In 2009, shortly after I started training animals on a more ongoing basis, one of the first concepts that I learned from fellow animal training colleagues was the concept of a jackpot. A quick google search yields the following definition “a large cash prize in a game or lottery, especially one that accumulates until it is won.” For animal training purposes the following definition is more commonly used “giving a dog a really big reward, often a large number of treats, all at once. It is usually reserved for a breakthrough moment or a desired behaviour that the dog only occasionally performs” (Schwarz, 2016). This is the first definition that I have been exposed to and the idea behind it is that by offering a large reward the behaviour that preceded it is somehow more likely to be remembered better and repeated in the future.
My first contact with this concept was in a context in which the animals were trained using a bridge or bridging stimulus (e.g. a click from a clicker) for both learning and maintaining known behaviours. Known behaviours do not necessarily need a bridge to be maintained, but that is a topic for another discussion. The theory goes that if Fido gets a click and three pieces of food for a perfect sit and a click and only one treat for a decent, but not perfect sit, he will be more likely to do perfect sits in the future.
We might be inclined to assume that an animal will be tuned in to the magnitude or quality of the reinforcer in a way that makes some variations of the same behaviour more likely to be repeated than others. However, is that really what happens? Does the animal actually remember the topography of that behaviour better because she got 5 or 10 food treats after the click instead of the standard one treat? In this text we will explore the function and the best use of jackpots in animal training by relying on the opinion defended by animal training professionals.
Jackpots are commonly used as a special reward for excellent behaviours. They are an attempt by the trainer to capitalise on a behaviour (or a variation of the behaviour) that the trainer particularly likes. This seems to be based on the assumption that a particular special reward will increase the chances of similar responses in the future. For example, Kazdin (1994) mentions that “The greater the amount of the reinforcer delivered for a response, the more frequent the response will be.” However, research confirming this rationale regarding animal training, with a bridging stimulus, is hard to find. If you click and pay more than one treat there are a few things happening that are helpful for your training program, but those things might differ from the traditional interpretation of jackpots. So, let’s start by having a look at some quotes by international references in the world of animal training and how they contrast with the common understanding of jackpots in animal training.
“A jackpot serves to charge up future performance but does little to communicate to the animal that his previous actions were special.” (Reid, 2012).
“If you click, and then deliver the treat afterwards, an especially large, numerous, or wonderful treat is no different from any other treat, in terms of its ability to reinforce behavior.” (Pryor, 2006).
“Click means treat is coming. If the treat is sometimes a kibble and sometimes chicken, sometimes small and sometimes huge, that's fine, it keeps your clicker nice and strong; but it doesn't tell the animal anything different about the behavior.” (Pryor, 2006).
“When it comes to training a new behavior, it's rare that a jackpot would work in having the dog repeat the jackpot earning behaviour.” (Fisher, 2009).
“Jackpots make the giver feel good, but they interrupt the flow of training and focus the dog on the food, rather than the task. (…) Overall, it's clarity of criteria and a consistently high rate of reinforcement that leads to a solid behavior.” (Alexander, 2006).
As you can see from these quotes, there are several animal training specialists suggesting a different interpretation for what really happens when we bridge a behaviour and offer a bigger reward after. Let’s explore their rational and look into what really happens when we use such an approach.
Clicking and paying several treats can increase the value of the clicker. Given that there is some variability regarding what happens after the click, that stimulus (the click) remains nice and strong from the animal’s perspective (Pryor, 2006). When a large reward is offered in the beginning of a training session it can motivate the animal and increase interest in the task. It can make the animal increase its activity level and it can trigger subsequent variable behaviour (Fisher, 2009). So, as you can see offering several rewards after the click can actually accomplish a few handy things. These are some of the things that happen when we use the traditional interpretation of jackpots in animal training. Now let´s have a look at a few things that do not necessarily happen.
Clicking and offering several treats does not provide the animal with any additional information about the behaviour that she just did. Offering more than one treat after the click is also unlikely to strengthen a behaviour over another; what ultimately accomplishes that goal is when you choose to use your clicker: clicked and rewarded behaviours are more likely to occur in the future when compared to behaviours that do not get a click and reward. For many practical situations in which we are training our pets, offering several treats after the click simply tells the animal that sometimes it gets more treats than usual (Pryor, 2006; Farricelli, 2014).
