Knowledge, Experience, and Reinforcement

Hierarchy of Behavior Change Procedures

There are currently two discussions happening in the animal training world that are related to how people should choose what approaches to try first when they decide an animal’s behavior needs to change. One of the discussions involves a magazine columnist—at the magazine where I held my first job, though our tenures did not overlap—who wrote a personal essay about his decision to use a shock collar on his dog after first failing at certain goals with unspecified positive reinforcement approaches. The other discussion is about the Hierarchy of Behavior Change Procedures, a visual tool developed by one of my mentors, Dr. Susan Friedman, to help animal caregivers choose the effective approach that lets the learner retain the most control and also produces the most behavior that we might associate with good welfare and “positive” emotions.

These guidelines and the more nuanced article originally intended to inform their use have been used most recently as the basis for the "LIMA standard," a set of professional ethics adopted by a number of prominent animal training organizations, and as such of course should be revisited and reexamined regularly. For me, one of the most salient points of a new article Dr. Friedman published about the hierarchy this week is that the use of procedures that are ethically preferable, such as reinforcement (procedures that strengthen desired behavior), may require more expertise and experience than the use of punishment (procedures that decrease or suppress behavior), which we all get plenty of practice at from the day we are born, and which, when it is effective, can provide immediate gratification. While learning is a lifelong process for trainers as well as animals, she argues, those who are paid to dispense behavior advice in particular should seek education, experience, and mentoring to make sure animals don’t pay too dearly for whatever our current shortcomings may be.

The idea that how effectively one can use reinforcement depends on the knowledge and skill one has at the time came to mind when reading criticism of the magazine article by professional dog trainers (which is what called my attention to it), and then again when editing the following short before-and-after video (shared with permission from my client). Although the video shows only the beginning of a training process, I think it is a good example of how reinforcement-based strategies may succeed or fail, or fall somewhere in between, based on nuances that rely on education, experience, or both. (I would also add that while most humans are experienced with punishment, using it while minimizing the risk of fallout depends on similar nuances and most of us are not so good at that.)

The dog in the video below generally behaves to escape and avoid having her paws handled and her nails trimmed with any kind of handheld device. Her devoted and patient owner is a previous client, and with very minimal help from me a couple years ago, she taught the dog to file her own nails using a scratch board. But she still needs a way to tackle the dew claws, and so she had recently embarked on a plan to help her dog feel comfortable having her nails Dremeled. She asked for my help because she got stuck. I recognized where she got stuck, because I’ve been stuck there before. In fact, I got stuck in a similar place with my own dog years ago, and, full disclosure, I didn’t work through it in as pretty a way as you’ll see below. But I have more knowledge and experience now than then I did then, and I hope I’ll have more next week, next month, and a year from now.

Here’s the video. The first clip is the beginning of our session, conducted via Zoom on May 28, 2020. The second clip was sent by the owner on May 30:

Some excellent thigs the owner was already doing:

• Starting with the dog in a settled base position that was already trained and associated with lots of good stuff
• Working in a space where the dog could leave at any time
• Breaking her goal down into small steps--she has also already separately associated the sound of the Dremel with bacon and begun using the Dremel to grind penne to introduce the way the sound will change when it is used on a nail (a technique I like to call "pasta alla Torelli" because I learned it from Laura Monaco Torelli)
• Moving deliberately so she did not surprise the dog
• Using food items that her dog loves. Bacon!
• Asking for help when she got stuck

Here are all the little things we changed after reviewing where things stood, in order to get to where things are now:

• Skip the clicker and just use hand movement to mark behavior, which will improve timing of reinforcement.
• Elevate the mat on the couch so that the dog’s paws are hanging over edge. That way, she doesn’t have to stand up or lift her tiny leg up awkwardly from a down position on the floor to offer a paw. She can then remain in a comfortable and stable position while her paw is in the hand.
• Aim for the dog to initiate contact, rather than to “allow touching.” (Note: The dog’s willingness to do this this likely came from my client’s prior simple pairing of touching + treat.)
• Wait to start the next trial until the dog finishes eating, so that the prior treat is not still in play before the next contact is made. That way the dog doesn't have to choose between eating and leaving and we can more clearly see whether she is participating voluntarily. We are also less likely to make her suspicious of treats because they're not predicting being touched.
• When she’s done eating, watch what she does. Pick a behavior, or several, that will cue you to present the next trial, such as looking at you or lifting a paw. If you don’t see the behavior you’re looking for, or if it comes hesitantly; adjust the plan.
• To get rid of the “swiping” behavior, the product of withholding the click to get duration, temporarily reduce the requirement for the paw touch to smallest success point (h/t Laura Monaco Torelli), the smallest perfect behavior (h/t Mary Hunter, though I may be misremembering her language), the shortest antecedent-behavior-consequence loop (h/t Alexandra Kurland), or whatever you want to call it. Reinforce the moment the paw makes contact with the hand, which will actually mean anticipating when it is about to make contact and moving the treat hand a little early.
• Deliver the treat low, under her nose, so that she does not have to lift her head or stand up to eat, and can easily keep the paw in your hand.
• Rebuild duration from this point by treating for just a second or two of leaving the paw in your palm.
• Begin to gradually increase the time between treats.
• End the trial while the paw is solidly resting in the hand by removing your hand and stopping treats.
• In the space created when treats can be delivered farther apart, introduce approximations toward holding and manipulating the paw. (More nuances of this step have since been discussed, but are not fully demonstrated in the video.)

