I was reading with interest an article in a newsletter from NSDA about how a dog’s performance can be affected by subtle clues received from the handler. Things like nodding, pointing, etc. can “push” a dog to alert. This was also discussed in depth at the NAPWDA training.
How do K9 Handlers work to prevent this? I assume even having a conversation with a support person could lead a dog to an incorrect alert so what do we do to keep it from happening? I know we’ve been taught as support people to back off from both the handler and dog and let them work but to also watch the dog as closely as the handler in case you catch clues the handler may miss. We also try to stay off to the side and behind the dog to prevent our scent from interfering as much as possible.
Do you see interaction between the handler and support person as interference to the dog? We try to minimize the amount of conversation we have while searching but sometimes you need to discuss topics related to the search.
Early in the training any interruption can cause a dog and handler to lose focus. As the bond between handler and dog is solidified and the certifications continue, it allowable to be disruptive, especially when K9 is in scent. A well-trained dog should be very difficult to distract when working, which is why we always talk about consistency and reliability of the teams we deploy. In general the handler should always manage the support personnel to the advantage of the search effort which sometimes means moving, silencing and even flogging support and LE to keep the distractions to a minimum while working; however this does not mean silence is required, only discretion.
One of the things I did after I realized I was cueing my K9 when working problems I set up, was to work blind problems but try to behave the same.
Another thing is if I have him check an area, I start him further away than my area of primary interest – it allows me to see the difference between a negative area and a possible positive.
I whole-heartedly agree with Stuart – I think it is tough to distract a working K9 when they are in scent.
Cueing – can’t speak to how SAR dog handlers deal with it, or to what extent they individually see this. I do know that it has been a perennial question for narcotics and explosives detection dogs. For the reasons that it’s a problem for drug dogs, it may not be such a big thing for SAR.
We were concerned about it persistently since, in the field, we could never tell whether an alert was valid if drugs weren’t found. Did the dog alert because there was residual odor? Were the drugs so hidden artfully that we failed to find them?
Another thing had to do with the fact that training was usually framed around the dog getting immediately some sort of primary reinforcer (play, food, etc.) upon an alert. You were always shooting for the strongest, most persistent alert behavior attainable. What developed, once the dog had made the odor/reinforcer association, was what the behavior chain would become.
Detect odor – give alert behavior – look to handler for primary reinforcement. As the handler you had to be primed to give that primary reinforcer as soon as the alert behavior reached the desired criteria. Your body posture, attitude, etc. changes as you’re getting closer to the drugs if you know where the drugs are. If you couldn’t get around that, you fell into the trap of the dog watching the handler for those cues that the handler was preparing to give the reinforcer. There were some beautiful false alerts.
Ok, to your point – dogs appeared to pick up on tone, etc. of accompanying officers when you were doing these. On a street search, we really didn’t want the officer there if he/she would cue us as the handlers. And in training, the assistant who placed the article for a “blind search” had to be careful to not cue a handler. A lot of the individual drug dog crew’s reputation relied on the integrity of how the handler did the search:
- For a locker search, did the handler know in advance which locker was suspected of holding drugs?
- For a vehicle search, the handler has already been cued to the fact that drugs are in THAT car. Has the dog been reinforced on alerting for marijuana in an ashtray so many times that the dog begins alerting on EVERY ashtray?
It took a lot of blind searches (training aid placed by someone other than the handler). Even that was complicated. You were training in public sites (schools after hours, junkyards, etc.). Well, you could never tell what had or had not contained drugs recently.
Everything seems more clear-cut with wilderness SAR. Either you have a subject or you don’t. Until you go HRD, you’re really not looking for tiny quantities of training aid. The primary/secondary reinforcement situation is different in context to the behavior chain. Building searches in detection work magnify enormously the wilderness SAR problem of the subject being at “X” but the subject’s scent is pooled or transported by air currents to a separate/discrete location. In buildings, you’d literally have large warehouses in which the dope would be mid-building but the HVAC would lift the odor and dump it at the base of a “blank wall.” Your dog would literally alert on a wall with no windows, containers, etc.
There’s a lot written out there about behavioral training methods and how the trainer can insert behavior unintentionally (such as false alerts) through unconscious cueing. An oft-cited example is the “Clever Hans phenomenon.”
Bringing it all around to Stuart’s point… By the time a SAR team puts forth a dog crew as mission ready, the dog and handler have done so many searches that the dog has a firm feel for what it should alert to. And the handler has a good feel for a legit alert vs. the dog doing a false alert.
For SAR we’ve seen both sides. The handler thinks the dog is out of area, calls him and the dog give him/her the paw. The dog proceeds on because unbeknownst to us all the subject actually moved out of area. The same happens on the flip side. We can pull a dog off live or HR at any given moment. Stuart’s statement of course is true, the more seasoned dogs learn to ignore humans when they are on task. The trick is to find out what works for each human and each dog. All of us are support people at any given time. For searches I explain what the dog is capable of and include LE and my support person. I tell them if they see something of interest coming from the dog then tell me about it as we are all in this together.