Scent discrimination: advanced Scent discrimination: advanced technique made simple technique made simple technique made simple
By Anne Lill Kvam, dog trainer, Norway
DURING my life as a dog trainer, I have learnt a lot from the many dogs (and owners!) I have had the honour and joy to know. Some of the dogs I have trained myself, others I have supervised until fully trained by their handlers, and even more have attended some of my weekend seminars or camps. Both the dogs and I thrive most when we work with the dogs' olfactory sense. Actually, so do the owners! After having attended and given many courses around the world, where nose work has been in focus, I have realised that much is still to be learnt and explored.
Did you for example know that the dog's nose is capable of recognising molecules in such a low concentration as 10-18? Exactly like you right now, I had no idea what it meant. The researcher who told me this, explained it like this: a trained dog is able to find two grains of sand on a beach 500 metres long, 50 metres wide and 50 centimetres deep. Incredible, isn't it?!
And we are still astonished by the fact that our male dogs get restless when a bitch living three miles away is in season.
My intention with this article is to share some of my experience with you. I will explain how I do the things, and to the extent where I can find space for it, I will also make an effort to explain why I do what I do.
To illustrate my "recipe", I will use peanut oil as our training object. Our dogs will learn to pick out peanut oil from other edibles. There are people who are deadly allergic to this oil, and every time I mention it, people will tell me stories of deaths in Japan, Germany, Norway, Canada, probably all over, hence my choice.
What do you need to get started?
Many believe that the demands on the dog are the most important aspects of this training. I strongly disagree! First and foremost, there are demands on yourself that you have to cope with. You must be patient, be a little systematic and most important: have the desire to succeed. Training of dogs is great fun, but it is never only straightforward; you have to reckon with both stagnation and downhill periods, at least now and then.
If you have the qualities needed, there will rarely be a problem to find a capable dog. Either you have it in your own house, or a friend, neighbour or a colleague
will have one you can train. You can guarantee a worried owner that the training is totally free of pain, inconvenience or threatening behaviour towards the dog. This kind of training is absolutely harmless and very stimulating for dogs. You will see that the dog is going to love it.
My bold assertion is that, with this programme hereby given to you, any healthy dog can be trained for scent discrimination.
Plan your training
Be aware that it is the very first track, the very first search for hashish, the first search for chanterelle that is the most important in the dog's life. The way a dog chooses or manages to solve a given task for the first time, will be the preferred method whenever the dog meets problems solving a similar task later in life, regardless of the amount of training that has been carried out in the meantime.
Never start a training session without having thought carefully through it and made a plan. Have you not planned properly, it is better for all that you delay the training for later and go for a nice walk with the dog.
The cheapest food; priority of the senses
When working with dogs, it is important to be aware of the dog's inheritance from the wolf. Dogs have most of their instincts and behaviour intact from the time when they were wild. A wild dog (wolf) will be concerned to stay alive and will consequently wish to gain most proteins whilst spending the least energy possible to achieve this.
As we humans rely mainly upon our vision, the dog will trust its olfactory sense. Nevertheless, the olfactory sense will not be a hungry dog's first choice when searching for food. The "cheapest" food will be within range of the eyes. If nothing edible is visible, the dog will normally listen for a possible prey. When both these have failed, the dog will engage its olfactory sense. And yet, the dog will, first of all, sniff against the wind. Only after the failure of all these attempts will the dog start sniffing the ground to solve the task.
Don't you ever forget this when planning the training: if you want a solution by the scent, you have to eliminate any possibilities of solving the task by vision or hearing!
So what is "scent discrimination"? Simply put, it is to discriminate between scents, or to pick one scent out from others, like when a drug dog can recognise the scent of a particular narcotic through the smells of sweat, fuel, leather, food, tobacco, or anything you can think of.
When thinking of wild life, it is easy to imagine the dog (wolf!) being able to smell the difference between the track of an elk, a hare, a fox or another dog. Depending on the dog's motivation, it may pick one of these tracks.
If the dog is hungry and alone, chances are it will track the hare; if it is hungry and has company the choice could be the elk track. A lonely dog may pick up the track of another dog. A fed and satisfied dog may choose to do absolutely nothing!
The motivation
To ensure that your dog will work for you whenever you want, it is crucial that the dog is highly motivated to work on this particular task. High motivation is achieved through giving the right treat in the right amount at the right time. This can be as easy or difficult as it sounds.