Further clarifications I saw the opportunity to make after watching the second video:

• Whatever you’re going to do, that’s the deal for that trial. If you are ready to increase your criterion, do it on the next trial, and not in the middle of the current one. Don’t change the deal mid-handshake.
• When you hear yourself thinking “man, she’s doing great, let’s get one more,” toss a treat off the mat and see if she comes back quickly and enthusiastically to ask for more. If she doesn’t, end the session and adjust the plan.

We still have a ways to go before I would call the dog “comfortable” with paw handling, but this is such great progress for a day’s work, and small changes made a large difference. Further, I’m certain that my peers and mentors, including those named above, would see even more opportunity for improvement. (In fact, I can hear Laura Monaco Torelli in my head now, reminding me to intersperse other fun and well-known behaviors.) There’s almost always a way to do better.

Lest this inspire despair, know that in many if not most cases, your ever-changing best will be plenty. This is evident from the number of people currently living more or less happily with dogs and vice-versa, despite a lack of training and less than perfect behavior on both sides and yes, even the occasional aversive stimulus.

But my point here is: If you’ve tried reinforcement and it has failed, there may still be a lot you can do before you start assuming you must use more aversive procedures, and some of it can make a big difference fast. (Notes: 1. This client, as far as I know, was not considering more aversive procedures. 2. I'm not saying the magazine writer did or didn't try such things, or how the use of the shock collar affected his dog's welfare; I have no idea.)

As far as the application of the hierarchy goes, it is worth noting that this process of giving the learner more control involves both positive and negative reinforcement procedures. The dog can take her paw away and that behavior will be allowed to serve its function of escape or avoidance; she won’t need to squirm or bite in order to stop the procedure if we miscalculate what she’s ready for. Or she can meet the current training criterion and get treats, which also happen to be followed by release. And she can leave the session entirely any time she likes.

Ideally criteria are set carefully, so that she doesn’t choose to take her paw away or leave very often, if at all, and we see loose body language and low latency. But I would also be interested, based on some very cool work by both Ken Ramirez (2017, teaching a beluga whale to say “no”) and researchers in the human world (Rajaraman et al., 2019, “enhanced choice model”), in seeing what might happen if we gave her the same treat for taking her paw away or some other alternative behavior as well.

Are there arguments to be made that a process like this is not always the least intrusive, or least restrictive, effective approach? Sure. First, it remains to be seen if this will be effective in the long run. Second, if this were a dog who had come from the shelter with nails curling into her paw pads, it would probably not be the option that immediately provided the best welfare. I would likely recommend having a kind vet or groomer get the job done in some other context, and then start work in a different setting on cooperating with nail trims. But happily this dog has nail-filing in her repertoire, and an owner with the resources and time to improve her own skills.

To Teach Loose Leash Walking, Loosen the Leash (and Other Insights From Years of Doing It Wrong)

There are many ways to teach a dog to walk on a leash, and I have tried a lot of them over the past 15 years, ranging from sharply yanking the dog’s collar every time she pulled (this was before I was even a student of dog training, and I don’t recommend it) through passively using various halters and harnesses to make pulling uncomfortable or counterproductive to clicking and treating when the leash was slack.

As I’ve learned more about both dogs and behavior, it has become apparent that "loose leash walking" is really a complex set of interactions between the dog, the handler, and the world--not a single or simple behavior.

How I teach it has evolved accordingly, some of it relatively recently, and I'm sure it will continue to do so. But here are what I currently think are some of the key pieces. Note that no one by itself is a magic solution (though sometimes the first one can seem like it). Some set the stage for others to work, and they are ideally implemented together.

Use a longer leash.

I was pleased to see this article yesterday about a study by two French trainers who found that dogs sniffed more on longer leashes, and that as they sniffed, their heart rate slowed. Here's their own report. It's cool way to try to measure what many dog trainers have observed more anecdotally, which is that giving dogs more control over their walks appears to have positive effects on their emotional behavior. Alexandra Horowitz’s canine cognition lab at Barnard has also published peer-reviewed research showing a connection between nosework and behaviors interpreted as “optimistic.”

What I want to suggest here, though, is that giving dogs more room to roam and sniff on walks may also have positive effects on their walking behavior.

When we attach dogs to a 6-foot rope and march them down the middle of a sidewalk, 8-10 feet from what they find interesting, we are setting them up to learn to pull. Basically, we tempt them with treasure just out of reach, and then when they hit the end of the leash on the way there, they learn that a little extra oomph will get them closer. For some dogs, a strain-and-sniff on a tight leash, even on a painful collar, even with an owner barking orders at their back, is still worth it.

So, how long a leash do you need? Maybe just longer than what you are using now, but not so long that you don’t feel comfortable or safe. Just a foot or two can make a difference for some; if you have a 6-foot leash but currently carry it gathered up or knotted, try just using all 6 feet. My personal preference is for a thin, flat biothane (like the one in the picture up top) leash somewhere between 8 and 15 feet long, depending on where I'm walking. I generally prefer nonrestrictive harnesses, like the Balance or Ruffwear harnesses, over collars, but the longer the line the more important it is that it is attached to the back clip of a harness, and not a collar or the front clip of a “no pull” harness like the Easy Walk. (Also note: retractable leashes don't work for this, because when they are extended, they are always taut.)