For her Masters Thesis, dog trainer Elizabeth Kershaw (2002) conducted a dog training experiment that tried to measure the effects of magnitude of reinforcement after the click when dogs are learning a new task. She had two groups of dogs learning to touch a cone with their nose and with their paw. One group progressed through criteria with one click and one treat all the time (constant group), whilst another group progressed through criteria with one click and one treat most of the time and an occasional click and delivery of larger reinforcement amounts (jackpot group). Overall, significant differences in performance between the two groups could not be detected.
Kershaw (2002) also mentions that using a jackpot to reward a breakthrough when the dog is learning a new behaviour might be a better option when it marks the end of the session. Using a jackpot halfway through a session, when you intend to continue immediately might be counter-productive. This can cause the dog to not be able to associate the larger reward received with the intended behaviour because a longer period of eating can disrupt the learning flow.
Fisher (2009) offers some interesting additional considerations about the traditional use of jackpots in dog training. She mentions that the longer it takes for the animal to eat the reward, the more the behaviour might be subject to memory decay (a disconnect between the reward and the behaviour that caused it). Instead of strengthening a behaviour, jackpotting can elicit the dog to follow it with a different behaviour. For speedy learning, a short time span between reward and the next repetition might be ideal and the training will progress faster with a rapid rate of reinforcement (many repetitions, each resulting in quick to ingest treats).
So, what if we still want to incorporate jackpots in our animal training sessions? What are the properties of a real jackpot? A real jackpot should function as an event marker (no bridge required) and almost startle the animal, it should consist of an unusual primary reinforcer and it has to make that behaviour more likely to happen again. A jackpot, when used correctly, should be an astonishingly big reinforcer, delivered contingently. The jackpot has to appear while the animal is doing the behaviour, not afterwards (Pryor, 2006). If the reward is offered after the behaviour we enter the realm of the non-contingent reward.
A non-contingent reward is a reward that is offered after the behaviour has occurred as opposed to while the behaviour is occurring. A non-contingent reward is not necessarily associated with any specific behaviour, it can be used to encourage the animal in a given situation and it can increase motivation (Pryor, 2006). Pryor offers the example of a slot machine jackpot, which is always delivered contingently (while you are playing), so that the act of playing is heavily rewarded. Compare this with a situation in which you play the slot machine, you then go out for dinner, then you go to a music concert and finally you return to your hotel bedroom to find a huge sum of money on your bed. This is the same amount of money as the slot machine jackpot, but this time it was not delivered contingently. Hard to say which behaviour would increase in this case... Perhaps the going back to the hotel, but not necessarily the slot machine playing.
To conclude, when we click and offer a bigger reward (say 3, 5 or 10 treats) we might be maintaining the animal’s motivation high or increasing it, which means that the following behaviours can be more enthusiastic. The bridge is also kept nice and strong, because there is some variability in what happens after we use it. This procedure is not a real jackpot though. A real jackpot should be totally unexpected and almost startle the animal, it should be a rare event, and you should not click the behaviour (clicking and offering lots of treats will make changes in the connection between the bridge and the reinforcer; not in the behaviour that you bridged).
Animal training is a fluid technology that is constantly being updated. 20 or 30 years ago we were probably looking at jackpots in animal training in a different way than we are today. A few years from now, today’s knowledge might get updated and refined. That is the beauty of the animal training world and I can’t wait to see what the next chapter brings us.
Alexander, M. (2006). Should You "Jackpot" Outstanding Responses? Clickertraining.com. Retrieved 1 May 2017, from https://clickertraining.com/node/632
Farricelli, A. (2014). Using Jackpots of Treats in Dog Training. hubpages. Retrieved 5 May 2017, from https://hubpages.com/animals/Using-Jackpots-of-Treats-in-Dog-Training
Fisher, G. (2009). The Thinking Dog: Crossover to Clicker Training (1st ed.). Wenatchee, Wash.: Dogwise Pub.
Kazdin, A. (1994). Behavior modification in applied settings (5th ed., p. 147). Pacific Grove, CA: Brooks/Cole Publishing Company.
Kershaw, E. (2002). An evaluation of the use of magnitude of reinforcement, i.e. “jackpot” rewards, during shaping in the training of pet dogs. (MSc). University of Southampton New College.
Pryor, K. (2006). Jackpots: Hitting it Big | Karen Pryor Clicker Training. Clickertraining.com. Retrieved 29 April 2017, from https://clickertraining.com/node/825
Reid, P. (2012). Dog Insight (1st ed.). Wenatchee, WA: Dogwise.
Schwarz, S. (2016). AgilityNerd Dog Agility Blog : Better Jackpot Rewards. Agilitynerd.com. Retrieved 4 May 2017, from http://agilitynerd.com/blog/dog/training/Jackpot.html