Cat and Dogs seeking solutions with sniffing canines and science

Hi Mia and Julie, 

First of all, I LOVE your blog! 

After meeting at SPARCS this past summer (summer for us in North America.. I take it summer is just beginning in Australia!), I’ve followed it closely. 

You do amazing things for the promotion of  canine science. Serious love.

A bit of background for the readers: I’m currently doing my PhD at Dalhousie University in Halifax, Nova Scotia in Canada, under the supervision of Dr. Simon Gadbois. 

Dr. Gadbois has an amazing amount of knowledge and experience in the science of sniffing (just check out Gadbois & Reeve, 2014 link below!). 

He’s trained sniffer dogs for the conservation of ribbon snakes and wood turtles, to track coyotes, and to detect invasive pests in lumber. He and I have taken on a different type of project and are studying the intricacies of biomedical detection dogs, specifically, the very interesting phenomenon of Diabetic Alert Dogs. 

Cat Reeve at #SPARCS2014 where she won the Best Emerging Researcher prize

I say interesting because there’s anecdotal evidence suggesting that some dogs alert their owners to hypoglycemic events (low blood sugar). In 2008, Deborah Wells published a series of case studies where dogs were reported as signalling (barking, licking, pawing etc. the individual) while their owners were awake, while they were sleeping, and even when their owners were in a different room with the door closed! And this is with no previous training! 

Isn’t this fantastic! Severe hypoglycemic events can be extremely dangerous for individuals with diabetes. If not treated, they can lead to seizures, comas, and even death. The fact that dogs may be able to alert an individual before a serious hypoglycemic event means less worry about hypoglycaemia unawareness, and blood sugar dropping over night when individuals are unconscious.

Given that dogs are signalling through closed doors, it is assumed that the dogs smell something that alerts them to a change in the physiology of their owner (as opposed to behavioural cues, as is believed to be the case with seizure alert dogs). There are many companies that have taken advantage of this supposed ability, and have trained Diabetic Alert Dogs (DADs) to sell to individuals with diabetes. 

In my own searches, I have found no company that publicly provides information as to how they train their dogs. However, according to recent studies (see Gonder-Frederick et al., 2013 and Rooney et al., 2011 below) these trained DADs dogs contribute greatly to the families of individuals’ with diabetes; they signal consistently and, consequently, significantly reduce the number of hypoglycemic events an individual experiences. 

Now, if it is in fact an olfactory cue that dogs use to identify a drop in blood sugar in their owners, one would expect that if you presented one of these trained DADs with the “scent” of hypoglycemia without the individual present (just like having the owner with diabetes on the other side of a door), the dog would still signal. 

Dehlinger and colleagues recently tested three DADs in a lab setting, presenting the dogs with human biological samples that were obtained identically to the way the samples used to train the dogs were obtained. In this study, none of the three dogs could pick out a hypoglycemic sample from a normoglycemic sample! DARN!

So what is it that these dogs smell? How is it that some DADs are accurate at signalling hypoglycaemia but evidently cannot signal in the lab? 

Enter ME! This is what I’m trying to figure out!

One possibility is that, at least in the study outlined above, the use of sweat samples was misguided. Maybe the volatiles are more strongly present in an individual’s breath. I am currently in the process of testing whether there is one channel more likely than others to expel the volatiles needed to detect a change in blood sugar, i.e. breath, sweat, or saliva. 

So far we’ve found that, after being trained to detect extremely low saliency scents, our dogs can detect and discriminate between human breath samples with incredible ease and accuracy (see the video of Koda discriminating between breath samples). 

Little Koda discriminating between breath samples. The jar that has the check mark on it is the target sample. The jar with the x+ is a different breath sample. When Koda first enters the room, he smells a jar that is the “sample” (it matches the target sample)... This is the smell we want him to find amongst the three jars that are closer to the camera. He chooses the right one!