For puppies, I think a longer leash is especially beneficial. Their developmentally appropriate and necessary exploration can quickly become pulling when you add a short leash, and the ensuing struggle can put a early dent in the relationship between person and puppy. I've been lucky in recent years to have access to a few puppies over extended periods of time, and what I've seen so far is that as they learn about the world on a longer leash, get a lot of treats for checking in, develop a broader responsive relationship with their owner, and maybe just get a little older, they start to wander less and walk closer.

Even for older dogs who seem to be way more interested in access to the environment than any food, toy, or attention you have to offer, using a longer leash often seems to make walks less of a battle and can help the dog learn that pulling is not the only way to get those reinforcers. The dog may in fact be able to get enough of them on a long line that they become slightly less important, opening the door for you to use things like food or toys to teach more attentive behavior.

In short, by allowing dogs more free access up front to environmental reinforcers, we may be able to lower the value of those reinforcers through satiation (a move known in behavior analysis as an abolishing operation), and thus make pulling less likely.

We can also avoid reinforcing pulling, because the dog doesn't need to pull to get to those reinforcers. Instead those reinforcers may strengthen whatever he does instead. And we can minimize or avoid punishment, which may create other problems, because we aren't putting ourselves in a position where we have to react to a dog who is already pulling.

Using a longer leash doesn’t mean that the dog should be allowed to run up to anyone and anything. There are of course times when you need to keep your dog close to you, and if you don't want your adult dog to run up to everyone, you shouldn't teach her to do it as a puppy. But you can shorten a long line, and teach the dog what she should do instead when she can't reach something of interest. You can’t, however, lengthen a short line when it's OK to explore.

There may also be some situations, e.g., where a dog is a threat to people or other animals, in which a longer leash may not be appropriate—at least not immediately, without professional guidance and extra safety precautions. But these dogs, too, often benefit from being carefully given more leeway in their movement through the world. If you were worried about your surroundings, would you feel better or worse if you were routinely walked through them on a short, tight leash? Close restraint limits the behavioral options a dog has if she wants to escape or avoid something; if you can't flee in those circumstances, you are more likely to fight. This problem, where you need to control the dog for safety but restraint makes the situation worse, might mean you choose to walk where you don't need to restrain, and/or that you need additional intervention for the problem behavior.

However long it is, try to keep the leash slack.

When dogs continue to pull on a tight leash, it is often attributed to the probably misnamed “opposition reflex.”

But it may be more useful to think about how a tight leash can become a cue for pulling or other voluntary behaviors--the very behaviors you are probably trying to prevent when you tighten the leash.

Let’s say the dog is pulling toward something she wants to smell, and feels tension on the leash. She then pulls harder, and gains a bit more access to that smell. Or as the dog is pulling toward a person, the owner pulls back, lifting the dog and making it easier to stand on hind legs than on all fours; the person then walks forward and the dog gets petted while jumping. Or a dog sees a threatening dog, the owner tightens the leash, the dog barks and lunges, and then the other dog goes away. Things that routinely happen just before a behavior is reinforced--in all these cases, the leash tightening--can come to evoke that behavior in the future. You already know this if you have ever intentionally “captured” a behavior, using a clicker and treats, and then added the desired cue by saying it just before you know that the dog is about to offer the behavior again.

Keeping a leash slack does not require jogging to keep up with the dog or letting the leash spool out longer than is safe or appropriate in a given setting. Light-handed but effective leash management is a set of skills that can be broken down into steps and learned by handlers. A key component of Grisha Stewart’s BAT technique, for instance, is the skilled use of a long leash, and she does a great job breaking those skills down. Two moves I like to pre-teach are a “gather,” in which the owner learns to shorten a leash quickly and efficiently without adding (much) tension to it, and a “dip,” in which tension on the leash becomes a cue for the human to lower her hands, removing tension without adding (much) length. I additionally like to pre-teach the dogs that both of these moves are cues for their behavior: if I do one or both of those things, I want the dog to orient to me, because I’m probably getting ready to do something that will be easier if the dog is paying attention.

I teach both by doing the move, then immediately (and initially regardless of the dog's behavior), clicking and treating close to my body. Eventually when the dog feels or sees you doing the thing, he will orient toward you before you even have a chance to click; at that point, the move has started to become a cue.

Reinforce behavior, not products of behavior.

A common piece of advice in teaching loose leash walking is to reinforce “when the leash is loose.” But what is the dog actually doing to keep the leash loose? You can shape up some really odd behavior when you focus on the end product. Put in human terms, do you want to reinforce good test scores, or do you want to reinforce answering questions correctly and all the prerequisite behaviors needed to do that? Pick the former and you may reinforce cheating or guessing.

Teach prerequisite skills.

OK, then what behaviors do you need to teach your dog to do in order to have a nice outing with you on leash? My ideal end product looks like the dog meandering calmly (as in, not zipping across my path unexepctedly) on whatever length of leash is appropriate for the situation, sniffing whatever he likes (within safety parameters), checking in with me frequently, and trotting within 3-6 feet of me on a slack leash when not otherwise occupied; I don't really care which side, but I feed a lot on the right so usually that becomes the default. I also like to be able to cue a short-term close-walking behavior if needed.

Your ideal picture may be different from mine, but whatever pieces there are to it, identify them and teach them in low distraction, so both you and the dog can get fluent before there are lots of other things competing for your attention.

