Friday, July 24, 2009

Penn Current: Research: Questions about ‘aversive’ training

Penn Current: Research: Questions about ‘aversive’ training


  1. Very interesting!! Good to finally have people start to do studies on behavioural aspects - too much anecdotal stuff floating a trained scientist myself, I always believe in evidence-based practices.

    Having said that, though, there might be questions about the robustness of this study - a) in terms of numbers and b) in terms of any inherent biases - as the surveys were returned voluntarily therefore, does the type of person likely to complete and return such a survey have any influence in the results - ie. are the results representative of the entire population of dogs trained using aversives or is this just part of the story?

    Similarly, while the dogs trained using positive techniques "showed no aggression" - what does this actually mean? No aggression towards their trainer in that situation or less aggresion overall? Because it's all very well saying they showed less aggression towards their trainer but if their overall problem of aggression towards other sources is not reduced, then this raises questions about effectiveness too.

    There also needs to be clearer definitions of the various categories - ie.what this study looked at: pinning/alpha rolls & beating are at the EXTREMES of punishment-based techniques - but there are a lot of grey areas in between - eg, using noise aversives - would that trigger more aggressive behaviour too? Anecdotally, yes, but scientific proof?

    Also, what exactly does it mean that positive-trained dogs showed no aggression - towards their trainers? But what about at other times? What if they were already exhibitiing aggression (ie. this is the problem that needed help) - did this aggression reduce significantly with positive techniques only? What about dogs trained with a combination of positive AND aversives? Or with aversives only?

    There needs to be a controlled experiment looking at similar-temperament dogs with identical issues tackled with these 3 different types of techniques and to see what significant differences in results there are (results being defined as a reduction in aggression & reaction to the stimulus.)

    Not criticising the positve method but just pointing out that if you want to look at things the scientific way, then it has to be done completely properly the scientific way - otherwise it would be as bad as an anecdotal approach used to illustrate only one side of the argument.


  2. Hi Hsin-Yi,
    Wow, you are awesome at how you totally picked apart that study. I wish that I could do that as I read these studies. I just read them and think wow, that is really interesting. You need to write a book!!! I would buy it!
    Have a great day.

  3. All very good points. What needs to happen is a meta study on dog training and behavior, not just one or two clinical trials. Usually a study like this is designed to get a desired result. You could get the same data if you looked at any training program.