One thing that distinguishes basic from excellent positive reinforcement training is reinforcer delivery. You may be hearing a lot about this in the dog world lately, including Mary Hunter's brilliant demonstrations of building "reinforcement systems" for more effective teaching (she's cohosting a summit on the topic, registration for which closes today), as well as Hannah Branigan's presentation on how to build what's often called "drive" and Eva Bertilsson and Emelie Johnson Vegh's presentation on "starting points" at last week's ClickerExpo.

Here are a few practical considerations and tips that I originally wrote for my professional students in the Karen Pryor Academy on how to thoughtfully deliver just one of the many types of reinforcers that positive reinforcement trainers can use: food, aka "treats."

The delivery of the reinforcer (e.g., the treat) for the behavior that just happened is also the antecedent for whatever the dog is going to do next.

So consider how to deliver the treat so that when the dog is done eating, they are perfectly in position to offer the next thing you want to see.

Some simple examples:

• Want the dog to be ready to move its feet toward a target again? Deliver the treat away from the target.

• Want the dog to remain in position and offer eye contact after eating? Deliver the treat right to the dog's mouth so the dog’s front feet don’t move.

• Want the dog to immediately walk a few more steps by your side? Deliver the treat at your hip or slightly behind.

• Dog shoots ahead after each treat? Toss the treat for attending to you or walking by your side out into the grass, then walk forward and wait for attention or movement to catch up.

• Want the dog to look at a surprise trigger from a little farther away? Toss the treat for the first glance (or have the dog follow it a few steps) in the opposite direction from the trigger.

Behavior between the marker signal and the treat will also be reinforced—and after a lot of predictable reps, you may start to see it occur in anticipation of the marker instead of after it.

So consider how to deliver the treat to avoid things you don’t want between the marker and the treat, and encourage things you do.

Some simple examples:

• If you don’t want a puppy to jump up your leg between the click and the treat, preload your treat hand and deliver treats quickly to the floor after the click. This leaves no time for jumping, and no reason because it’s wasted effort if the treat is predictably going to appear down low. You'll then be able to move the treats back into your pouch once the puppy is waiting down low.

• If you don’t want a dog to curve in front of you every few steps on a walk, deliver treats from the side of your body that the dog usually walks on, rather than bringing them across your body with the opposite hand.

• If you don’t want your dog to turn toward you after going over a jump, mark as the dog begins the jump and send the treat (or toy) in a straight line ahead of them.

• If you'd like your dog to start to sniff when they see another dog, then mark for looking at the dog and sprinkle your treats on the ground rather than handing them to the dog. Your dog will likely begin to look down on hearing the marker, and may then begin to look down on seeing the other dog.

How you deliver reinforcers can add value to or subtract value from the reinforcer itself.

So consider whether:

• Your delivery might be boring: for instance, offering a treat (or toy--this is a common problem in tug) by holding it still rather than moving it away from the dog.

• Your delivery might be aversive, e.g., shoving food into the dog’s mouth or backing a dog up with the treat.

• Your delivery is getting delayed because the dog didn’t see where you tossed the treat or it went under something.

• Conversely, sniffing around in the grass for a treat might add value in some situations.

• A bowled- or “catch”-type treat delivery might be more exciting than a treat delivered to mouth (I often like this for helping food compete with birds and squirrels!)

You may want to change where you deliver the reinforcer at different points in your training of a behavior.

If you deliver in the same place every time without thinking about it, the behavior is likely to drift in the direction of reinforcement delivery. So be ready to watch behavior and constantly adjust.

Examples:

• If you toss your treat to reset every time you click down, your dog may start to pop up in anticipation of the click, or not lay down all the way. You can fix that a number of ways, including being more careful with the timing of your click, but another way is to begin to deliver the reinforcer in place once the dog understands that the behavior is to put their torso on the floor.

• When I attended Bob Bailey's chicken workshops, one exercise was to shape a chicken to walk a figure-8 around two cones . . . after shaping them to walk an oval around them. I clicked and treated a lot for the first step the chicken took to cross through the center, and after a while the chicken would pause there and glance back at me.

• This can be used for good as well as for evil; if you want to teach a dog to shift its weight back when it reaches a threshold (like a door or curb), mark them for arriving at the threshold and deliver the treat behind them.

Consider teaching the dog how a treat will be delivered in a given session before starting to work on the goal behavior.

• The click is frequently talked about as a conditioned reinforcer, but it may be better thought of as a cue to do whatever behavior is required to collect the treat or other "terminal" reinforcer.

• When cues are not clear, their value as reinforcers is weakened. So make your cues about where reinforcement will be delivered clearer.

• Let’s say you are going to click and toss a treat behind the dog. Start with just that—reinforce the default stand, say, with a click and a clearly telegraphed toss to a spot on the floor.

• Once the dog has that pattern, then add in your cue for the established behavior you want to work on, or start capturing or shaping a new behavior.

Check back later; I'll try to add videos to illustrate these various suggestions. But I wanted to get the post up today in case anyone wants to sign up for the Behavior Explorer event.