Some detrimental effects of conditioned reinforcement on the maintenance of dog behavior
A clicker loses power fast if food stops coming—keep the treat ratio high.
01Research in Context
What this study did
Two dogs learned an odor-detection task. A click always came after a correct nose touch.
In some phases the click was followed by food every time. In others food came only half the time.
The researchers flipped these conditions back and forth to see how the dogs reacted.
What they found
When food followed only half the clicks, correct touches dropped and the dogs walked away more often.
Skipping the treat too often crashed both accuracy and willingness to work.
How this fits with other research
Barlow et al. (2015) saw the opposite. Shelter dogs got free treats with no response required. Kennel barking and jumping still went down.
The key difference is setting. Shelter dogs lived in noisy kennels and saw people as relief. Lab dogs worked for every click and expected pay.
Older pigeon and rat studies (E et al. 1968; R et al. 1975) show that a conditioned signal keeps working if it has a rich food history. Peiris et al. (2022) show the signal quits when that history thins.
Why it matters
Your client’s clicker or token is a promise. Break the promise too often and the behavior will sag.
Keep the backup rate high at first. Thin it later only if the response stays strong.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Track how many clicks you pay with food; aim for at least 80% this week.
02At a glance
03Original abstract
In animal training, there is disagreement regarding whether a conditioned reinforcer, such as the sound of a clicker, should always be followed by an unconditioned reinforcer, such as a treat. Proponents for clicking without always giving a treat argue that the click can substitute for the reinforcer, due to Pavlovian conditioning and the partial reinforcement effect. Those who advise always following the click with a treat argue that the clicker will become an unreliable predictor of food if it is not always followed by an unconditioned reinforcer. In this study, a within-subject reversal design with two dogs was used to compare the behavioral effects of always following a click with food (one click condition) and only sometimes following a click with food (two clicks condition). Results showed that the two clicks condition disrupted the frequency, accuracy, and topography of the behavior and increased noncompliance and other unwanted behaviors. While the detrimental effects of this condition may seem paradoxical at first, they can be explained by the discriminative properties of the conditioned reinforcer.
Journal of the Experimental Analysis of Behavior, 2022 · doi:10.1002/jeab.790