How to actually run the Dont Shoot The Dog Game at home without it becoming a disaster
I ran this activity with a group of eight year olds last spring for a behavioral psychology workshop I was helping lead. We used it to demonstrate Skinnerian principles in a way that stuck. The Dont Shoot The Dog Game is fundamentally a clicker training exercise adapted for human learners, usually children, where one person acts as the "trainer" and another as the "subject." The trainer reinforces specific behaviors using a clicker or marker word, and the subject gradually shapes into a target behavior without any verbal instructions about what is expected. The whole point is that the trainer never tells the subject what to do. Reinforcement happens only after the desired behavior occurs. This is pure operant conditioning in practice. Most people think they understand it when they read about it. They do not understand it until they are the one clicking randomly and watching something weird happen by accident.
Setting up the Dont Shoot The Dog Game
You need a quiet room, a clicker or something that makes a consistent sharp sound, and two willing participants. One person is the trainer, one is the subject. The subject sits in a chair facing away from the trainer initially, or you can use a blindfold to remove visual cues entirely. I recommend the blindfold approach. It forces the trainer to rely purely on auditory feedback and makes the session shorter because the subject cannot pick up unintended visual signals. The trainer picks a target behavior before the session starts. It should be specific enough to shape but simple enough to achieve in a controlled environment. Examples include touching the left knee with the right hand, clapping exactly three times, or standing up and sitting down in a single smooth motion. Do not pick something vague like "be nice" or "act calm." That will not work. The behavior needs to be observable and measurable. Here is the mechanics of it. The subject performs any behavior. When the trainer observes the target or a close approximation, they click and provide a reward. The reward is usually verbal praise, a small treat, or a point on a scoreboard. Over successive trials, the trainer only clicks for behaviors closer and closer to the final target. This is called shaping. You are gradually raising the criterion for reinforcement.
I have found that a single 20 minute session is usually enough for a competent subject to learn a moderately complex behavior chain. Simple behaviors take about five to eight minutes. Complex chains with three or more steps can drag to 40 minutes before the subject cracks it. Time varies heavily based on how consistent the trainer is with their clicking.
Get the Full Details
What actually goes wrong
The most common failure mode is accidental negative reinforcement. The trainer gets frustrated and starts giving verbal corrections like "no" or "not quite" whenever the subject does something wrong. This inadvertently punishes the subject and slows learning significantly. I watched a colleague do this on day two of a training workshop and we lost an entire hour rebuilding the subject's willingness to engage. The subject just sat there staring at the wall. We had to restart from scratch. Another frequent error is clicking too late or too early. The click must land within one second of the desired behavior. If you click after the subject has already stopped doing the thing, you have reinforced the wrong action. This creates confused behavior patterns that are annoying to untangle later. I learned this the hard way when I was training my own dog and accidentally reinforced him for looking away instead of making eye contact because I clicked a half second late. Took me three weeks to reverse that mistake. There is also the issue of satiation. If you are using food rewards, the subject will stop caring after a certain number of repetitions. In my experience with children, this happens around 15 to 20 rewards per session. Switch to non-food reinforcers like points, tokens, or social praise to avoid this bottleneck entirely. I keep a bag of dry roasted almonds for dog sessions and a jar of stickers for kid sessions. Works every time.
Advanced considerations
One thing most beginners miss is the importance of capturing the exact moment the behavior begins, not just when it ends. If your target is "touch the nose to the table," click the instant the hand starts moving toward the table, not after contact is made. This distinction matters because it teaches the subject that the initiation of the action is what counts. Without this precision, you end up with subjects who perform the action but only at the very end of a long hesitating pause, which defeats the purpose of clean conditioning. A second nuance is the use of a bridging stimulus. The click itself becomes a conditioned reinforcer. After enough pairings with the primary reward, the click alone carries value. This means you can eventually phase out the treat and rely on the click as sufficient reinforcement. Most people skip this transition and keep reinforcing every click forever, which is unnecessary and wasteful. After about 30 to 50 reinforced trials, the click should carry enough weight to sustain the behavior independently.
Where this breaks down completely
The Dont Shoot The Dog Game does not work for every population. Children under five lack the working memory capacity to form the association between the click and the specific behavior. Adults with certain cognitive impairments may also struggle. ADHD subjects tend to perform unpredictably because their attention shifts faster than the trainer can capture it. I once tried this with a participant who had untreated ADHD and we spent the entire session waiting for a single correct response that never came in a meaningful pattern. We switched to a different format entirely and moved on. It also fails when the target behavior requires cognitive understanding rather than motor execution. You cannot shape someone into understanding philosophy through clicker training. Pick behaviors that are physical, observable, and repeatable. Everything else is just entertainment dressed up as education. If you want to download materials or find printable guides, the official Clicker Training resources from the Karen Pryor Clicker Training website have free downloadable worksheets and session tracking templates. They are straightforward and cover the exact protocol I described above without the marketing fluff. Search for the free animal and human training PDFs on their resource page. No account required.
Bottom line
The Dont Shoot The Dog Game is a practical demonstration of shaping through positive reinforcement. It works when you are precise, patient, and consistent. It fails when you get sloppy with timing or try to force it into situations where it does not belong. Use it to teach concrete behaviors to subjects who can sustain attention for 15 to 20 minutes. That is about the useful range for this kind of exercise before fatigue sets in and the data becomes unreliable.
