Does intention matter for positive reinforcement?

Question

What makes something positive reinforcement: The intention or the outcome? I have two examples I'm trying to understand.

First, if I am teaching my dog a trick and I give her a treat whenever I say "sit" and she sits, I'm intending to reinforce that behavior. Let's say my dog is deaf, so she isn't learning from my verbal command, and she doesn't learn to sit when I say "sit". Is this still considered positive reinforcement because of my intention, or is it not positive reinforcement because it is not effective?

Second, say I'm trying to teach my dog not to bark whenever the doorbell rings. Whenever the doorbell rings, she barks, and I give her a treat to distract her. Instead of learning that the doorbell is no threat (which was my intention), she learns that if she barks she gets a treat. Does this count as positive reinforcement, even though I did not intend to reinforce that behavior?

Essentially, I'm trying to understand whether intention and/or effectiveness have any impact on what we can consider positive reinforcement.

Any sources related to this would be extremely helpful, thanks!

Answer 1

The short answer is that only the outcome matters:

These terms are defined by their effect on behavior. ... Positive reinforcement and negative reinforcement increase the probability of a behavior that they follow ...

So:

• If your dog receives a treat from sitting, and sitting behaviour increases, then you have positive reinforcement.
• If your dog receives a treat from barking, and barking behaviour increases, then you have positive reinforcement.
• It doesn't matter if the dog is deaf, only if the dog sits or barks more. If the dog being deaf means she receives no treats, then no reinforcement has occurred.

However, intention does play a role in measurement. When evaluating reinforcement, you measure the frequency of the behaviour that you intend to affect - eg, barking. Thus, all other behaviours are effectively ignored - eg, running. So while unintended reinforcement might happen, if it isn't measured, then you won't know!

To use the example in the question, if you intend to affect barking behaviour in the presence of a doorbell ringing, then you measure the frequency of that behaviour throughout the conditioning process, and find that it does not change in frequency, and hence no conditioning has occurred.

But in doing this, you inadvertently increase barking frequency outside of the measurement context (no doorbell), which you don't know, because you are not measuring that! If you instead intend to affect all barking behaviour (with or without the presence of a doorbell), and measure that instead, and see that it has increased in frequency, then indeed, you have positive reinforcement.