Here are my biggies:

• Looking at me voluntarily produces good stuff, here, there, anywhere, in a house, with a mouse. Start at home, but then green-eggs-and-ham that shit. If a dog likes to look at you a lot, when he's not otherwise occupied, he'll probably hang out in a position that lets him look at you easily. Start teaching this for the first time with your new longer leash attached, so that the leash can help tell the dog this game is on.
• Trotting next to me on cue. Reinforce for looking at you from close to your side a few times, then move. The dog will likely move alongside so he can get in position to look at you again, and you can reinforce the movement. You can then add a verbal or visual cue to ask for this behavior when you need it.
• The dip and the gather are cues to look at me (described in a previous section).
• What to do when I stop. When I slow down, as when coming within about 10-15 feet of a blind corner or a curb, I start to gather the leash. If I've taught that as a cue, the dog will start to pay attention, and if I want (and have taught these skills), I can cue him to sit or wait. For turning your own stopping into a cue for the dog to do something particular, see this older post.
• What to do if the leash goes tight: No matter how hard you both try to do what you're supposed to do, there will likely still be times when the dog ends up on a tight leash. Like most trainers of any stripe, I don’t go forward on a tight leash. The dog may feel tension if I start to slow down and he's not paying attention at that moment. Or you may not be able to completely break your own old habits (a scenario that should engender some empathy for your dog, who's been reinforced for pulling for years). For instance, if I see, too late that a dog is about to encounter something she simply can’t explore--chicken bones, synthetic marijuana, a person who did not ask to be explored, or . . . Jesus, is that what's left of a rat?--I’m going to put on the brakes. Even if I know intellectually that the dog has a strong response to "leave it" or "drop," I will probably still tighten the leash. I might then “dip” the leash--but sometimes I might not, say, if even a few inches will mean the dog could reach the booty.

So what should the dog do when that tension continues? As with the dip and the gather, I want to teach that ahead of time, and I want it to be an opportunity, not just an annoyance--a cue for the dog to turn away from the thing and get something potentially greater than what he leaves behind. I teach this more or less the same way I teach the dip and the gather, by having the smallest possible amount of tension predict a treat behind the dog until the tension begins to cue the dog to look for the treat. Start with mild or no distractions at all, not around flattened rat carcasses.

• What to do after checking out a "distraction": The innovative trainer Leslie McDevitt has worked out lots of fun and effective "pattern" games for teaching dogs heavily practiced sequences where conclusion of one behavior (e.g., sniffing) becomes a cue to do the next (e.g., checking in with the owner). Your preferences may differ, but in my book the dog checking out stuff (without needing to pull to get to it), and then checking back in, rather than shooting forward to the next thing he wants to investigate, is a very pleasant pattern for both of us. Again, you should teach these patterns in an easy setting first, before taking them on the road.

Deliver treats (and other reinforcers) with purpose.

Ideally, the delivery is timed so that it catches the dog in a behavior you want to see more of (say, trotting next to you with head forward), and also sets him up to do the next movement you want to see (be it continuing to move forward in the same position, or taking a sniff break). I can’t say I never screw this up, but I don’t want the timing to reinforce, or the position to set up, awkward behavior like walking down the street with neck craned backward to look at me. I most commonly deliver treats in two places on walks: next to my hip on the side I want the dog to default to when he walks close (Hannah Branigan calls this the “takeout window”) or out in the grass slightly ahead of the dog. But considerations include where the dog currently is (if he’s next to me and I want that to continue, the treat will appear in front of his nose) and where I want him next (if I want him to learn that sniffing is part of the reinforcer for checking in, I’ll toss the treat where he can continue walking forward to gather it and then also sniff).

UPDATE: Here’s video (jerky screen captures of FaceTime during the COVID-19 shutdown, sorry) of treat delivery with purpose: The grass is full of smells that are powerful reinforcers for Booker, so in tossing each treat back out into the grass, the goal is to double the reinforcement for catching up and checking in: a treat plus a return to the grass. The treat placement also sets Booker up to “catch up” easily. And of course the longer leash lets Booker sniff without setting him up to learn that pulling is worth it. In the second video, Jerry is increasing the amount of walking nearby required to get the treat, still tossed into the grass:

View this post on Instagram

A post shared by Kiki Yablon (@kiki.yablon)

View this post on Instagram

A post shared by Kiki Yablon (@kiki.yablon)

General rules for good training apply.

There are any number of other arrangements that might be important depending on the dog and the situation. Some might involve what to do when novel distractions appear (see this older blog post about teaching the dog what to do instead of running toward them). Most would apply to training any skill. For instance: start new skills in lower distraction settings; the dog gets to decide what is reinforcing; what she finds reinforcing will change depending on what else is available; reinforcement should be plentiful when learning new skills; and new skills should be broken down down into small achievable steps.

Note that's skills, plural. Most helpful in my thinking about walking has been this realization, that loose leash walking is a set of skills, for both handler and dog, that will need to be perpetually combined and recombined while navigating the world. Once we break down and teach these skills, we can also practice the ways in which they might intertwine.

Trainer Worries About False Claims About False Claims That Dogs Lack Emotion

photo: Taro the Shiba Inu, CC by 2.0

In an October 12 blog post for Psychology Today, “Trainers Worry About False Claims That Dogs Lack Emotions,” evolutionary biologist Marc Bekoff reports that some dog trainers (and at least one ethology professor) have written to him to express dismay at claims supposedly originating from neuroscientist Lisa Feldman Barrett and behavior analyst Susan Friedman that dogs do not have emotions.

