Studies to assess the effect of pet training aids, specifically remote static pulse systems, on the welfare of domestic dogs: field study of dogs in training

Cooper, Jonathan, Cracknell, Nina, Hardiman, Jessica and Mills, Daniel (2013) Studies to assess the effect of pet training aids, specifically remote static pulse systems, on the welfare of domestic dogs: field study of dogs in training. Project Report. Department of Environment, Food and Rural Affairs.

Full content URL:

[img] Microsoft Word

Item Type:Paper or Report (Project Report)
Item Status:Live Archive


The project had a single aim, namely to assess the impact of use of remote static pulse electric training aids (e-collars) during the training of dogs in comparison to dogs referred for similar behavioural problems but without e-collar training. The specific objective was to use appropriate behavioural and physiological measures to make inferences about the welfare of subjects including their aversion and anxiety during and following training. A secondary objective was to evaluate the efficacy of treatment and possible explanations for any differences between treatment groups. These included owner assessment of severity of problem, population characteristics, trainer factors and temperament of dogs. The study design was developed following consultation with trainers to ensure a protocol that was representative of advised best practice in use of e-collars. Adult dogs (n=63) undergoing training by professional trainers were recruited to investigate the welfare consequences of behavioural modification incorporating e-collars, in comparison to positive reward based training. Dogs had been referred for problems commonly addressed using e-collars (recall problems and livestock/wildlife worrying) and were divided into three populations; one using e-collars and two control populations where dogs were not exposed to e-collars. Treatment groups were defined as follows

Group A: E-collar group (N = 21) ECMA recommended trainers with experience of use of e-collars, using e-collars as part of their training programme.
Group B: Control 1 (N = 21) ECMA recommended trainers with experience of use of e-collars, not using e-collars as part of their training programme.
Group C: Control 2 (N = 21) APDT affiliated trainers who do not normally use e-collars in training, not using e-collars as part of their training programme.