Preliminary tests with patient samples suggest that our dogs can also tell the difference between breath samples taken from the same individual when their blood sugar was low, when it was normal, and when it was high (unpublished). 

Also, it’s possible that in the Dehlinger et al. (2013) study, the sample collection procedure was simply not sufficient enough to contain the volatiles produced during a hypoglycemic event. With the help of a fantastic chemist at Dalhousie (Dr. Peter Wentzell), we have perfected a procedure that coats cotton balls in silicone oil. This is believed to help contain the volatiles through the *magic of chemistry.  

*may actually be science, not magic - check with the chemists

Another potential is that these DADs are simply responding to a general stress response in the body. It’s possible that DADs (and untrained dogs who signal to their owners) are picking up on the physiological change associated with stress (changes in cortisol, adrenaline etc.). Kind of how people say animals can smell when you’re scared. 

If this is truly what’s happening, you would expect to see DADs giving a lot of false alarms. 

And in fact, personal communication with friends of friends who own DADs tell me that their DADs alert to not only hypoglycemic events, but to asthma attacks, anxiety attacks, etc. If you read carefully, few studies ask owners of DADs if their dog gives a lot of false alarm signals (dog signals to their owner and after the owner checks their blood sugar, they discover that they are not low). 

Isn’t this fascinating? 

The dogs I work with are incredible (shout out to the amazing owners that let me work with their dogs every week), and have incredible work ethic and sniffers. 

Our dogs are extremely motivated to work (because we make it super fun!). This is Nutella on her “break”. She doesn’t want a break. She wants to keep working! She’s whining and pawing at the door of the work room.

With the assistance of these amazing dogs, hopefully Dr. Gadbois and myself will be able to shed light on how exactly DADs do their job in the near future - Ill let you know how we go!


Cat Reeve
PhD Candidate
Dalhousie University
CANADA

Follow Cat on Twitter

Follow the Dalhousie Canid Behaviour Research Team on Facebook

Check out #DogsOfDal on Instagram

Further reading:

Brown S.W. & Strong V. (2001). The use of seizure-alert dogs, Seizure, 10 (1) 39-41. DOI: http://dx.doi.org/10.1053/seiz.2000.0481 

Dehlinger, K., Tarnowski, K., House, J.L., Los, E., Hanavan, K., Bustamante, B., Ahmann, A.J., & Ward, W.K. (2013). Can trained dogs detect a hypoglycemic scent in patients with Type 1 Diabetes? Diabetes Care (Observations), 36, 98-99. 

Fier, B.M. (2004). Morbidity of hypoglycaemia in type 1 diabetes. Diabetes Research and Clinical Practice, 65, 47-52. 

Gadbois, S., & Reeve, C. (2014). Canine Olfaction: Scent, Sign and Situation. In A.  Horowitz (ed.). Domestic Dog Cognition and Behavior. New York: Springer. 

Gonder-Frederick L., D. Warren, K. Vajda & J. Shepard (2013). Diabetic Alert Dogs: A Preliminary Survey of Current Users, Diabetes Care, 36 (4) e47-e47. DOI: http://dx.doi.org/10.2337/dc12-1998  

Rooney N.J., Morant, S. & Guest, C. (2013). Investigation into the Value of Trained Glycaemia Alert Dogs to Clients with Type I Diabetes, PLoS ONE, 8 (8) e69921. DOI: http://dx.doi.org/10.1371/journal.pone.0069921

Matyka K.A. (2002). Sweet dreams? - nocturnal hypoglycemia in children with type 1 diabetes, Pediatric Diabetes, 3 (2) 74-81. DOI: http://dx.doi.org/10.1034/j.1399-5448.2002.30203.x   

Wells D.L., Lawson S.W. & Siriwardena A.N. Canine responses to hypoglycemia in patients with type 1 diabetes., Journal of alternative and complementary medicine (New York, N.Y.), PMID: http://www.ncbi.nlm.nih.gov/pubmed/19040375

© Cat Reeve | Do You Believe in Dog? 2014

Read More..

