Classically, you explain the rule to your kid and give them a reward every time they get it right. When you feel that the trigger is "embodied", you slowly but surely suppress rewards. The link between trigger and reflex disappears but the reflex stays: the trigger is learnt.
Technically speaking, it's called classical conditioning (and no, it does not mean taking your child for a dog) !
Now, this has a downside: your kid is quite smart so it probably won't take it that you suppress rewards. Using a wording similar to yours, it's not a negative training (giving the stick) but it has an indirectly negative part (not giving the carrot).
As an alternative, I'd suggest instead what one could call "child-oriented conditioning": you use their wishes as a reward.
It goes somewhere along those lines:
- Your kid comes to you with a request (go play outside, get a candy, can I have a little sister or whatever).
- "Sure but let's play our little game first, okay?" You give the trigger, they show the expected reaction so they get their reward: what they asked for in the first place.
- Repeat over and over, possibly every time they come to you with a new request.
- When you feel that the trigger is "embodied", you slowly but surely stop asking for it when they come to you with a request. They simply get what they want (or not), like before.
See the difference (bold emphasis)?
Instead of taking their reward away, you stop asking for a reward in return. This turns the indirectly negative part into an indirectly positive one (not setting a condition in exchange for the carrot anymore).