Bad CAT day for Iggy
This week we tried again CAT protocol with Iggy, this time with Justine as the decoy (thousands thanks to her patience). And it went wrong and we actually reinforced the opposite of what we were looking for.
CAT protocol
The principle
The CAT (Constructional Aggression Treatement) is basically a negative reinforcement based protocol to treat aggression. The idea behind this protocol is that the dog reacts to make the stranger go away. That is the function of the aggressive behavior. So in this protocol, we reinforce when the dog looks at the stranger (desired behavior) by having the stranger leave, which increases distance, and which reinforce the dog good behavior.
There are two reasons for this protocol to work :
- The dog's good behavior is reinforced with something he really wants : that the stranger leaved
- The dogs gains control over the threatening stranger : I can have it leave if I look at it for long enough.
After a few repetitions (they say about 45 minutes in a few sessions), the dog is supposed to :
- have understood that looking at the stranger makes her leave
- start displaying positive interest towards the stranger (sniffing the air) and even affiliative behavior that indicates that he wants to interact nicely with the stranger.
The protocole
I have read Snider's book on CAT, and listen to more recent podcast on CAT. Our protocol for this session was :
1. Determine the starting point
The stranger walks towards the dog until the dog stares at it, then leaves. This defines our starting point.
2. Step by step approach
We would then progress step by step following the exact same approach :
- The stranger walks to the waiting point until the dog stares at her, then immediately leaves until the dog can no longer sees her. We wait for 15 secondes. We do 3 repetitions.
- The stranger walks to the waiting point until the dog stares at her, waits for 1 seconds, and leaves. If the dog do a desirable behavior such as moving, looking at the handler, the stranger leaves to tell the dog that it's great behavior. We do 3 repetitions.
- Same then with 2, 3, 4 and 5 seconds. And then we move to the next step.
3. If the dog reacts
In the original CAT protocole, the stranger would stay when the dog reacts, and leaves when the dog displays a better behavior.
This got strong criticism as it leads the dog into an extinction burst where the dog is upset to see that his beloved technique (reacting) does not work. Because it has worked so well for him in the past, the dog tries harder and become even more aggressive: this is the extinction burst.
In a more recent podcast (2022), the founder of this method says that he no longer stays when the dog reacts, and that he goes away, and it still works because they spend most of the time below threshold.
A few things were unclear to me when we started :
- Is there a difference between trials where the dogs sees the stranger arrival (and then the duration of looking at the stranger gets much longer) and trials where the dog looks at the stranger once she has arrived.
- When the dog looks at the stranger and then looks away, the stranger leaves. But does this count as a successful trial ?
- Should the dog be relaxed at a step before going to the next step ?
The context
We decided to do CAT protocole at Julie's canine pension to be in protected contact with the dog free to move. There is also a kennel in which the dog can hide. The stranger approaches sideways, which wasn't optimal since it decreases the further distance at which we could start.
It's important to note that Iggy is well trained following the Click to calm approach in which he has learnt that looking at something scary and then looking at me is reinforced by a treat, and often by us moving away from the trigger (thing or person he finds scary or too much interesting).
First session
The first session was WAY too long. It started very well and ended very bad. And yet, I did other sessions after it.
The very beginning is perfect
At first Iggy is seated near me asking for scratches and cuddles. He looks relaxed (mouth opened, body supple) but I can see some kibbles on the ground that he has not eaten showing that he is not so relaxed.
When Justine arrives, he looks at her, still panting, then he looks at me. And she leaves.
I don't treat, I just pat him when he asks for it. The first 5 minutes, he looks ok with it.
Starts wondering
Then, it looks like he starts wondering what's happening. He looks less confident, he starts closing the mouth, he stops asking for pats (which he never ask for for too long even at home), and starts lying down.
It seems to me that when he looks at me, he is asking for help or validation that he is doing well. And I'm not giving that.
Lying down
For the next 25 minutes, he is lying down, his head often on top of his paw. He checks on Justine when she approaches but he no longer look at her for more that a glimpse. There are not many stress signals. He takes a treat I toss to him, but without much motivation. He is a bit “off”.
During the session, I take it as relaxation, and there is not much that can tell that it's not the case at this point, even looking back at the video.
Too much
After about 35 minutes, he is lying down, Justine is approaching for the second time at the same distance, and Iggy looses it. He gets up and lunges at Justine. I hadn't seen it coming.
Justine leaves as we had decided before, and she is a bit shocked.
A more vivid dog
After that event, Iggy is more vivid. He explores his box and discovers some kibbles left by the previous dogs and starts eating them. He sniffes the grass, pees on the wall, explores the environment. The change in behavior is very surprising to me.
Then Justine come back at a larger distance, and we resume to the protocole. Iggy explores his environment for a bit less than 10 minutes without reacting. Then he sits, and we have 1 successful trial followed by one failed where he reacts again, but with a lower intensity. And just after reacting, he drinks, which to me is a behavior he does when he is relaxed.
After that he lies down beside me, we have a last successful trial and we stop it, finally.
Afterthoughts
Interpreting the apparent relaxation of the beginning
I have 2 interpretations for why he looked so calm.
- He is following the click to calm — game, expecting that this would finally get Justine out of his way. And then he realizes that it's too much for him – or that the game is not working, and he panics.
- He is in an avoidance / dissociating mode, and then it's too much: too close and too long, and he panics.
Duration
The first take home message is: it was way too long!!! We should have stopped after he first lied down I think. So after 5ish minutes, when it started to become difficult for him as his closed mouth indicated.
I should not have had it last more that 20 minutes.
Engagement
A lesson of that session is about engagement. In the beginning, Iggy is engaged mostly with me, then mostly with the stranger, and then he disengages. Now I will consider it as a sign that he is no longer learning from the situation and take it as a sign that we should pause.
Treats and movement
Iggy knows very well the click to calm game. Here we want him to look at the stranger, I could/should clic and treat that. I don't think that it would hide Iggys' signals since there are already not many signals to look at. Plus I could toss treats on the ground to encourage him to move and this might be more instructive for us about his emotional state, and more helpful to him.
Look at that
I could also take advantage of that setting to teach him a “look at that” cue, and to reinforce looking at the person for a longer duration. The reward could be both the treat and the person leaving.
Start from a larger distance
Julie has boxes that you can approach from a larger distance while being seen by the dog. I would try there next time.
And in this case, I would keep the person visible by the dog in between trials, but at a very large distance so that it's not threatening.
Don't reinforce reaction
We don't see it in this video, but in the following 5 sessions (yes it's much too much, but they were shorter), he reacts 6 other times. I find that after each time, he seems more confident about doing the right thing. So next time, if he reacts, I plan on doing the old cat protocol and have the stranger turn her back and walk away only two strides and then wait until he does something better. And then the person leaves far away.