Mr B the Cat Learns to Weigh Himself

Above: Mr B gets on the scale and looks up at me while he begins to sit.

The above video shows Mr B getting on the scale. I am not sure why it is not possible to hear all the clicks except the one where I clicked the very beginning of the behavior.

I found that trying to manage the video camera, clicker, and container of treats while giving an SD (especially before coffee) was not conducive to astute training–but the benefit is you can see him on film.

Mr B took about three days to learn the behavior. I didn’t really track the number of sessions.

The first day was just click and give a treat. I also captured a “sit” behavior in the process.

My training sessions tend to be very short and frequent.

I stop when the animal hits the success level we are aiming for and bonus the animal with extra treats, attention, or play.

The next session after we progress tends to be a review with a repeat up to the desired behavior we left off on and maybe a couple of reps to make sure the animal understands that the new level of behavior is the one I want.

Depending on the animal, I might ask for a little more.

For the scale, Mr B sniffed it and I clicked and rewarded.

Then after he did that consistently, I watch for the next criterion I wanted–which is movement of a front paw.

He got that quickly and put the paw on the scale.

The first paw took the longest to train. Once he got the click for two paws on the scale he consistently offered it without going backwards.

Okay, stay with me while I digress to illustrate some things.

Mr B then started variations to check if he was on the right track.

He would step on the scale with two paws and lay down, the next time he would sit with his rear off the scale after putting his two paws on it, then he might walk around the scale before putting his paws on it.

That is what you want to see when training.

Every animal is different and some might “get” that the movement onto the scale is what is the ultimate goal and they will continue to move onto it more quickly.

Some animals, like Mr B, will experiment to see if other things will work too.

Another more familiar explanation can be done with dogs that are learning not to jump on a human.

Some will replace the jumping with a sit right away–because it has been reinforced and the jumping is not–so there is a good chance that will be rewarded.

Their test works and that is the end of it.

However, some dogs will jump up and put one paw on the person, or perhaps jump and touch another object nearby with the paws (or paw), next they might just jump up in the air in front of the person without touch anything, and finally they will offer another behavior in replacement of the jumping.

Mr B was doing the same type of thing.

He is also not a fast acting guy–unless you tell him to get out of the kitchen!

So once Mr B began to consistently get all four paws on the scale and sitting down for the weighing, I began anchoring the “scale” command–which means making sure he understood what was wanted.

The anchoring worked with asking him to perform the “scale” behavior. If he offered any other variation or did not respond in a specific window of time, the session ended.

This means his opportunity to receive reinforcement ended.

This taught him to respond with the behavior and to respond within a window of opportunity rather than on his time schedule.

What then happened is that he would get onto the scale without offering other behaviors.

Also, he began getting on the scale faster because slow responses no longer received reinforcement.

I knew he understood that he was to get on the scale in a certain location but would he do so in another?

Some animals can do this easily while others have difficulty with it.

Finally I began to work the behavior in other locations and with another scale and that is when I discovered that the variable schedule of reinforcement was too much with all the other changes.

If you look at this body posture in the video you can see how his whiskers curve forward in that happy manner cats have when they are affectionate or playful.

In this “action video” (LOL) he walks past the scale to see if I will reward it. He is also purring which I was sorry you cannot hear.

What I found in changing the location (above we are working in a third location) is that he did not do well on a variable schedule (used in this video).

My final assessment through two weeks of work with Mr B is that he is one of those rare animals that needs to have consistency.

So, he is back to a one-click-equals-one-treat model and now the owner is learning to work with him.

I hate using consistent schedules as a trainer–but unless Mr B is on a consistent schedule he will offer a behavior he knows is a solid one for reinforcement (in this case, “sit” and wait) instead of what is asked.

Part of this might be because the SD is too similar between the “sit” behavior and the “scale” behavior but I think it is more about his process of manipulating people that has worked so well for so long.

I also considered if I attempted to introduce too many variables too early which I ruled out.

When an animal does not perform as you would like, it is good to step back a reassess the situation.

Training is communication and so if an animal does not respond in the way you would like–redirection in training might be needed.

In his entire repertoire through the training he now knows:

  • sit when asked,
  • sit and wait before being fed,
  • come and come eat,
  • stay aka wait (marginal),
  • get out of the kitchen, and
  • scale (get on the scale and sit).

Above: Mr B as he starts to sit on the scale.

Above: Another view of Mr B on the scale in location #3.

I’ll have to ponder if I left anything out but now I am slowly working the owner into managing his behavior with the clicker.

Many people have a hard time coordinating the clicker, food, commands, and (in this case) a syringe all at one time.

She did good so I gave her a mini milk chocolate bar as a reward!

I have more to say but that will be in antoher post.

Back to Basics & Forward Again: Training with Mr B

Above: Mr B midway through his training to step and sit on the scale.