I can’t speak in depth about Feldman Barrett’s work, as I still haven’t finished her book, How Emotions Are Made (though Bekoff seems to indicate he hasn't read it either). But I teach fellow animal professionals about behavior analysis under the mentorship of Susan Friedman. As many of her students have already pointed out on social media, there are some grave misconceptions about what she teaches in the post.

Bekoff quotes an e-mail from a dog trainer, identified only as well-known:

“Susan Friedman, for instance, insists that terms like ‘protective’, ‘pushy’, ‘playful’, ‘reserved’, ‘grumpy’, ‘extroverted’, ‘curious’ etc. should never be used when speaking about animals, with these being examples of improper verbiage...It is a powerful force in the dog training world right now that's moving away from an integrated understanding that sees the whole phenotype of the animal and considers both proximate and ultimate explanations for behavior."

Bekoff then links to a longer list of adjectives included in one of the free handouts about behavior assessment available on Friedman's website. In the handout, Friedman advises that these words don’t “describe behavior in a functional way.” Bekoff writes that he can't understand why she says these words are "functionless."

First, I don’t think anyone who has taken any of Friedman’s courses or attended her seminars could report that she has said that animals do not have emotions. Should you be interested in her perspective on emotions (and have $60 to spend well), you can watch her give a 90-minute lecture on the subject to an audience of animal trainers, drawing on a multidisciplinary basis of evidence, on ClickerExpo’s video-on-demand site.

Second, the reason the words on the worksheet are described as “not functional” is not that they serve no purpose. What Friedman means here is that these words do not describe the function of the behavior.

In behavior analysis, the function of a behavior is more important than the form, which means it is never completely described without reference to its context. We all seem to agree on this. Context is what Bekoff suggests that we look at to determine how animals are feeling:

“I've argued that there's chimpanzee joy, dog joy, and human joy, and elephant grief, gorilla grief, and magpie grief, and even among members of the same species (conspecifics), these feelings and the ways in which they're expressed may vary. However, there's little to no doubt about what the individuals are feeling when their behavior and the context in which it's observed are detailed.”

Bekoff almost sounds like a behavior analyst here: The form of “grief” looks different not just between species but between individuals within a species, and so the only way we can reasonably infer that, say, crying is an expression of “grief” in humans is by seeing it in context. UPDATE: Case in point:

Then again, how are we to even tell whether crying in the context of a friend’s recent death is “grief” and not, say, “regret” at not having spent more time with her before she passed? We could ask, but how will we then verify that the answer (itself behavior, and influenced by its own conditions, let's say a public setting) accurately represents the private behavior of the crying person?

Describing what an animal is doing, rather than what we perceive it to be, what we think it is feeling, or even what it says it's feeling, is what allows us to locate the behavior within the conditions that are likely responsible for it. Immediate conditions are one of the three main factors influencing behavior in the behavior analysis model, along with past experience and genetics. And they are notably the only one of the three we have any control over.

So, just to be clear: The description of what the animal does, under what conditions, to produce what outcome, is a functional description in that it describes the function that the behavior serves. Not in that it functions as a description, though . . . that too.

Why is a functional description important? A couple of examples from my own work:

If you tell me that your dog “hates guests,” before I give you any advice, I’m going to ask “what does that look like? What does the dog do? When he does it, what do people do in response?”

If you tell me that when people come in the house, if the dog stiffens, lowers his head, peels back his lips, and growls, then people then stop approaching him, we can probably make an educated guess that he is behaving to avoid proximity to people. Do we want to agree to call that “hate?” What about “fear”? Or is it “dominance,” as some would argue?

These labels are important, not because they describe anything other than behavior in conditions, but in that they may influence how the owner (or trainer) will attempt to address the problem. They may be useful in engendering empathy (and yeah, we have to ask "what does that look like?"). They may be harmful in encouraging solutions directed at the “hate.” But they do not add anything to the understanding that if we want the dog to change his behavior, we will need to change the conditions so he can maintain distance from guests without threatening them, and further, so that maybe maintaining distance from guests will no longer be an outcome worth growling for.

On the other hand, if you tell me the dog “hates” having guests over, and when the specifics are investigated it turns out that he barks repetitively in a high pitch at you whenever you sit on the couch with a guest, and each time he does, you get up and give him a new bully stick or Kong to work on, we’re going to make some different environmental changes from the first scenario.

Is it any more useful to say a dog (or for that matter a person) behaves because he “hates guests” than to say he behaves because he is "pushy"? (For the record, Friedman has never said such words should never be used to describe animals; what she has frequently said is that labels for behavior—of animals and humans alike—can be useful, but only once all stakeholders know what they're shorthand for.)

In both cases above, I will need to know: what behavior are we talking about, and under what conditions?

Do we disregard emotions by focusing on behavior we can observe and measure, instead of what we presume to be in the dog's mind or brain? No, because the behavior we can observe and measure is the same behavior that we use to infer emotions. When the behavior changes, we look at it under its new conditions and infer different emotions.

Another misconception: The well-known trainer refers to what trainers are being taught of applied behavior analysis as “methodological behaviorism.” That’s incorrect. Simplified, methodological behaviorism (a) generally assumes that there are processes occurring in some unseen dimension, e.g., the so-called mind, that mediate between the environment and behavior and (b) that since these processes can not be observed or measured, they are not appropriate subject matter for a science of behavior, and are to be set aside. (My source on this is here.)

