Cows, Goats, and Raccoons Oh My!
Check out these fun training videos and step by step tutorials!
Teaching fish to "touch."
Want to learn how to train your cat, dog, or even chickens, fish or coworkers? Then read on and learn all about a few easy, safe, effective and fun methods to use on your human and animal friends!!
For this post I thought it would be fun to talk about a few different principles of learning theory in the context of how I trained the Koi and Shubunkin fish in my pond. Training fish, that sounds crazy right? Well even animals such as bees can learn if you find the right motivation and means of communication. Unfortunately dogs, horses, and circus animals have in the past been subjected to aversive training methods such as positive punishment (jerking on a dog's neck when they pull on leash), negative reinforcement (taking away painful bit pressure once a horse performs the behavior you want) and flooding (forcing a fearful dog to go to a crowded festival). With animals such as fish, dolphins and lions you can't use a choker chain or yell at them with any success. It would be dangerous, isn't as effective, and can come with a sleu of negative behavioral consequences such as increased fear, anxiety and aggression. So what methods should be used instead? The most effective method is called positive reinforcement and it even works on surgical residents. Positive reinforcement is a type of operant conditioning and it means learning voluntary behaviors by consequence. To reinforce means to increase the likelihood of a behavior occurring again in the future. Positive indicates that we are adding something. So logically, to increase the likelihood of an animal or person repeating a behavior, you must give them something they like such as food, play, or praise (or for people, that silly thing called money).
There are also two other techniques that I used with my koi and shebunkin to help them feel comfortable in my presence and learn that I not only was safe, but also a good person to be around. This was essential before starting operant conditioning which involves voluntary behaviors such as swimming towards my hand and eventually targeting aka touching my finger tip.
The first technique I used is called desensitization. This is the process of helping and animal learn slowly over time that something that once scared them is actually nothing to be afraid of. Animals are born with an inherent neophobia, or fear of new things. Anything we don't expose them to as babies (puppies 3-12 up to 16 weeks and kittens 2-7 up to 14 weeks) will likely be scary to them as adults. When I first introduced my fish to my pond, they were presumably past their sensitive period for socialization (the time when their young brains are like sponges and they can learn all about what is safe and not safe in the world). They had a significant flight distance (how close I could get before they would swim away) and they were fearful of the approach of people. For quite some time I did not even know if the Koi were still alive because they never came out when I was outside. I saw the occasional twitch of a Lilly leaf that would give me hope that they were still hiding in the plants but I was never sure. Over time I was able to sit by my pond and after several quiet minutes of my sitting like a statue the fish would come out and grab some fish food from the opposite side of the pond. This process of me slowly allowing the fish to become accustomed to my presence, without causing fear, is called desensitization. Over time they became desensitized to the sound of my voice and my movement near the water.
These videos show the initial flight zone on my fish and the dramatic improvement they have made.
As I mentioned earlier, the technique of flooding is not a very friendly thing to do to an animal you love. Flooding means to expose your animal friend to the full force scary stimulus in hopes they will overcome their fear. This is the equivalent of locking an arachnophobic person in a small room full of spiders. Terrifying right!? While some people might learn that spiders are ok after a day of not having any crawl on or bite them, others will have a bad experience or just enter a state of mental shutdown called learned helplessness, a state that some mistake for calmness or "submission". Not so fun right? The technique of flooding is not recommended because it is seldom effective and is inhumane. Desensitization on the other hand aims to keep the scary stimulus far enough away or quiet enough that it doesn't not elicit a fear response for the animal. For most animals that means that they can still eat treats readily and are showing calm and relaxed body language.
The next technique I used is called classical conditioning. This technique involves pairing a neutral stimulus with food or something else that induces an involuntary positive emotional state. With the koi this meant shaking a food pellet jar and then immediately dropping a pinch of food into the water. Over time they learned through repetition that the sound of the jar meant food was coming.
This can also be used for classical COUNTERconditioning, which involves changing an animals involuntary response of fear or anxiety to a relaxed and calm emotional state. This is the technique that is used to change an animals response to a muzzle from anxiety about a scary or painful procedure to excitement about a Kong filled with meat. We will use this technique in this summer's basket muzzle training challenge.
If I remained still and calm they would come up to the surface and gladly enjoy the food. But they were still not completely comfortable with me. I eventually added some shubunkin fish to the group, they are known to be more outgoing fish. This social facilitation helped my koi relax and allowed the processes of desensitization and classical conditioning to work even faster and more effectively.