Training with Mr B, a siamese cat from a rescue facility, has been filled with interesting variables.

First, when he first “got” training he began purring and purrs through every training session.

When you start an animal training program the end goal is established and then most people do a “training plan” or outline the steps to get to the final behavior.

The lesson plan looked like this:

  • train Mr B to understand the marker,
  • get a stationary behavior away from the food bowl,
  • maintain the behavior until released with a word,
  • look at an object,
  • investigate an object,
  • step on the object,
  • move another paw onto the object,
  • move a third paw onto the object,
  • move all feet paw the object,
  • sit on the object (scale).

To accomplish the training I use a noise making device to “mark” the behavior.

Most people use a verbal maker–usually the word, “good” so that the animal knows the behavior is the right one.

The marker signals that something good is coming.

A reward can be praise, petting, play, food, or something else an animal enjoys.

Most people make the mistake of thinking that there always has to be a food reward. This is incorrect.

A reinforcer just needs to be something an animal enjoys.

When you use a marker, such as a clicker or a whistle, it signals the precise moment the animal is engaged in the right behavior–so the behavior is marked.

A recent study indicates that clickers might be better than using a voice due to previous conditioning to voice intonations etc.

In Mr B’s case I have been using a clicker as an IOU for the reward. The “click” means that is right and a reward is coming.

His primary reward is food but the clicker becomes a secondary reinforcer (reinforcing in itself) through its use.

Most animals do better if you go to a variable schedule of reinforcement versus a consistent one but you have to be skilled in order to use variable reinforcement.

Let me explain:

  • In the beginning you must always do a consistent schedule. This means one click gets one treat.
  • Later, when you introduce a variable, the reward might come on a second, third, or first click but you vary it each time.

There are other types of variables you can use (like the time interval I mentioned as the “window of opportunity) but the most common is the variation on the delivery of the primary reinforcer or reward.

Mr B understood the concept of training quickly and began purring because he liked training.

He invents games and so training school was really to his liking–plus he could now manipulate my behavior when all his previous attempts had failed!

I started him on the training program to differentiate feeding away from the food bowl (where he is frenzied) and into other pursuits.

He is also in a weight loss program due to his gross obesity and is slimming down nicely…which is why his normal kibble is being used as the rewards–no extra treats for this big boy who was having trouble breathing and moving around when we started.

Once he understood the one click equals one treat, he could experiment and then get rewarded for it.

The first thing I did was teach him to sit away from the food bowl outside of the kitchen.

He got that quickly.

This is replacing an undesirable behavior with an acceptable one.

Previously he would run into the kitchen in a frenzy thinking he would be fed.

He would also bat and bite if he was not fed…drawing blood from the owner is not acceptable.

Now he knows that if he wants to be fed that he must go sit and wait…and so he does.

BUT what he also learned is that he should be on the carpet not on the hard flooring.

This was clear when I began training the scale in the kitchen on the hard floor and then transferred this behavior to another location while we were working on variables–but only after he learned the “step on the scale” lesson.

If you do anything the same way at the same time it conditions an animal to that behavior at that place and time.

Variables add a variety and take away the cue (aka discriminative stimulus or SD) that was established by accident previously.

Before the kitchen and presence of the owner, sometimes at a specific time, was the signal for feeding.

Now the SD is the release word followed by the phrase, “come eat” or “come.”

When you have a pattern of behavior that has been reinforced for years, it is harder to extinguish than something more recent.

Mr B has to still be reminded with an, “out of the kitchen” command but he is getting better daily.

Anyway, Mr B is very diligent and persistent. So, he did not do well at first with the variable reinforcement schedule.

When I introduced the variable he stared at me.

Then, unsure what to do, he just sat down. Which is a behavior that works for him in many cases and the first one he learned.

So, I went back to a consistent schedule and raise the criteria for the behavior instead of using a variable to mark the various stages of behavior that he was doing correctly.

That means he had to do more to get a click and then a reward.

One foot on the scale was no longer enough for a click–he had to put two, and then he was rewarded at that level a few times before three feet were required.

Some animals do better than others with training concepts and since this is so new to him I had to step back in the training process for it to be clearer to him.

Part of the problem was that instead of focusing on training during the initial variable, he began to see how he could get rewarded rather than focusing on the training.

He has been a master of manipulation so he moved through stages and my interpretations:

  • Mr B staring at the clicker for long, hard durations: “If I stare hard enough and will it–she will click.”
  • Next, staring at the counter and food bucket, “If I can successfully jump onto the counter, I can get ALL the food rewards in the dispenser.”
  • Finally, “It is easier to experiment around the object than to try and focus on other things.”

The training of the scale was a fun process and now we are using another scale.

Variables can confuse an animal sometimes so moving locations, changing the target object, and other activities can help establish the behavior and get it in other circumstances.

I’ll talk more about that next.