B.F. Skinner’s radical behaviorism (radical as in thoroughgoing, far reaching, and fundamental, not as in extreme) distinguished itself from methodological behaviorism in part with the gorgeously parsimonious assumption that private behaviors, including those we might call emotional, are still behaviors, and that though they are hard to verify and sometimes hard to identify even within ourselves (do I feel sick? Or am I just tired?), they are subject to the same influences as public behaviors. They are obviously important to the organism who experiences them, who can observe them, and a thorough behavior science is obligated to try to account for them. The pioneers bringing ABA to animal trainers are, to my knowledge, radical behaviorists. (Friedman addresses other misconceptions about Skinner and behaviorism in another video on demand here.)

Radical behaviorists also certainly take into account phenotype, contrary to what the well-known dog trainer asserts, and both immediate and distant explanations for behavior. In fact, Skinner wrote a lot about evolution. He saw selection of behavior within an individual’s lifetime as both similar to and intertwined with selection of behavior over a species’ existence, and it was from this perspective that he strove to account for complex human behavior and its possible origins (see, for instance, The Evolution of Behavior).

The question among dog trainers who operate with more than a casual knowledge of behavior analysis, then, should not be whether dogs have emotions.

Bekoff suggests that we ask not whether emotions have evolved, but why, which is an interesting question whether you’re an evolutionary biologist or a behaviorist. Skinner sure thought so.

But we might first have to answer this one: What are emotions?

Is there any evidence that they exist as something separate from behavior and context?

How do we come to an agreement on what behaviors go with which labels? Is a stiff upper lip “stoic” or “angry”? How can we verify that a person who says he is "happy" and smiles and a person who says she is "happy" and weeps are experiencing the same private behavior?

If we can't find emotions in the mind (which probably doesn't exist), or in the brain (Feldman Barrett argues convincingly here, in lay terms, that emotions don't live in particular brain regions), or in consistent facial expressions (Feldman Barrett again presents falsifying evidence), or anywhere other than in the behavior we observe and the context we observe it in, does it exist anywhere else? And if not, can emotion "underlie" behavior, as dog trainers so often say?

One last and more practical question: What should dog trainers focus on if they want to ethically and effectively change behavior or, allowing for the possiblity that they’re distinct entities, emotions? What is in our power to directly change?

The concept of functional description that Bekoff misunderstands is a key component of functional analysis—which, on the subject of ethical behavior change, has been correlated in human behavior with a marked shift in human practitioners’ choice to use reinforcement-based procedures over punishment. As behavior analyst Gregory Hanley writes, it is a "humanistic" approach: it “dignifies the treatment development process by essentially asking the [organism] why he or she is engaging in problem behavior prior to developing a treatment.”

And while observation of behavior can arguably be supplemented with actual “asking” in humans (as unreliable as their answers may be), the observation of behavior in conditions is all we have with animals. So it’s critical that before we agree on whether and what to label it, we see it with clear eyes.

Training With the Grain

Edited to add, on February 14, 2023:

Teaching “alternative behaviors" might be most ideally described as teaching animals how to more successfully behave to get what they want or need.

This concept, that we can provide new routes to important reinforcers, rather than punishing the routes we don't like, is a brilliant, life-changing concept. But traditional and even early positive reinforcement dog training have involved some very crude, suppressive interpretations: Give me laser eye contact instead of looking at that dog you usually bark at; lie perfectly still on this mat instead of jumping on visitors; come, then touch, then sit with duration instead of running off; sit instead of ... jeez, everything. These are often hard asks in those situations, and the reinforcers on offer (avoidance of corrections, food) often have nothing to do with what the dog is going for.

(Note: I use a ton of food in my training, especially in teaching new skills. There are times when I cannot figure out how to give the dog what it is actually behaving for, and food often helps there too. And, to the best of my ability, once the dog has some usable skills, I'm looking out for ways to transition to reinforcers that are more likely to be produced directly by that behavior.)

Selecting better alternative behaviors--easier, faster, fewer at a time, more comfortable, likelier to access naturally occurring reinforcers---is next-level training. It starts with observing the behavior in context and also observing what skills your dog already has. They have a lot!

Now, on to what I wrote originally:

Most recently in her brand-new podcast, British trainer Kay Laurence repeated a point she's been making for years now, recommending that we rethink “sit” as the gold standard for a well-trained dog. When we ask a dog to place its hindquarters on the ground to prevent it from doing some other behavior, she argues, we are squeezing “positive” training into a traditional mindset. In applied behavior analysis, this is referred to as a pathological approach (geared toward "curing" unwanted behavior), as opposed to a constructional approach (envisioning what is wanted and then figuring out how to build it).

Don’t care what “mindset” your training is squeezed into, so long as it works? Well, Laurence is practical as well as dog-centric. At a seminar at For Your K9 several years ago, she noted that when we teach an "unnatural" behavior, it will require more maintenance (read: continued reinforcement by the trainer) over time. And for many dogs, in many situations, she argues, sitting is an "unnatural" response. It “goes against the grain.”

