A few years ago i wrote a blog post about LRS: https://djurtranarskolan.wordpress.com/2011/12/01/lrs-least-reinforcing-stimulusscenario-2/ After attending the ORCA/Wicked Minds conference this weekend, listening to an excellent LRS presentation by Steve Aibel (Sea World San Antonio) and getting input from numerous brilliant trainers and behavior analysts, I feel it’s time for an update on the subject!
So, the basic question is “What do I do when my animal makes a mistake?”.
What’s a “mistake” in this context? Well, anything that doesn’t meet my criteria for reinforcement. So I’m looking at my learner, and my plan is to reinforce as her behavior meets criteria – but instead I see another behavior that I don’t wish to reinforce.
A little aside: Labeling it a “mistake” (or wrong, or incorrect, or error) is really a bad habit, I know– it’s all just behavior, stemming from the individual’s and the species’ history, and we shouldn’t put a right/wrong value to it. But for convenience, in the current discussion I’ll use the term “mistake” as a label for “behavior I don’t wish to reinforce”.
Then there’s the timeframe question. Here, the discussion revolves around what to do immediately after the behavior in question. A different (yet obviously related) question is what to do more long term, on the larger time scale – but that’s for another discussion. (In parenthesis, the best answer would be “focus on what the desired alternative would be and make sure to build that”…)
Whatever I do when my animal makes a mistake is both a C and an A
OK, back to the question: “What do I do when my animal makes a mistake?”
The answer to this question has to reflect the way behavior flows; the stream of behavior is multi-layered and unstoppable, and the A-B-C contingencies constantly builds into each other.
So whatever I as a trainer choose to do when my animal makes a mistake will function both as a consequence (to the “mistake” behavior) and as an antecedent (setting the stage for what the animal will do next).
On the consequence end (the rear end of the elephant, as Steve Aibel put it) we wish to avoid reinforcing the previous behavior.
The consequence as an antecedent
However, too often trainers get a bit hung up on “not reinforcing” and forget to look at what happens next – what does the animal do then, in the constant stream of behavior? And how is that reinforced? These behaviors that occur after a mistake, do they help or hinder our future training?
So, we have to be mindful of the antecedent end of whatever consequence we provide. In a “positive reinforcement system”, I wish to set the occasion for desired behavior to occur and I wish to build and maintain them with positive reinforcement. This goes also for whatever behavior I wish to see my animal doing after a mistake!
What if there is no opportunity for behavior+reinforcement after the mistake?
If I forget training for nice behaviors following mistakes – then what might my animal do? Well, any situation when reinforcement is not available through me opens up either to behavior being
If there’s not anything in the environment signaling that reinforcement from the trainer is available for some behavior – well, what’s the animal then to do?? In such a situation we might see frustration/aggression/displacement behaviors (behaviors that typically show up under extinction conditions). And/or we might see the animal working for other possibly available reinforcers – looking at or moving toward something, exploring the environment, and other behaviors that we as trainers might label “lack of focus” or “distractable”…. What we’ll end up with is double trouble; first a mistake, then a mess of undesired behaviors following the mistake.
Since I do not want to see neither frustration/aggression/displacement behaviors nor scanning for reinforcers in the environment anytime in my training sessions, I of course want to avoid this. Thus I have to make sure reinforcement is always available through some behavior – after mistakes, and at all other times.
The question and some possible answers
Therefore, a basic question when deciding what consequence to provide after a mistake is: What behavior should then follow, and how do I plan to build and reinforce that.
There are of course many different ways to solve the puzzle of “what do I do when my animal makes a mistake, taken into consideration what behavior my animal should do then and how I in turn reinforce that behavior?”. I’ve tried to gather a bunch of examples (coming from various brilliant trainers):
- Trainer alters nothing – animal keeps trying
- Reward (give a reward – maybe lower value)
- Redirect (give a cue)
- Adjust the environment
Note that the list is not in any organized order…. In my own training I do use all of the above, in different situations.
My plan is to elaborate on each of the items on the list… but I realized that’d make this more into a book that a blog, so the compromize will be to make one blog post for each item 🙂
So, stay tuned, more is on its way 😉