Posts filed under ‘Carly and Louise’
As I mentioned in a previous post, I decided to change Louise’s trick from teaching her to jump over an obstacle (ruler) to trying to get her to “beg” on her two hind feet at the word “up.” Since Louise stands up so much on her own, all I had to do was capture the behavior by clicking and feeding her as soon as both of her front feet left the ground. I also tried to coax her by holding a piece of food over her head. Thus, I often just gave her the food directly from my hand; however, I also intermittently delivered the food through the tube in the food corner.
I then began adding the “up” command on the very first day of training. I say “up” right before it looks (to me) like she’s about to stand on her hind feet. However, I’m often wrong in my assumption and end up saying it a bunch of time before she actually stands up. As of now, she still is not showing much response to the “up” command. I’m hoping if I just continue to do it over and over, she’ll eventually begin to stand up on command, but at this point I think she is just so interested in the food that I’m holding over her head that the “up” command has little salience. I’ll continue to work on not holding the food to coax her to stand up, and only reinforcing her through the tube in the food corner. We’ll see where that gets us!
Here is a brief video showing the different ways I reinforced Louise, as well as what the trick is supposed to look like!
I will continue to work with Louise as much as possible. However, Erika and I will be starting in on our experiment as soon as possible, so trick training time may be limited!
On Thursday I began trying to teach Louise her first trick: jumping over a piece of wood (aka, a ruler.) I began by simply letting her go up to the ruler to sniff and clicking which she touched the ruler. I had planned on gradually clicking when she put her feet on it, then when she put her head over it, and then when she put her feet on the other side, but Louise was ten steps ahead of me and just began jumping over the ruler without much coaxing.
I had been placing a small piece of food on the other side of the ruler, and once she jumped I would click and she would find the food. This seemed to be working well and Louise continually jumped over the ruler, until she simply…stopped. For a couple of days I couldn’t get Louise to jump over the ruler without some sort of coaxing (holding the food in my hand and luring her over, or just giving her a nudge.) With some help she started jumping again, but she then seemed to become very overstimulated by the reappearance of my hand in the tank.
I believe she also was slightly confused by the fact that she had to jump over the ruler in two different directions, especially when one direction pointed her toward her usual food corner and the other didn’t. Also, since I began teaching the trick with a ruler that stretched the entire way across the tank, after I clicked, she would try to jump back over the ruler to get back to the food corner, which was just too complex for the both of us!
Thus, I’ve decided to try a different, simpler trick. Today I am going to try to teach her simply to do a “begging” stance in which she stands on her back legs to the command “up.” I was hesitant to do this at first because she so often stands on her back legs and sniffs when she’s off task, but I think it is a much simpler trick for not only her first time learning, but my first time teaching. We’ll see how it goes!
Over the past two and a half weeks, I have been using classical conditioning to “clicker train” Louise. The aim is to pair a clicking sound with food, in hopes of conditioning Louise to associate the click with food. In the end, Louise should be under stimulus control. In other words, when Louise hears the click, she should immediately look for food.
We have come a very long way! Our journey began the day Louise and I met and she tried to jump out of my hands onto the lab floor. However, after a week of daily, hour-long play sessions, Louise became much more comfortable being handled.
Once Louise and I were comfortable together, I began placing the food in the tank and clicking when she ate it; however, after only a couple of trials of this method, I noticed that Louise was running up to my hand and not attending to the food. I then perused the blogs of last year’s students and noticed that a lot of them used the method of clicking and then dropping the food in a particular corner. I thus adopted this strategy in hopes that it would work for me.
I began by dropping the small pieces of pellet (about 1/4 of the size of my pinky finger nail) into a corner of a glass tank that was approximately 12″ by 18″. However, this strategy was not successful because the pellets would hit the bottom of the tank and bounce uncontrollably. To solve this problem, I rolled a white 8.5″ by 11″ piece of paper into a tube (it can be seen in my videos) and attached it to the bottom left corner of the tank. By dropping the food through the tube, it always landed in the correct corner.
I then repeated trial after trial of clicking and then dropping the food into the same corner. Eventually, I started clicking again when Louise actually ate the pellet. I felt that this served as an added opportunity for Louise to associate the food with the click, and I also thought that perhaps the second click would inhibit the potential association between the food and the mere sound of the food hitting the bottom of the tank.