This observation is consistent with the basic laws of behavior, though we might argue it is the context, not the behavior, that is "natural" (or not). In the sense that behavior is an adaptive tool, all behavior is "natural" (a distinction Susan Friedman has emphasized, and that goes back to B.F. Skinner observing that the organism was always behaving as it should). It's through natural processes that behaviors like "sit" acquire new and counterintuitive functions, such as getting food. (What animal sits to get food in "the wild"?) If there are no naturally occurring reinforcers available for the desired behavior—or if it is especially effortful in a given situation, or if in order to do it the dog is consistently asked to give up reinforcers it could get for doing a different behavior—then a trainer will have to continue to contrive ways to make her preferred behavior reinforcing for the dog.

Deciding what we want the dog to do instead of behavior x or y is the core of problem solving in modern dog training. And if we keenly observe dogs—both in general and as individuals—as well as the behaviors they choose when they are allowed to choose, we may find that there are better options than the standard “obedience” repertoire suggests. By “better” I mean that they may both fulfil the dog's needs or desires and require less maintenance effort on our part.

To this end, the larger dog training community of which I am a part seems to be thinking more and more about facilitating vs. suppressing behavior. One lovely example appeared on the blog Eileenanddogs, where Eileen Anderson wrote about “bootleg reinforcement,” a name for competing reinforcement that undermines training. Bootleg reinforcement can range from a treat inadvertently dropped by the trainer to interesting odors wafting by on the wind to the relief of an empty bladder. Very often, it is naturally occurring reinforcement; we call it "bootleg" because it reinforces something other than what we are training.

Anderson was having a problem with a behavior she had reinforced with food in order to stop her dogs from doing something else: All three dogs had been taught to lie on their mats just outside her mudroom so that they wouldn’t all crowd into the small space near the back door. But one of them, Clara, kept getting off the mat in order to cross the threshold and sniff her. This dog's behavior demonstrated that access to sniff was more reinforcing than the treats available for staying on her mat. What's more, staying on a mat (though lord knows it is often a successful and helpful alternative behavior) would be a pretty unlikely behavior for an untrained dog to use to track a scent of interest.

Rather than punish Clara, which would damage the relationship she had carefully built with this formerly feral pup, or give her even more food for staying on her mat, Anderson observed what the naturally occurring reinforcer was for Clara’s unwanted behavior and decided to figure out an appropriate way she could give it to her.

“The standard advice for a competing reinforcer situation, such as the choice to ‘get on the mat for a cookie’ vs. ‘take a sniff and get some novel odor,’ would be to raise the value of the reinforcement for the desired behavior, and start over and practice in easier situations,” Anderson wrote. “And it generally works. We don’t think of positive reinforcement as a particularly intrusive solution, but often we do it as a substitute for the animal’s first choice of behavior. And the desire for that behavior has no reason to fade. So—what if we could make the competing reinforcer non-competing? What if we could make the bootleg reinforcer legal?” This wouldn’t do with “behaviors that are never acceptable, like eating cat poop or knocking over toddlers, but sniffing? Why not try it?”

Visit the blog post for a video of Eileen’s elegant, dog-centric solution.

So, some takeaways about how to "train with the grain": (1) Consider your choice of alternative behavior--is it easy or hard for this dog in this situation, and if it's difficult or effortful, is there an easier, less costly, acceptable alternative? (2) In searching for powerful reinforcers, watch the dog. When dogs are "misbehaving," they often might as well be holding up a big flashing neon arrow that points to the strongest reinforcers on the scene.

Just one example: Many owners ask their dogs to sit to get their leash put on for a walk. Not only is this not necessarily the most convenient thing for the owner, depending on the equipment (I find it harder to buckle a harness around a dog who's sitting than a dog who's standing), but it may not be a behavior a dog would likely do when getting ready for active movement. In the video below is one way I found to train "with the grain." Rather than picking Scout up and stuffing him into his harness because we could, or even placing it on the floor and teaching him walk onto it, we let him do what he normally did when the harness came out—jump up—and put the harness in his path to start capturing and shaping this behavior.

This post has been significantly revised from a post written several years ago for One Tail at a Time.

Redirect or Preempt?

Well-intentioned training "tips" can really go south if you don’t understand the underlying behavioral principles. One of those is the advice to “redirect” unwanted behavior, e.g., that when your dog jumps, or puts his paws up on the counter, or barks at you while you’re watching TV, you should direct him to a more appropriate behavior.

If your dog is already jumping on a frail, unsuspecting old person or small child right this very minute: by all means redirect him to another behavior. But know that if you make a habit of this, your dog will probably make a habit of jumping. Here’s why:

Cues are signals that predict reinforcement for a certain behavior. If you have trained your dog (as I hope you have) that sitting on cue will produce a treat, petting, play, or something else he wants—then through close, repeated pairing, your sit cue has acquired some of the value of that treat, petting, or play, becoming what’s known as a conditioned reinforcer. Money is a powerful conditioned reinforcer for humans—we work for pieces of paper or numbers on a screen because they have been endlessly paired with food, shelter, clothing, Netflix, Moroccan rugs, airplane tickets, and everything else we need and want.

Cues can thus be used to reinforce other behaviors. Which is fantastic news if, say, you’re trying to maintain all the behaviors in an agility run, where food or play can only come at the end.

But it also means that if the dog throws his paws up on the counter, and you give him a cue that means sitting will be reinforced, you may reinforce putting paws on the counter. Even, potentially, if you don't give him whatever that cue has been paired with. And behavior that is reinforced is repeated.

