Dog training as gaming
November 4, 2010 at 5:31 pm 10 comments
Several different recent online alerts pointed me to this video from game theorist Tom Chatfield. Take a few minutes to watch his presentation and then let’s talk about how dog training is like gaming.
Modern computer games offer a stunningly wide range of carefully designed rewards. They also provide us with some really fascinating, and incredibly strong, tools to measure exactly what kinds of things people find rewarding.
Games keep us engaged largely through masterfully designed schedules of variable reinforcement. And game designers don’t just vary the timing of rewards, the value of the rewards varies greatly and smart game designers also offer different kinds of rewards including abstract things like karma and experience.
To keep our attention, a game can’t just offer rewards, it also has to offer some aspect of risk. We only stay fully engaged in a game when there is a real risk that losses (or aversives) will occur along with rewards.
Based on his work, Chatfield has come up with seven different ways that well-designed games reward our brains. His list bears a striking resemblance to the ways that I think a well-designed training program rewards our dogs’ brains.
- Complex games give us a way to measure our progress. When we play a game we want to feel like we’re getting somewhere. That we’re accomplishing something. And a good game gives us a way (or better yet, several ways) to measure that. This innate need to feel that one is making progress is one of the reasons why it’s important to break a training exercise down into discrete steps and give your dog meaningful input at each one of those steps rather than just at the end of a task.
- A game provides players with an array of different long- and short-term goals. Making progress on smaller goals helps maintain our motivation as we work to achieving the big ones. Small successes help prevent burn-out and frustration. This is something that people commonly lose track of when they work with dogs. Humans appear to be unique in our obsession with forward thinking and planning ahead. In advanced training as well as in day-to-day life, there are times when we’re focused on a complex and/or distant end goal that our dogs simply aren’t capable of seeing. This can be a source of much interspecies miscommunication – and frustration. And it’s another reason why it’s important to break training work up into a series of discrete steps that make sense to your dog.
- A well-designed game rewards effort along with skill. This is another place where we commonly create confusion in our dogs. There’s a big difference between making a sincere effort that puts you into the wrong place and deliberate defiance or misbehavior. As I commonly remind my clients, being wrong is not the same as being bad – and the two absolutely should not be dealt with in the same kinds of ways.
- A game needs to provide players with timely, frequent and clear feedback. Do I need to clarify how this ties into dog training? I hope not. (Although this idea does tie in nicely with my recent post on goals, learning and the emotional regulation of behavior).
- It is vitally important that a game includes some element of surprise to bring excitement into play. Many trainers focus on the importance of surprise in using jackpot rewards to maintain a dog’s interest. While jackpots can be valuable, we also need to incorporate suprise in a less obvious way- through the use of contrast. Contrast allows us to give the dog a way to compare one thing to another in a way that is simple for him to figure out. Contrast is an enormously valuable tool because it lets us tell the dog whether he should focus on sameness or difference in a given situation. It can also help show a dog which features he needs to focus on and which he can safely ignore. This is vitally important in most complex problem solving exercises.
- A game provides players with windows of enhanced attention. This state of enhanced attention or being completely involved in an activity simply for its own sake is sometimes referred to as flow. When you’re in the flow state you engage all of your physical and emotional resources to act and learn. Flow is important in play because it’s a very strongly intrinsically rewarding state of mind. I believe that humans and other animals have a natural play drive because the flow in play is intrinsically rewarding. A good training exercise should provide you and your dog with these ‘windows of enhanced attention’ – and leave you both wanting more.
- Games are interactive. Team-mates and opponents play a vital part in games. Dogs and humans are social creatures and competition and collaboration are often more rewarding to us than cash or treats. I see this in Audie who works mostly just for the reward of interacting with me. I rarely use treats or toys when I work with him because praise, petting and the opportunity to collaborate meaningfully with me are what the boy lives for. Though he also seems to love the competitive rush he gets from chasing (and sometimes catching) squirrels and other prey.
I thought it was interesting that while Chatfield brought up the importance of risk and loss in creating a good game he left that idea off the list. We seem to be developing such a strong (and in many cases, irrational) distaste for fear, stress and other kinds of aversives in today’s world that many people seem not to be capable of seeing the important and necessary part they play in our lives. Without yin there is no yang. If we could erase all aversives from life – joy would disappear too.
A really great game is addictive (though not always in a good way). Really great dog training should be addictive too, so if you and your dog haven’t become addicted to the work you’re doing, take a few tips from game theorists and get lost in the flow.
Entry filed under: dog training, dogs, science. Tags: cognitive science, game theory, play.
1.
Rick | November 4, 2010 at 5:41 pm
Very interesiting post! I’ve never thought of training in that way, but it makes perfect sense. My dogs both work just as hard for treats as they do for the opportunity to play. Got my wheels turning!
