Wednesday, 3 October 2012

Canine Neuroscience

The main problem with the neuroscience of dogs is that they would have to be sedated to be in the scanner, and then their brain wouldn’t be doing its normal stuff. Until now.

A team of scientists led by Gregory Berns at Emory University has successfully trained two dogs to go into the fMRI scanner and keep still long enough for a brain scan. Prof Berns says he got the idea from realizing what military dogs are trained to do – if a dog can parachute out of a plane with its handler, he thought, then surely it could do an fMRI.

The dogs are Callie, a two-year-old feist (small hunting dog), and McKenzie, a three-year-old border collie. And while McKenzie does agility, Callie had only had basic obedience training, and is a rescue dog. (If anyone ever tries to say negative things about rescue dogs, tell them about Callie!).

Dogs can haz brainscanz?

The dog training was complex and took place over several months. A mock-up of the scanner was made for each dog, including a replica head coil, a tube of the same size as the real thing, and a patient table within the tube with a head rest for the dog. Because dogs’ hearing is so sensitive, they also had to learn to wear headphones to protect their ears. The first times the scientists actually started the scan, the sound caused a startle reaction, so they played the sound throughout the training sessions to get the dog used to the noise that the machine makes.

The dogs were trained using positive reinforcement, and had to lie still with their head in a chin rest. Once the dog was used to this, the chin rest was moulded to give a custom fit, because it’s important for the dog’s head not to move in order to allow the scan to be interpreted. Finally, they were ready for the real thing.

What task was used for this ground-breaking experiment? The researchers wanted to know if there is a different pattern of activity in the brain when the dog sees a hand-signal that means their owner is about to give them a treat, compared to when they see a hand-signal that means no treat. The hand signals were left hand held up to mean the dog was about to receive a piece of hot dog, and both hands pointing horizontally to each other to mean no hot dog was forthcoming.

The dogs were trained on this task for ten minutes a day for several weeks prior to the actual experiment. The dog had to hold still for ten seconds while the hand signal was made, and then if it was the ‘treat’ signal the owner reached into the scanner to deliver some hot dog; the dog could move to get the treat and then had to go back into position.

Seven border collies posing for a group photo
Do you think these border collies are anticipating a reward?
 
Previous work on people and monkeys has shown that the caudate region of the brain is activated in expectation of a reward. The scientists found activation in the ventral caudate region of the dogs’ brains in response to the hand signal that denoted a treat, but not for the other hand signal; this shows the dog had learned to expect a reward.

This study is exciting because it demonstrates that canine neuroscience is possible. A future study might look for any differences in activation depending on whether it is the owner or someone else who makes the hand signal. This would tap into the same wider debate as that considered by Feuerbacher and Wynne in their investigation of whether dogs and wolves prefer treats or social interaction as a reward. 

The experiment was conducted with great care for the dogs’ wellbeing and they were free to exit the scanner at any time. The researchers make some useful comments about the ethics of this kind of study; for example, they say it wouldn’t be appropriate to use laboratory-bred dogs as they would have no choice whether to take part in the experiment. 

Emory University made a short video about the research which you can see here.

If you could train your dog to go in an MRI scanner, what would you test?

Reference:

Berns GS, Brooks AM, & Spivak M (2012). Functional MRI in awake unrestrained dogs. PloS one, 7 (5) PMID: 22606363

5 comments:

  1. Interesting study, though I'm pretty sure I could have predicted the results of the reaction to hand signals. My dog knows without a doubt what signs and sounds (clicker) mean a treat is coming. He's not just a pretty face, for sure!
    What would I test for if Eddie could be trained to undergo a brain scan? Probably whether his brain turns to fluff every time he sees stairs, hallways, or elevators (he's terrified of them) - which probably doesn't bode well for ever training him to submit to an MRI!

    ReplyDelete
    Replies
    1. I would have trouble getting one of my dogs to stay still for the duration of the hand signal - he would be so excited knowing that a piece of hot dog was coming his way! So I could tell for mine too. But I think the important thing is that it shows the brain scan was working; so it had to be a simple test to begin with.

      Poor Eddie, I hope he doesn't have to do stairs or elevators too often! I think it's fairly common for dogs to be frightened of them, especially if they've never come across them as puppies.

      Delete
  2. The major goal of our initial Emory study was to prove methodologically that we could obtain quality fMRI images of dogs without the use of sedation or restraints. Therefore, we purposely selected an experiment with predictable results. Thus, if our findings, as confirmed by the fMRI images we obtained of the two subject dogs, duplicated the outcomes of prior research that indicated an expected anticipatory reward response in the caudate, then we substantiated our belief that we could successfully produce valuable neuropsychological research that incorporated highly humane animal handling and training protocols.

    To obtain a quality fMRI brain image, we needed to train the animals to remain virtually motionless for a period of up to 30 seconds. Cranial movement as small as 3 mm in any spatial plane would create archetypal noise that rendered an image useless.

    Prior to our study, to create static positioning, animals subjected to fMRI were either anesthetized or restrained by pillories, braces, straps, and/or surgically implanted halos. However, our Emory team desires to study canine brain activity, cognition, and emotion. Sedation precludes our ability to accurately study cognition. Restraints create an anxious and/or fearful emotional state that inhibits our ability to accurately study emotion during normal cognitive processing. Consequently, to meet our research goals and our humane methodological objectives we needed to use solely positive reinforcement to train our subject dogs to remain motionless within the fMRI tube and coil. Concurrently, we also designed customized anatomical chin rests for each dog that provided comfort for the animals while facilitating greater static positioning.

    Although our research findings of canine caudate activation upon the anticipation of a reward were not unique or surprising, the methodology we incorporated was groundbreaking- and we hope trendsetting. Now that we have proven the validity of our methodology and the value of humanely treating animals participating in fMRI research, we look forward to producing novel research that provides groundbreaking information regarding canine brain activity, sensory processing, cognition, and emotions. Stay tuned!

    ReplyDelete
  3. Thanks for your comments, and for explaining how previous studies were done. It's definitely much more humane to train the dogs to keep still, and I am so impressed that they were able to do this. 3mm is such a small amount of movement, so it really shows they were absolutely motionless. I can't wait to see what you do next!

    ReplyDelete
  4. Hello! I just wish to supply a enormous thumbs up with the great information you may have here for this post. I am returning to your blog for more soon.

    ReplyDelete