Which Treat Value Is BEST?
There’s so much variation in how food reinforcements are used that we have to look at the objective of the training.
Generally there are two styles of training. We can primarily work in classical conditioning or we can primarily work in operant conditioning. It is correct that Pavlov is on your shoulder and Skinner is on the other. Hopefully we have a plan when training. Our plan should be clear as to which strategy we are using in the moment. You can’t simultaneously be marking behaviour and paying to create only an association without looking for behaviour. Either classical or operant should be front and centre.
Classical conditioning is all about emotions. I want the dog to like something. I want them to like the sight of nail clippers, like the sight of the neighbour’s dog, like the bath. There isn’t a skill that I am creating (yet).
Operant conditioning is all about skills creation. I want the dog to sit, to lie down, to come when called. My concern is primarily teaching the dog to do something. If done well, the dog will also learn to like it. But the main goal is skill creation.
Which treat is best for the task is determined by which of these strategies you are using.
If you want to focus on creating associations with classical conditioning, then use high value food. The fastest route to robust, strong conditioning is supremely high value food, paired tightly with the trigger.
High value food means stinky, juicy, the best ever… RARE. It’s the rarity of the high value food that wakes the brain up. The rare, special food slaps the brain and says, “You really need to figure out why this happened.” You might create positive associations with lesser value food. It’s harder to do and much slower – if you get it to happen at all.
Think of regular treats like nickels – a low value coin. Most of us have found loose change on the ground. It’s a nickel. It’s nice, but not memorable. Regular dog treats are like nickels. Kibble? Think pennies. Your ability to get an association is so compromised you might as well not even try.
Imagine finding a hundred dollar bill or a wallet. Most people can remember the event for decades. The rarity, the value, wakes the brain up and says, “Remember.” Succulent meat is like a hundred dollar bill. It’s memorable.
For people who become nervous about fat dogs or too much junk, remember that in classical conditioning, fewer repetitions with longer breaks between leads to more robust conditioning. The long breaks mitigate the overall quantity of food used. Think consistency, rarity and value. If you do it by the books, the breaks reduce the quantity of food used overall. People who over drill, without breaks, should be concerned. But skilled trainers don’t do that…so nothing to worry about.
Operant conditioning is different. There are certain hallmarks of good skills training. We want a high rate of reinforcement – paying quickly and frequently for responses. Good operant training is a rapid fire series of yes/treat, yes/treat, yes/treat, at least during initial training.
It’s common for people to work with a dog for several minutes, at this rapid fire rate, and then end the session. You do not want breaks between repetitions. Rather you work in rapid fire sets during skills acquisition.
For this type of training, you want the, “optimal level” of arousal. This comes from Yerkes-Dodson law. If arousal is too low, performance declines. If arousal is too high, performance also declines. The brain becomes so focused on the wow factor that the dog isn’t really focused on the task.
The main takeaway is that the optimal level of arousal varies. Some cognitive tasks do better with lower levels of arousal. Some demanding tasks require more arousal. There is no set answer. Which treat? It DEPENDS.
However, you do want to be in-between extremes. It’s the “Goldilocks” treat. Not too high. Not too low. Just right. It can vary a bit from dog to dog from task to task.
The basic rule of thumb is use high value for classical conditioning. Use optimal value for operant conditioning.
Beyond this, some other things may need to be considered.
Some skills require larger amounts of effort than others. A close range hand touch is easier than a long distance recall off a squirrel. When asking an animal to do extra work, pay better.
There are times when we may not be prepared to feed immediately after the “yes.” The longer the delay between the yes and the pay, the less perceived value that food has. If you have to pause or wait to get to your food, then level up on what you give. That time lag comes with a price to pay. Or better yet, be prepared next time.
Need your dog to focus mentally? You may need to level down a little. Keep the dog’s arousal level lower.
Finally, some training strategies mash classical and operant. Autoshaping does this. It starts with classical conditioning, then swaps over to operant.
We start with Pavlov, and Skinner shows up to the party. Here’s where we have to look at each training plan and think. There is no magic treat. The answer is always “it depends.” How difficult is the skill? How much effort does the dog have to expend? Are you working with classical or operant conditioning in this moment?
Do avoid chasing the magic treat. Magic lies in technique. However, you do want the right treat for the task. So what’s the BEST treat? The one that suits the task. Go big on classical conditioning. Optimal value for operant conditioning. From there, adjust based on common sense and a thoughtful training plan.
Before someone says it, “What about the toys and the jackpots?” There’s a special place in a training bag for one, and the other is best thrown out. Both deserve their own blogs.