2.
Linda Kaim | November 4, 2010 at 5:48 pm
Interesting segue to some of the things showing up on the lists these days, especially regarding dog training. I just did a revision on a post I wrote a couple years ago about chewing that this ties in quite nicely with. I love TED!
3. Tweets that mention Dog training as gaming « Smartdogs' Weblog -- Topsy.com | November 4, 2010 at 6:16 pm
[…] This post was mentioned on Twitter by Jeannie Morris and dogtraining_101, Janeen. Janeen said: Dog training as gaming: http://wp.me/p7sOv-2mC […]
4.
JackPDB | November 5, 2010 at 8:58 am
A fantastic way to look at it. Game-design theory has been very useful in modifying personal behavior in humans — in weight loss and exercise, for example — and with appropriate adaptations for what we know about the way that dogs think and learn it could be a huge tool for training.
– – – – – – –
Jack@PDB
dog beds and more
5.
Catherine Nelson | November 6, 2010 at 2:34 pm
I’m mildly surprised that you do not know that I gave a poster presentation on this subject at the second Canine Science Conference this summer in Vienna.
6.
SmartDogs | November 6, 2010 at 2:50 pm
I’m more than mildly surprised that you did not post information on your presentation on many of the dog-related lists we’re both on! I spoke to Jill about your experience and asked her for information on how to get a copy of the proceedings or other information but haven’t heard back on that.
Send information on on your presentation and I’ll be happy to post it. Or – feel free to slice me publicly into pieces here in the comments.
7.
LabRat | November 8, 2010 at 6:09 pm
Speaking as one who is more gamer than dog trainer, I think I can say with a fair degree of confidence that Chatfield left risk and loss off his list because game designers have been trying to get this right for decades- especially in the MMOGs who can create a viabe ongoing business model for years (11 in the case of Everquest)- and have far more of a track record of failure than success in this mode.
The most obvious example is death penalties, of which there’ve been dozens of variations, from lost experience to having to make a corpse run to the possibility of losing items if the model includes the ability for other players to loot player corpses as well as enemy corpses.
In theory the risk increases the reward, but especially when you’re designing content that’s complex enough to create that satisfying challenge, what winds up happening in effect is the penalization of risk-taking and exploratory behavior, which is VERY frustrating for the player. If a player incurs a substantial penalty every time he or she tries something new or risky, the result is either you grind until you can eat the costs without too much devastation (boring!), or you stick only to safe activities (equally boring).
*makes a note to self to possibly expand on this later*
8.
SmartDogs | November 8, 2010 at 9:08 pm
Speaking as a dog trainer with no experience in gaming I wonder if the problem with risk penalties in gaming is that they’re delivered only as after the fact punishment with no meaningful input. I dunno about gamers, but when one deals with living breathing dogs, retributional punishment is not terribly effective.
Perhaps if games were designed such that when you incurred a negative penalty you also concurrently received some piece of constructive advice on how to avoid that penalty in the future, the value of punishment (and your interest in continueing the game) would be greatly enhanced. There’s a world of difference between rubbing Fluffy’s nose in poop and scolding Fluffy, in the act, for pooping in the house then taking him outside and praising him for pooping in the right place.
9.
LabRat | November 8, 2010 at 10:44 pm
Part of the problem is that in most cases, learning that portion of the game is itself part of the design/intended fun; you learn by failing and figuring out what works and what doesn’t. In the particular case of MMO raiding, teamwork is also a significant part of the equation- getting ten or twenty-five or in the old days forty people to do the right thing at the same time to survive the encounter.
Come to think of it, in such games the other players actually provide the risk/loss factor; no one wants to let their teammates down, and no one wants to carry a player who’s not doing their best to help the team succeed.
10.
SmartDogs | November 9, 2010 at 7:09 pm
Interesting.
I think that teamwork and social connection are a bigger drive in training dogs than a lot of people realize. \
I’m very lucky in that my pack spends 99% of its time offleash and unfenced (well, ok, we do have a fenced backyard but it’s full of free range chickens, tossed out kitchen waste and marauding squirrels so as far as distraction levels go, it may as well be unfenced). Anyway the great majority of the training work I do with my dogs doesn’t involve leashes, treats or ecollars. I use body language and verbal hot and cold cues along with the occasional “OMG YOU’RE AWESOME” or “DON’T MAKE ME KILL YOU” cue. And they love to work.
The training we do goes well beyond simple come, sit, stay, cut that out stuff too. The dogs herd chickens, kill vermin, find and carry tools, help me get dressed and undressed and do a host of other handy tasks around the place. And I think they do it because they want, no need, to feel like they have a purpose in life and are part of our extended human/canine ‘pack’.