Methods Behavioural and physiological data that can be related to a dog’s emotional state were collected to assess immediate impact of exposure to e-collar stimulus in comparison to the control groups, as well as adaptation during training procedures. This included field observations of behaviour, cortisol samples from saliva and urine, judgement bias tests and questionnaires relating to impulsivity and temperament. Pre-training data were collected to assess if there were significant differences in treatment groups prior to training and follow up assessments were conducted 3 months post training to assess longer term welfare consequences of training methods and their efficacy. Dogs were allocated to treatment groups to control for the nature of referred problem and owner assessment of severity.
During training, data were collected over a period of up to 5 days covering the period of initial behaviour modification. For Group A, dogs’ training regimes followed recommended practice for resolving the problem under referral as advised by e-collar manufacturers. Trainers used e-collars with a variable intensity setting and a pre-warning cue which, with time, would allow dogs to modify their behaviour to avoid exposure to the e-collar stimulus. Dogs in this group had the working level of e-stimulus determined on day 1 of training. On subsequent days non-compliance with tasks was associated with exposure to e-stimulus. The pre-warning stimulus was used to predict potential application of e-stimulus if the dog did not respond appropriately. Dogs in this group also received positive reinforcement such as food, play or praise for compliance. Dogs in control groups B and C followed training regimes without exposure to e-collar stimulation. Dogs in these groups wore a dummy collar (de-activated e-collar) to control for collar wearing and ensure observers of video tapes were blind to treatment. On the final training day, dog owners conducted the training under instruction from the trainers.
Following completion of training, owners were surveyed for their assessment of the ease and efficacy of the training programme. At three months post-training, owners returned to the training centres and took dogs through a series of training commands including the recall command that they had been trained to follow. Dogs’ behaviour was video recorded and saliva samples taken on first arrival and following training with and without a deactivated collar.
Results There were no differences in the reasons for referral between dogs in the three groups or their owner’s assessment of the severity of the referred problem. Data collected prior to training days found no differences between the treatment groups in dog’s performance in cognitive bias tasks or in their temperament and impulsivity scores. There were no day effects on behaviour except more owner directed behaviour on final day of training.
Dogs in groups A and B spent roughly half of their time walking during training, whilst dogs in group C spent more time standing and spent less time sitting during the training sessions than dogs from groups A and B. Group C dogs also showed more food related lip licking than the other treatment groups. Dogs in group C engaged in more environmental interaction such as sniffing, than dogs from Groups A and B and were less often observed yawning than Group A dogs. These treatment effects were found to be significant following Holm-Bonferroni correction for multiple comparisons. Dogs in group C may have spent less time tense during training sessions (4% of scans) than dogs in groups A (25%). Dogs from group C may have had their tail in a low position less often (0.8% of scans) than dogs from groups A (9%) and B (6%) and may have moved away from the trainer less often. Dogs in Group A may also have yelped more (0.5 per training session) than dogs in groups B and C (0.1 per session) and panted more (20% of scans compared with 10%), however these latter measures were not found to have a significant treatment effect at the sample sizes of this study.
Dogs in group C had higher salivary cortisol levels throughout the training period compared to group B, whilst group A dogs’ salivary cortisol was no different to groups B or C. There was a decline in salivary cortisol in all groups over the 5 training days, but no difference in the size of this decline between the three groups. There was no significant difference in urinary cortisol:creatinine ratios between treatments and no changes in this ratio over the five days of training for any treatment.
Owners were generally satisfied with the training advice they received, and no differences were found between the treatment groups in owner satisfaction at one week or 3 months post-training. It would appear from owner feedback that spending an extended period of time with experienced trainers handling their dogs was generally beneficial in terms of improving owner’s understanding of dog behaviour and training. These measures were however solely based on owner assessment, and an independent assessment of the severity of the referred behaviours by for example a behavioural consultant may have been useful.
During observations on return to the training centre after 3 months, dogs tended to spend more time running (15.7% of scans) and more time excited (25.9% of scans) than in the training days and less time relaxed (24.1% of scans). Activities that might be associated with aversion or anxiety were rare and there were no significant behavioural differences between treatment groups. There were no differences between treatment groups in cognitive bias scores, nor in measures of temperament. None of these measures had changed from the pre-treatment assessments.
There was no effect of treatment on the ratio of urinary cortisol to creatinine at 3 month follow up. Group A dogs showed higher levels of salivary cortisol on arrival at training centre and there was also a decline in salivary cortisol from first arrival at training centre to end of training in all three groups.
Discussion and Conclusions. In previous defra funded work (AW1402) there was evidence that experience of e-collars had long term negative welfare consequences in some dogs from the pet community. However, the retrospective nature of this previous work meant it was not possible to establish a causal relationship. In this study we sought to overcome this limitation and also examine industry recommended practice concerning the use of e-collars in the field. Reason for referral, and severity of referred behaviour were also successfully controlled in this study. It was already known that training dogs with e-collars can have severe welfare consequences (E.g. Beerda et al 1998, Schilder and van der Borg 2004), however, advocates of e-collar training have argued that such studies are not representative of the appropriate use of modern e-collars for correcting behavioural problems, and for this reason this study focussed on the welfare consequences of use of e-collars within industry recommended training protocols compared with reward based training.
Dogs in the three treatment groups were found to be comparable in all measures prior to training with the exception that group C dogs had a higher salivary cortisol measure than group B before experiencing their training programme, which persisted throughout the five training days. There were, however, a number of behavioural differences between the groups observed during training. Some of these appeared to relate to trainer related factors, rather than the use of e-collars. For example, trainers for groups A and B tended to use more commands than those in group C and encourage sitting rather than standing. Dogs in groups A and B engaged in less environmental interaction than Group C dogs and there was some evidence that group A and B dogs also, showed lower tail carriage and moved away more often from trainers than group C dogs. There were, however, differences between behaviour of group A and C dogs of welfare relevance, for example group A dogs performed more yawns during training than Group C dogs and they were more tense, whilst group C dogs engaged in more environmental interaction. The incidence of yelping and panting noted in group A dogs related to high frequencies of these activities in a small number of dogs in this group. Overall these differences in behaviour suggest that some dogs in group A exhibited more negative emotional responses than dogs in Group C, though there were few behavioural differences between group A and group B
At follow up, the only treatment group difference was the higher salivary cortisol level of group A dogs on first arrival at the training centre, which may be related to anticipation of events based on previous experiences. However, this level declined with time and signs of aversion or anxiety were rare on return to the training centres, and not significantly related to treatment group. Dog owners were generally satisfied with the training advice they received, and although numerically a greater proportion of group C owners thought their dogs’ behaviour had improved, this was not a significant effect.
The results of this study show that that both the trainers’ general approach and the tools they use in training affect the dog’s emotional responses to training. It would therefore be of value to further investigate the welfare consequences of the skill levels of e-collar operator as well as the tools they use. Nevertheless the study did find behavioural evidence that use of e-collars negatively impacted on the welfare of some dogs during training even when training was conducted by professional trainers using relatively benign training programmes advised by e-collar advocates.

Keywords:Dog training aids, electronic collars, Dog welfare
Subjects:D Veterinary Sciences, Agriculture and related subjects > D328 Animal Welfare
M Law > M110 UK Legal Systems
Divisions:College of Science > School of Life Sciences
ID Code:14567
Deposited On:22 Jul 2014 15:52

Repository Staff Only: item control page