In addition to this, I also had to work to extinguish the association Louise seemed to be making between my hand and the food. Louise often began sniffing the corner at the sight of my hand moving toward the tank, as well as to the sound of my fingers touching the paper tube. Thus, I began placing my hand at the usual corner over and over without the click and without the food. I mixed this method with actual trials and Louise began responding less to my hand.
Additionally, the amount of food Louise ate had a large impact on her training schedule. To start, I only managed to get in about 20 trials on average. I then began to withhold food from Louise except for when we were training, and after a few days I realized that Louise needed at least 12 hours between when she was fed and when she was trained. In response to this, she began to eat ravenously during sessions and on good days we could complete between 60 and 100 trials.
Furthermore, I weighed Louise’s training pellets before and after each training session in order to keep track of how much she was eating. Once I knew this number, I knew how much to give her for her daily “meal” because I made sure she ate at least 22 grams per day. I also weighed Louise daily in order to be sure that I wasn’t depriving her too much.
In addition to keeping track of her food, I also had to begin bringing water to training sessions. I had noticed that upon being put back in her cage after training, Louise would go directly to her water bottle and drink a ton! By providing her water during training, I was able to get in at least 10 to 15 more trials on average.
Another extremely important factor worth mentioning is that Louise and I worked best in a completely quiet, undisturbed environment. Any slight noise made us both get off task, and having other clickers around made training nearly impossible!
After weeks of training, I am now able to recognize the behaviors that mean Louise is distracted and will not react to the clicker. These behaviors include running, sniffing the air, standing up on her hind legs, grooming, and even sleeping! I even had two days at the beginning of the third week during which Louise seemed to completely regress. She only exhibited the above behaviors and she showed little to no response to the clicker. I was very worried that her previous training had become obsolete, but as I began working with her again she began picking up on her previous training.
Overall, it took about a week and a half before I began to see much progress. Although Louise seemed to know which was the “food corner” because she tended to hover there for long periods of time, she didn’t begin reacting to the click until the middle of the second week. Typical early responses to the click included ear perks, head turns, searching for food, and sniffing the air.
Once I noticed these initial responses, I began trying a “shaping” technique. This strategy consisted of allowing Louise to turn or walk away from the food corner slightly and then clicking and dropping the food. After about a session of trials of this method, Louise began to react to the click. She would turn abruptly toward the corner or even walk toward where the food was normally dropped. I continued to let Louise get further and further away before clicking. I would also click right as she began to walk over to the food corner of her own volition, hoping this would help “make the connection.”
Louise eventually responded to the click from all the way across the tank. This “responding” behavior began as a delayed response in which she would sniff around a bit or finish grooming before she “moseyed” to the food corner. After hundreds of trials of this, Louise eventually began reacting quicker and quicker until she was running to the correct corner upon hearing the click. At this point in time, Louise responds to the click about 90% of the time. It’s been a long road, but it was worth it!
As a supplement to the above description of Louise’s clicker training, I have also put together a number of clips that briefly chronicle our clicker training journey. Enjoy!
I also created a video of some of Louise’s most unproductive behaviors. Although they are pretty straight forward, I figured it would be helpful to show some of the things we don’t want our rats to do!
After a restless night of worrying that Louise had suffered some horrible memory loss and would never be clicker trained again, we had an extremely successful couple of sessions today. Although we did have to take a step back and re-train a bit (I had to go back to the shaping phase and work back up to clicking when she was across the tank) Louise was on task, hungry, and energetic. During our second session we got in over 110 trials! It was awesome and I think we’re back to the point we were before her “off days.” She is coming to the food corner when I click about 75% if the time and most of the other times are when she is sniffing around or grooming. However, I did notice that she stopped grooming at least once to run over to the correct corner; and she also ran to the corner as soon as she finished doing whatever she was doing a couple of times. Regardless, we’ve definitely taken a giant leap forward!
I’m writing this after two particularly frustrating days of clicker training. On Sunday, I had Louise coming to the “food corner” after I clicked about 70% of the time. I had almost completely phased out the second click (when she reached the food) and her behavior was looking very promising. I even have two great videos of her responding to the click… I was convinced that we had it down!
