Timing of reinforcement during dog training

Research on dog-human communication has demonstrated that dogs are extremely responsive to human cues such as pointing, eye gazing and vocalizations. Because dogs are so receptive to such cues, it is reasonable to assume that subtle feedback from humans has an effect on the efficacy of dog training. Timing of reinforcement in the field of dog behavior has not been researched extensively. Research on other species has demonstrated that although animals can learn tasks with delays to reinforcement, longer delays result in longer average times to task acquisition and relatively lower rates of responding. The aim of this study was to examine owners’ latencies to providing reinforcers for their dogs’ responses during basic dog training. Video observations were made at three New Zealand dog obedience clubs. Fifteen people volunteered to take part in this study with their dogs. All participants were members of beginner classes and were videoed while training their dogs in class. Behaviors appropriate for examining the timing of reinforcement required a clearly-definable start and finish point, so ‘sit’ and ‘down’ were chosen for analysis. Times were measured between the owners’ commands, the dogs’ responses, secondary reinforcement (verbal praise), and primary reinforcement (food). Preliminary analysis of data from seven participants shows that dogs were reinforced for responding correctly to 27% of commands. These events were used for timing analysis. In addition, the dogs responded correctly but received no reinforcer of any type for 31% of commands, they responded incorrectly to 2% of commands and 40% of commands elicited no response from the dogs. There was considerable variation across participants’ timing of reinforcement with the latency to deliver the first instance of reinforcement (secondary or primary) ranging from 0 to > 5 seconds. No significant difference was found between owners’ latency to deliver secondary or primary reinforcement (t (5) = −1.66, P = 0.16) or between the times taken to reinforce ‘sit’ or ‘down’ responses (t (5) = −0.65, P = 0.54) (first instance of reinforcement). Results showed a positive correlation between the time to the first instance of reinforcement and the proportion of incorrect responses (r (7) = 0.65, P = 0.12). We conclude that given the wide range of latencies to reinforcement, it is possible that the dogs’ task acquisition was suboptimal at times. Delays to reinforcement may also give room for unintentional feedback to occur, particularly as dogs are so receptive to human-given cues. Ongoing research will attempt to address these questions.
Journal Article
Type of thesis
Browne, C.M., Starkey, N.J., Foster, M.T. & McEwan, J.S. (2011). Timing of reinforcement during dog training. Journal of Veterinary Behavior: Clinical Applications and Research, 6(1), 58-59.
Publisher version