Trial Times

Why is it important to notate the length of each trial?

I don't know.  But it is collectible data.  And at some point in the future of the project as we are compiling the data and making hypotheses, it may be important.  So we collect it now.  Rather than wishing in retrospect that we had.

The time each trial takes the dog to complete is impacted by a wide variety of conditions:  time spent searching for the scent, time spent searching for the treat, time spent turning the wheel, time spent resetting the dog, time spent on duration of the indicator.  The number of repetitions. And the obvious, the strength of the scent.

Just as it is human nature for the trainer to want to see rates of success of 90% and 100% on each trial, it is also human nature when the trainer wants to see the dog complete the trial efficiently, confidently and quickly.

But that's not necessarily the name of the game.  And it doesn't necessarily indicate a mastery of the job at hand.

So, whenever possible, we record the length of each trial.  And trust.

Variables vs Distractions

This is how I define these two terms in light of testing the viability of the seizure scent.

If I am testing a sample that is on a hydrated cotton square and if I am testing on a six container scent wheel, then the SS sample is in one container and hydrated cotton squares without SS are in the other five.  One Sample.  5 Dummies.  Variables.

If I am testing a sample that is on a dry gauze pad and if I am testing on a six container scent wheel, then the SS sample is in one container and dry gauze pads without SS are in the other five.  One Sample.  5 Dummies.  Variables.

If I am testing TWO SS samples on a single scent wheel, then the remaining four containers on the wheel contain unscented sample dummies.  Variables.

I might even conduct a test with SS sample in one condition and have the dummies have different conditions.  A hydrated SS sample on cotton and dummies of hydrated cotton, dry gauze.  But all would be SS sample related.  This might be a test to tell if the dog is alerting to SS or to wet cotton, for example.

In my mind, this is testing the scent.  This is the job of Phase 1.

When it comes to distractions, I include kibble, toys, anything else that is different from sample in my definition of distractions.  In my thinking, teaching the dog to search out the scent from other optional scents is Phase 2: Teaching the Dog to Search and Locate.  To Monitor and Locate.  But it is separate and apart from testing the scent itself.  No less valuable.  No less important.  But different.  Phase 2.

In Phase 2 we are using the scent that we know is viable because we determined its viability in Phase 1.

It's a fine line, I know.  And I may be totally off base and misguided in my thinking.  Diabetic Alert Dog trainers know that the scent they use is viable.  They have even measured it and tested it against their BG monitors.

There is no way to measure seizure scent and there won't be until, as a result of our Phase 1 testing of SS sample, it is used to train several Seizure Alert Dogs that are working successfully.

Phase 1 is the first step up the path.

CAVEAT:  Before the testing of live SS begins, the dog must first be taught how to play the "find the scent" game.  And, be introduced to the Scent Wheel.  During this training process it is assumed that distractions (as defined above) will be used so that the dog understands all the parameters of the game.  The criteria of switching from training mode to testing mode should be a) eager enthusiasm for the game; b) being able to find and alert to 3 different scents on the scent wheel; c) clear and consistent alerts.

Testing the Scent Sample

Robert Milner speaks to our challenge .....

"It sounds like you are doing a great job.  If I were going to train a dog to alert on impending seizures I would first try to determine whether a seizure alert dog is responding to an odor change or to a minute behavioral change in the patient.

If it were odor, I would want to know what the odor was and what its persistence was before I made the huge assumption that the clothing worn during a seizure was a valid target odor upon which to train a dog to alert. 

The easy part is that of training the dog.  The hard part is figuring out what target to train him on."

Read the entire conversation with Robert Milner found in the Research Index.