Please don’t leap from here, as so many do, to the conclusion that we must therefore punish unwanted behavior. Unfortunately things just aren’t that black-and-white. Punishment, especially (but not only) physical punishment, carries the risk of side effects including fear, aggression, apathy, and escape-avoidance behavior. The same conditioning that creates such good associations with cues trained with positive reinforcement can create bad associations with people, places, animals, and cues that may predict something aversive. (Read about “poisoned cues” here.) Punishment also doesn't teach the dog what to do instead to get whatever he was after, so even if you do choose to use it, you should also be teaching new skills--and if you are doing that well, punishment is likely unnecessary.

The fact is, if the dog is already doing the unwanted behavior, you’re starting out behind the eight ball. All your choices are less than ideal. You may be obligated for safety or other reasons to reinforce the behavior (e.g., dragging the dog away from whatever he’s using aggression to get distance from, or handing a barking dog a bully stick so you can finish your important work call).

If there’s no immediate danger, you may be able to simply make reinforcement unavailable (e.g., by continuing to scroll through Facebook as your dog barks in your face) or remove it (e.g., stepping away from the dog when he jumps to greet). With ignoring or otherwise removing reinforcement by itself, though, you risk creating frustration, which looks like escalation to more intense and emotional behavior.

If your dog has a rich repertoire of other well-reinforced behaviors that you like, however, he may try one of those next—be sure to notice and reinforce it.

Or you can redirect. Yes, you might reinforce the unwanted behavior. Is it the worst thing in the world? No, and sometimes it’s the best of your crappy choices. Probably anyone who lives with a dog will settle for it in some situations. I’ll confess right here that I redirect my dog from barking pretty much every night—when my husband is cooking dinner, and I finish up on the computer and walk down the stairs into the kitchen, she play bows and barks. I then cue her to her mat, and when she lies down (which for her tends to be incompatible with barking), I pull a fish-skin chew out of the treat basket and hand it to her, or play some fetch. Think she’ll bark again tomorrow? I do; however, I've accepted it as a reasonable request on her part.

If you do need to redirect, do it as quickly as possible, so that you only reinforce a little bit of the unwanted behavior. Then make sure the dog is not set up to immediately repeat the cycle—after all, setting up rapid repetitions and reinforcing each one is exactly what you do when you want to teach a behavior. Prevention of repetition might involve leashing, gating, or crating the dog, or just giving him something more appealing to do (say, chewing on a fish skin) BEFORE the unwanted behavior starts.

If it’s important to you to reduce this behavior in the future, that last item is one to focus on. Your first job is to figure out how to predict the behavior so that in the future you can preempt rather than react. Put another way: if you want to teach a different behavior in response to the cue that currently evokes the problem, you need to first identify that cue.

Ask yourself: What are the conditions that set the stage for the behavior? What time is it? Who’s present? What are they doing? From how far away? What seems to be the most immediate cue for the unwanted behavior? Can you prevent any of these conditions from even happening in the first place? (If the blinds are closed, the dog can’t bark at stuff he sees out the window. No “training” required.)

Next, what do you want instead? If the cue does happen, what would you like the dog TO DO? (This answer shouldn’t involve the word “not,” as in “not bark.”) A word of advice: Pick something easy. When the cue for the unwanted behavior occurs, that’s the best time to ask for (or just catch) a desired behavior. With repetition and reinforcement, the problem situation can begin to cue the new behavior instead of the old one.

Finally, what does the dog get out of the unwanted behavior? That’s the “reason” he does it again and again. Can you provide that same consequence "legally," either for free or so that it reinforces a more acceptable behavior? Can you provide something the dog likes as well or better? If the new behavior produces more reinforcement, over time, that’s what the dog will begin to choose when he has the option.

This systematic way of looking at the behavior in the context of its environment is called functional assessment. It lets us figure out when and why a behavior is probably happening, and gives us the specific information we need to get ahead of it—instead of waiting for it to happen and then trying to figure out what to do about it.

Now you’re ready to train. Does your dog have the needed tricks in his bag already? You may need to strengthen, or even teach from scratch, the behavior you want the dog to do instead. Best to do that in low distraction first, then introduce it into the problem scenario. While you're doing that teaching, you need to prevent the problem scenario if you can (it's of course much easier if you have identified it, as described above). And you may need to introduce the new skill into the problem scenario in small, achievable increments.

But sometimes it’s much simpler than that. If your dog routinely jumps to greet, and you routinely cue him to sit before petting him, can you instead reach down and pet your dog while his feet are still on the ground, or invite him up on a bench or chair so he can get closer to your face without jumping? If your dog puts his paws up on the counter to get tidbits while you’re cooking, can you put his bed nearby, ask him to lie on it before you start food prep, and toss him bits of what he wants straight from the chopping block?

As a treat for reading, here's 20 seconds of Crash, the dog pictured at the top of this post. Crash had developed a hapbit of barking at his owners for the duration of meal prep and eating. He's a recent rescue, so we don't know his history, but this behavior looks to have been reinforced, perhaps intermittently, leading to a rather persistent version with the hallmarks of "frustration" discussed above. Over three sessions in the past month, Crash has learned new skills, including going to a mat, lying down, and maintaining that position in different settings and for different durations. This week we put it all together and added a problem scenario: meal prep. In the entirety of this session, a whole carrot was chopped into tiny bits, and Crash got to eat some of them—a powerful reinforcer for his new behavior.

Revised and updated from a blog post written for One Tail at a Time.