Sometimes the dog's reward must be the very best you can give, other times it does not need more than a friendly word from you. You have to learn how to pick which achievement to be treated with what.
For myself, I follow the simple rule that I reward with the best treat available for feats that surprise, and something of slightly lesser value for normal, good feats. For the more expected feats I may only praise. To gain progress in your training, the dog (and yourself!) need a treat that is worth the effort!
Food as reward/reinforcement
Food is my treat of preference when teaching anything new. Food is the ultimate treat for all that live and eat. Any effort from an adult wild carnivore will mainly be to get food, in order to survive.
Training treats must be small, fresh and juicy, tasty and easily gulped down! I don't want a situation where biscuit chewing produces crumbs and my dog ends up searching for the crumbs on the ground while I am waiting for her to finish before the training can continue. Neither do I enjoy the situation where my dog runs off with a ball or other toy, and does not return. Rewarding with toys takes too long for teaching something new, and it moves the focus from the training to the play. Besides it will always end up with something "negative", as you have to take the toy away from the dog, either by force or by command.
My pupils are normally asked to find their dogs' "seven wonders of the world". When you have found a collection of foods that your dog really likes, you want to establish a ranking between them. Say you have sausage, ham, chicken, cheese, boiled liver, smoked fish and roast beef. Have a piece of roast beef in one hand and cheese in the other, and close your hands before you present your fists to your dog. The hand your dog pays most attention to, will contain the preferred treat. Continue
to compare the "winner" with all the others until everything has been compared, and you will have a ranking list of your dog's seven wonders of the world.
Variable reinforcement
In addition to varying the treats, I apply variable reinforcement. Simply put, it means that the dog never knows beforehand whether it will be rewarded or not. The idea is that your dog gets a treat every time in the beginning, until it starts to get a grip of what you want. Then you treat every second time, until the dog improves. When the dog starts to be reliable, you should reward every third to eighth time. This is variable reinforcement! This schedule is what you stay on forever.
If you stop rewarding, the dog stops performing. The point is, as you see, that the dog is not to know when or what will come. Thus, it will always be alert and excited, like some of us when playing the lottery. Each time your dog performs extraordinarily well, give the jackpot. This is how I do it: I will have five or six treats (sufficient for 5-6 repetitions) in my hand, or nearby. When my dog has a breakthrough, I throw what is left in my hand to the dog ... the Jackpot. It is an extra big or valuable reward to be given rarely and only for a special performance. If given too often it will lose its effect.
Shaping and clicker training
As you will see, my way of training is very much similar to shaping. My personal experience is that clicker training is the perfect choice for this kind of training. I do not discuss the method of shaping or clicker training here, as I expect you will find it described other places.
To choose the marking behaviour
Even if your dog has found peanut oil in your food, it is of no help to you if the dog is unable to tell you what it found. The dog needs a way to report (to mark) so that you understand what the dog has found and where it was found. The marking behaviour can be almost anything, depending on what your dog is searching for and in which environment. The behaviour must be trained independent of the search training.
Police dogs that are to find escaped persons will mark by barking. Civilian search and rescue dogs (at least in Norway) will mark by retrieving a particular marker that they have attached to their special collar. Dogs searching in ruins can mark by scratching or/and barking. A dog that has found a land mine will mark by sitting or lying down motionless in front of the mine. If your dog is trained to find mushrooms, you can choose sitting down, barking or to come back and fetch you.
Never mind what you choose, the behaviour must be happily executed and reliably performed before you can apply it in the search work. The marking behaviour must be trained in an exclusively positive way, free from any kind of punishment or unpleasantness. My experience is that a behaviour which is shaped is a happy choice. Another possibility is, of course, simply to wait and see which behaviour the dog itself offers.
Some practical advice
You may have realised the problem connected to training with peanut oil. It is liquid, and may not be "put" somewhere. I use clean tin cans, or jam jars into which I pour a few drops of oil. It may be wise to pick cans that are narrow as many dogs will try to eat the oil. The cans should be identical both in size and colour to keep the dog from solving the task by looking at the container.
Mark the cans to ensure that you will not mix them up, or contaminate them all with peanut oil. Self-adhesive labels and/or a permanent marker is handy. Be aware that the labels and marks are visible and have their own scent! Label and write on all the cans. If you are ever in doubt that you have spilled peanut oil on some of the other cans, clean them all and start over again. Cleanliness and order are the key words!