Do As I Do Copy Cat Social Imitation in Dog Training

Join us for another guest post, this time from Claudia Fugazza of the Family Dog Project in Budapest. Claudias here to discuss her recent publication in Applied Animal Behaviour Science on the efficiency of new methods in dog training.

Hi Mia and Julie,

Formal training methods used until now rely mainly on the well-known rules of individual associative learning. These methods work perfectly well for a very wide range of animals — pigeons, rats, dogs and even crabs — and human and non-human animals can learn by ‘click and treat,’ as noted in the popular training book by Karen Pryor.

However, recent research has found substantial evidence that dogs could be predisposed to acquire information socially via the ‘Do as I do’ method. Do as I Do is a relatively new training method for people to use, based on dogs’ social cognitive skills, particularly on their imitative ability. 

With this training technique, dogs learn new behaviors by observing and copying their handler. The dog is a copycat. This method relies on social learning, and it was recently introduced in the applied field of dog training. 

As this method has started spreading in the dog training world, we felt that its efficiency and efficacy needed scientific testing. We were also wanting to know whether this method would be more or less efficient than other current training methods in training for particular behaviors.

We expected that dogs would more easily copy object-related actions from a human demonstrator so we tested dogs’ efficiency in this kind of tasks. To do this, I travelled across Italy and the UK with my video-cameras as well as a heavy Ikea cabinet filled with objects (you can imagine the weird looks I got from security personal at checkpoints!). I used these objects to test dogs learning to open or close drawers and lockers, pick up items from it etc. Since training methods can be affected by the skills of the trainer, only experienced dog-owners pairs who achieved a certificate either for the ‘Do as I do’ method or for shaping / clicker training were included in the study. Each pair was tested using ‘his’ method for teaching three different object-related actions in three testing sessions.

We expected that the ‘Do as I do’ method would prove more efficient for teaching complex tasks, compared to the shaping method that relies on individual learning. This expectation comes from what we know in humans: we tend to rely more on social learning when required to learn something difficult.

Our research found that the ‘Do as I do’ method proved more efficient for teaching dogs complex tasks, like close a drawer, open a locker and pick up an item that was inside (i.e., the time needed by the owner to obtain the first correct performance of the predetermined action was shorter with the ‘Do as I do’ method compared to shaping). We did not find a significant difference in the efficiency of the methods for teaching dogs simple tasks like knocking over a bottle or ringing a bell.

Now that we know a bit more on how to efficiently teach complex object-related actions, we are curious to know what happens when we want to teach different kind of complex actions, like body movements. We also want to know whether introducing social learning in dog training could have an effect on learning cues for trained action. 

We are aware that learning rates can be influenced by many factors, and we acknowledge that this study is just a very first step towards a more scientific approach to training paradigms. However we believe that this kind of information can be very important for the practitioners working in the applied field of dog training. We hope that the readers will not misinterpret the results and will not extend them to different actions and situations that were not tested.

Furthermore we would like to emphasize that, despite being efficient for training some kinds of actions, the ‘Do as I do’ method does not replace the methods based on individual learning (for example think of how many actions are not imitable at all if the demonstrator is a human and the learner is a dog!). Instead ‘Do as I do’ is a useful (and fun!) addition to existing training paradigms. Experienced dog trainers may find effective ways to mix the different training techniques in order to obtain the best results with each dog. 

Claudia Fugazza
Do as I Do Book and DVD
http://www.apprendimentosociale.it/en/claudia-fugazza/
Family Dog Project 

Reference 
Fugazza C. & Miklósi Á. (2014). Should old dog trainers learn new tricks? The efficiency of the Do as I do method and shaping/clicker training method to train dogs, Applied Animal Behaviour Science, 153 53-61. DOI: 10.1016/j.applanim.2014.01.009

© Do You Believe in Dog? 2014
p.s. Check out this dogs excellent jump!

Read More..