I think you're approaching it from an operant conditioning model where the treat must come after as a consequence of good behavior. But what they describe sound

I think you're approaching it from an operant conditioning model where the treat must come after as a consequence of good behavior. But what they describe sounds like reciprocal inhibition: https://en.wikipedia.org/wiki/Desensitization_(psychology)#Reciprocal_Inhibition https://en.wikipedia.org/wiki/Desensitization_(psychology)#R... The idea is that you can't feel opposing emotions simultaneously. So if you can make the dog feel relaxed (by giving a treat beforehand), it interferes with their ability to feel anxiety. They aren't earning the treat for choosing not the be anxious. The positive emotional feeling from the treat prevents the anxiety from occurring.