create an EO by manipulating deprivation and make food available only under certain conditions. I learned that in the dog training world this is sometimes referred to as “ditching the bowl”; more technically it is referred to as a “closed economy”. I felt that that was only half of the story. I had to consider an additional MO, my lifelong enemy: the CMO-R. My attempts to engage him under conditions in which competing reinforcers were also present (e.g., the grass, other dogs, interesting smells) functioned as a set of conditions correlated with a worsening set of events (removal of reinforcers, response effort, reinforcer delay). Hence, establishing their termination as a reinforcer, in other words my attempts to interact with him under those conditions, over time, made escape more valuable. It wasn’t just a case of getting better food, being more interesting, increasing the state of deprivation… I had to increase the value of food in my presence and in the presence of all the stimuli that typically signalled some behaviour was expected.
Most behaviour is multiply-controlled, meaning there are multiple converging sources of control that make behaviour more or less probable in certain contexts. Consequently, reinforcers are also often synthesized. When I used to call my children to set the table because dinner was ready when they were little, my request used to signal the availability of multiple reinforcers: my attention, the task itself (doing a grown-up thing) as well as food delivery. Their latency used to be very short. Fast track 10 years, to my screen-worshipper teenagers. Rarely now does my request function as an Sd for the target behaviour. More often than not, it is correlated with the interruption of screen time, and if they are at the stage of passing a level, I am lucky to get the polite acknowledgement of “in five minutes.” In this case, my request alters the value of escape from the task to maintain reinforcement access: a synthesized negative and positive contingency. So, the behaviour of continuing to sit in front of a screen or saying “later” is maintained both by escaping the task itself (in the interest of time and to avoid an argument I often end up doing it myself) and continued access to the screen. It is also important to say that if food was only ever accessible under the conditions of screen time interruption and setting the table, I would probably get a different response. This would require a pretty extreme rearrangement of the household and blocking access to the kitchen. I would have to make sure that food was only available contingent on the behaviour I wished to select and not at other times for other behaviour. Clearly, too much effort on my part...
UMO manipulation – increase the value of food determined by changes in eating behaviour.
First, I discounted medical issues, and removed chicken from his diet. Darwin had chronic soft stools from the time I got him, two under the skin growths that had to be removed between 4 and 7 months of age (likely due to an allergic reaction), and never ate in the mornings (which at times led to vomiting). So, it wasn’t just a case of low eating behaviour during my attempts to train. In general, when presented with a bowl, he would eat a bit, then walk away, drink, sniff around, then maybe go back to it. When I removed the bowl after 15 minutes, it was often still half full. Of course, I then tried to compensate by offering food throughout the day. He went from kibble to a raw diet to cooked food back to raw and, now finally, back to kibble and the occasional raw bone in his crate or bed.
Food was delivered 4 times a day in small meals through a game of two bowls which I had learned from Hannah Branigan and Chirag Patel. As soon as a bit of food was consumed in one bowl, a second bowl was presented behind him, as he was consuming that, food was dropped in the other bowl. This process continued until he was ping-ponging rapidly between the two bowls without delay in between the trials. Gradually the distance between the two bowls was increased, until he was running between the two. As this behaviour became more fluent (the end of the bowl evoked turning to the next without wandering around or delay), competing EOs were gradually introduced. For example, we transitioned to the garden.
CMO-R manipulation: reduce the aversive value of the teaching setting.
During the first couple of days, I stopped giving any instruction, I did not intentionally ask, shape or lure anything. When we were out, he would be on a long line and be free to sniff, no recall, no attempts to engage him as I knew they would fail and further diminish their already almost non-existent discriminative properties. We would go out early mornings or evenings to avoid other dogs (major source of reinforcement for Darwin). I did not offer any food under these conditions to avoid food refusal. I simply manipulated the long line when necessary. Movement was valuable to him, so I ran a lot, and he would chase. I did not try to put a contingency on those reinforcers (i.e., sit or look first then I’ll run and you can chase me). They were given non-contingently (non-contingent reinforcement).
The easiest way to bring escape behaviour to zero levels of occurrence is simply not to present the EOs for it. In a negative reinforcement contingency, the termination of the antecedent is the reinforcer. If the antecedent stimulus is not presented, there is simply no EO. In other words, there is no motivation to escape because there isn’t a stimulus to escape from or to avoid. I simply did not ask him to do anything. No “sit” or “stay” or “leave it” or “come” “look” or responding to name. I stopped all the early curricular objectives I had been taught in the puppy classes. Trying to establish stimulus control when you are lacking the most important element of the contingency, an effective reinforcer, will simply increase the value of escape or avoidance. For two days we co-existed. The long walks made him tired, which meant he was less likely to try and look for other competing reinforcers in the house (e.g., chewing the dirty boxer shorts my boys keep leaving on the bathroom floor).
Then, I rebuilt the value of food in my presence in a low-demand teaching context and in the absence of competing EOs (a fairly empty room). I chose not to use a leash to prevent roaming; restricting movement would have been too easy. He always had the opportunity to escape, but I stacked the deck in favour of remaining with me by devaluing escape (sanitising the environment) and increasing the value of remaining with me through animating the delivery of food rather than simply feeding to mouth (bowling, flicking, catching) for the behaviours of:
· Chasing kibble
.Turning toward my direction
· Moving toward me
· Remaining in close proximity (within 50 cm) for increased durations
No verbal stimuli were used at this point: no instructions, no event markers. I said absolutely nothing throughout our sessions, which replaced the 4 bowl to bowl feeds.
The technique itself looks simple enough: pair the food with the “aversive” context and the aversive value will decrease. And we could leave it at that, but the beauty of a behavioural description of a relatively simple phenomenon is, at least to me, the dissection of all the variables at work. Even a simple intervention as the one described can be viewed at multiple levels of complexity and as the interaction of the different behavioural principles and processes at work.
My graduate ABA students will know that I have an intense dislike for when “pairing” is freely invoked either as the “cause” or as the “outcome”, because you still need to use a behavioural measure to assess if “pairing” has occurred. The contingency, I believe, in this case was operant. I wanted my presence and what I did to be signals for the availability of reinforcement for approach behaviours and for maintaining proximity for incremental durations. These were my initial targets, over a couple of weeks. I got increased speed, tail wags, shorter latencies between one presentation and the next and some eye-contact. In other words, I got “enthusiasm”.
Approach and proximity are the two fundamental objectives in any applied human early behavioural intervention and would ideally be established before any formal teaching is implemented. They are the behavioural measures of “building a relationship”. The instructor manipulations are based on creating the conditions for the target (social) behaviour to occur without any guiding or prompting, but through what is sometimes referred to as “pressing”. The fact that this also happens in the dog training world speaks to the generality and the beauty of behavioural principles.