Then Monday rolled around and something about Louise’s behavior was just off. I came in for two sessions, one in the morning and one later that evening, and both had about 10 minutes of productive training. During each session, Louise spent the majority of the time walking to various corners, standing up on her hind legs, and sniffing the air. When she wasn’t doing this, she was grooming, scratching, or falling asleep! I have learned to recognize these behaviors as times during which Louise will not respond to the clicker at all. I like to joke that I could do a dance with the clicker and she would simply continue sniffing or grooming.
I figured that Louise was just having an “off day.” Maybe she was tired, feeling sick from her food, or something had scared her in the lab prior to me coming to train that had had some negative effect. Additionally, since I was able to get her to come to the corner when I clicked at least a few times during those sessions, I hadn’t completely given up hope.
Unfortunately, tonight’s session was anything but stellar. It was almost as if I was working with a different rat, and it seemed like we were back at square one. Although it was clear that Louise still knew where the “food corner” was because she lingered there nearly the entire time, she was showing no response to the sound of the clicker. Usually I can at least get an ear perk out of her, but today she just sat still.
Louise still managed to eat at least 40 or so pellets; however, that was only because I was going back to the “shaping” training I had started last week in which I would let her get a short distance away from the corner before I clicked and dropped the food. She was responding well to the “re-training,” until she began sniffing the air again and not responding at all. Eventually, she also began to leap and jerk around the tank, and she even tried to jump out once!
Additionally, a lot of my trouble stemmed from the fact that she wouldn’t even leave the food corner (which makes it somewhat hard to come running to the corner when I click!) And when she would leave the corner it was almost always because she was sniffing the air or grooming, which makes her nearly impossible to work with at all.
This behavior is extremely odd for Louise (especially the lethargic sitting in the corner) and I really hope it passes. I am hoping that it’s just some sort of fluke or unfortunate circumstance and not some biological rat issue. I can understand that living things can be unpredictable, but I never would have expected her to completely stop responding after doing so well for a couple of days. I also hope that it’s not something that I am doing wrong. I haven’t changed anything regarding Louise’s schedule and my techniques have essentially remained the same. I’ll continue to stay as consistent as possible.
I’ll be heading back tomorrow morning and hopefully there will be a change for the better.
After clicker training today, I’m feeling much better about Louise’s progress. Although her responses are still not completely consistent, she is beginning to respond to the sound of the clicker. I have posted two videos showing her reactions to the clicker. I apologize in advance for the quality of each–I was in the lab by myself and had to improvise filming!
The first video shows Louise responding to a click after having turned away from the “food corner.” This is a bit earlier on in the training, but still after I began allowing her to move away before clicking. You can see that I’m not allowing her to get very far away from the corner before clicking, but the clip still shows a difference in behavior.
The next video shows Louise walking out of view to the right front corner of the tank, and then crossing back over to the “food corner” when I click. Although I waited until she was facing the front left corner before I clicked, she still made it across the tank. This is very promising!
The Tuesday deadline for having Louise clicker trained is fast approaching and I feel as if we’ve reached a plateau. Louise has been great during training–she is always hungry and she hangs out in the correct corner throughout most of the session. We even tend to get in at least 70 trials per session! However, it seems like our time is spent clicking, dropping the food, and eating, over and over. Not much else seems to be happening!
Thus, yesterday I decided it was time to switch things up a bit. Instead of just clicking and dropping over and over, I have begun waiting for a few seconds between each trial. This gives her the opportunity to attend to something else or meander away a bit. As soon as she begins to turn away, I click and drop the food. Over the next couple of days I’ll let her get farther and farther away from the corner before I click. Hopefully this will eventually lead to her coming to the corner when I click, regardless of where she is in the tank.
Right now, she sometimes comes to the corner when I click, but I think anytime she would move away before I started this new method, she was actually too distracted by some new smell or sound (in which case, she doesn’t seem at all affected by the click.) Thus, by giving her the opportunity to move, but not forget about what we’re doing, hopefully she’ll begin to get it! I’m nervous because it’s coming down to the wire, but I think we’ll get it soon.