After the fish were readily coming up to the edge of the pond when I was sitting there, I began to use the technique of operant conditioning that we talked about earlier. I wanted to target train them, which is an easy behavior to train with fish and is actually my favorite behavior to teach my clients and their pets. It involves teaching an animal to touch their nose, paw, beak etc to a finger tip, hand, target stick or other object. It is often easy to "capture" with cats and dogs who are naturally curious (especially when your finger smells like chicken) and chickens who naturally peck at things. You do this by positively reinforcing your pet the moment they perform the behavior. Using a marking device such as a clicker, whistle or verbal "YES!" helps improve the efficiency of the training process and improves inter-species communication. This clicker allows you to mark the exact second that the animal performs the touch behavior. Over time of using positive reinforcement, the behavior will strengthen and you can eventually add a cue word so that you can ask your animal friend to perform the behavior.
With my fish I used a technique called free shaping because I didn't expect them to just happen to touch my finger. This must also be used for target training fearful dogs or for teaching unnatural behavior such as backflips. This technique involves rewarding closer and closer approximations to the final desired behavior. I started by rewarding the fish for approaching my hand. I then rewarded closer and closer approaches and finally a quick gentle touch of their mouths to my finger tip. This happened over multiple short (2 minute) sessions over the course of weeks. We are now at the point of them firmly bumping my finger with their mouths. I may switch to a target stick in the future and train them to make circles in the water or something like fish agility. There is so much potential for what we can do together!!
Here is a video of Rorschach performing the touch behavior. Notice my treat delivery. At first I had to be cautious to make very slow and minimal movements when dropping treats in order to not frighten them.
Sam the Calm learns with Clicker Training!!!
This is a fun memory from one of my rotations during vet school. This is Sam the calf showing off his newly learned behavior of stepping up on a bag of compressed bedding. We used positive reinforcement, clicker training, operant conditioning and shaping to achieve this goal. Read on to learn how to train your pet or patient to perform a behavior and at the very end, what I could have done better in this video.
Step One: Sam learned that a click (can also be a verbal "yes" or whistle) equaled a food reward. I did this by clicking then immediately giving him a few pieces of grain. After 10-20 repetitions most animals will learn this association. You can ensure they have learned it by clicking and watching for their response. If they orient to you or look for the reward, you've accomplished your goal. If not, keep clicking and treating until they make the connection.
Step Two: Shaping!! Outline the behaviors you will reward in a stepwise manner. I didn't just expect Sam to step up on the bag on his own, it wasn't a normal behavior for him! So I had to create approximations towards the final behavior. First was any movement of a front leg towards the bag. Next was a step towards the bag. After that was touching the bag with his hoof. I then increased the criteria and only rewarded for placing his hoof on top of the bag, then for placing weight on the bag with one limb and finally for placing both hooves on the bag.
Step Three: Reward each step of the behavior. I started with the first approximation and rewarded Sam with a few pieces of grain until he was regularly moving one limb towards the bag. Once he was reliably and quickly performing that step I increased the criteria to him actually taking a step towards the bag, and so on.
Step Four: End your session after 5 minutes, or sooner, if you notice your pet is not as interested or is no longer having fun. Start back up again later in the day or even on another day.
Step Five: Start with the previous step than you ended on, with Sam that would mean rewarding for any movement toward the bag. Don't worry, the previously learned steps move quicker during each session.
Step Six: Increase the criteria until you have the final behavior. Give better rewards (higher value treats such as grain vs hay) for really good approximations and for difficult steps (placing weight on the bag).
Step Seven: Once your pet is reliably performing the behavior you can add a cue word, such as "step up." Say the word right before you know your pet will perform the behavior and reward once they have performed it. Using the same process of shaping outline above, you can start to cue the behavior from greater distances and in increasingly difficult environments (other people or animals around etc).
Remember to work at your animal's pace and always have fun!!
A few things I could have done better in the video:
You should click at the exact moment your pet performed the desired behavior, otherwise they might learn the wrong thing. Communication is critical! The click helps you mark the moment of the correct behavior faster than you can deliver a treat.
Sam probably did not enjoy the pat on the head I gave him. Some animals enjoy petting but we shouldn't count on this to be considered a reward. Praise is nice sometimes but treats, toys and play are usually better!! To fearful or under socialized animals, a pat on the head may even seen like a punishment, decreasing the likelihood of them repeating the behavior!
Pay attention to your own body language and what you say. Your pet will pick up on everything you say and do and might get the wrong message if you aren't paying attention.
Using a smaller bag would have been easier on Sam's body. Choose behaviors and props that are compatible with the species, age, and anatomy of the animal you are working with. A Border Collie might be able to learn a backflip through shaping but a Great Dane could never perform such a behavior.