Ok, we have established some tools to deploy, and the training may commence! Scent discrimination is a complex discipline: there is the willingness to search, the marking part, and the part which is the knowledge of what to search for. In many cases a particular search pattern will also be convenient. So, the dog must know what to find, where to search and how to present its discovery to you. Additionally, and not least, the dog must be motivated to work for you!
The training process
Before you start, bear in mind that it is important not to overdo the training. Don't do too many repetitions in a row. My golden rule for initial training is to do a maximum of five repetitions, give a short break before another one to five repetitions. The short break may only be some seconds.
Have you reached number five without any success, reward the dog anyway (for its patience with you!) and figure out what went wrong, e.g. too many disturbances, the dog is not well, you are in a bad mood, your dog is fed or tired, or the task is too difficult.
If your dog performs on number one, give Jackpot and a short break! Always terminate the training with the best you can expect. This will establish a positive learning atmosphere, leading to a dog that happily returns to the training sessions.
Whenever you have fulfilled three to five such sessions (of one to five repetitions), take a longer break, at least five to 10 minutes. During a day you can have plenty of these training sessions, as long as you manage to give the breaks at the right time. Gearing the dog for success is YOUR responsibility!
The break
A break means that the dog is supposed to rest, relax totally, by calming down or strolling on the leash sniffing on its own by itself. Play, trick training or any kind of activities are not a break. Stop the training while the dog still wants more. Make sure you always terminate with success!
The training step by step
Having defined some tools and established some attitudes to be applied during this work, we have come to the search training itself. You may already have decided a marking behaviour and started this training. From here on, I expect that the dog to be trained is clicker trained. If your dog is not clicker trained, never mind and praise and treat whenever I write "click".
To control the progress of the training, it is a good idea to divide the behaviour aimed at into smaller, achievable and measurable steps, as I suggest here. When your dog performs 80%, which is four out of five times, then you can go to the next training step. If the dog does it right two of five times (or less), you have to take one or two steps backwards.
The steps of the peanut oil training, are: 1. Establish a positive association to the peanut oil 2. Introduce a marking behaviour 3. Adding a "negative" scent 4. Adding two negative scents 5. One by one, increasing the number of negative scents, until the number or search length wanted is achieved 6. Negative runs, which are runs free from peanut oil. This is to confirm the dog's understanding that it is only peanut oil it is to find, and there is nothing wrong when the dog cannot find anything. 7. Endurance: gradually increase number of search runs in a row. 8. Generalise: train in various environments. Try to simulate (or go to!) such environments your dog will work in later. 9. Add a cue to the behaviour, and start playing around with it. 10. Get some help, to give you unknown tasks.
Step 1
The aim is to establish a positive association to peanut oil for your dog. Hold a can with peanut oil in front of the dog, click immediately the dog sniffs at it. Gradually hold the can further away, so that the dog has to take some initiative to be able to sniff the oil. You can, for example, leave the can on the ground, letting the dog take a few steps towards it. Repeat this until the dog by itself comes up to the oil to get its reward.
Step 2
The aim is to get the dog to report, or mark its find. Ideally the scent of peanut oil should become the cue for marking.
Present the peanut oil to the dog. Whilst the dog sniffs it, delay half a second before the click, and give the cue for the behaviour required. Reward even if the dog does not sit or bark or whatever it was supposed to do. Be aware that there is a lot on your dog's mind right now! After three to five times the dog will normally start to obey your cue.
A shortcut to the marking behaviour required, can be to warm up by training the marking behaviour immediately before the search work starts. With your enthusiasm for the chosen behaviour fresh in its mind, the dog will tend to offer this particular behaviour to please you again and thus make you supply more treats!
Repeat this until the dog marks spontaneously when it sniffs the peanut oil. The word "no", or any other kind of negative reaction is totally wrong during training of scent discrimination
Step 3
Where the aim is to teach the dog that only peanut oil counts, and that anything else is to be ignored.
Present peanut oil and sunflower oil to the dog, and arrange the cans so that the peanut oil comes closer to the dog. Click for sniffing the peanut oil. The first 3-4 times you don't demand marking even if the dog is able to. Ignore any sniffing at the wrong ("negative") oil. Whatever you do, keep yourself from saying "no" or "wrong" or "different" to the dog. If the dog won't leave the wrong oil, get an assistant to hold the cans and hide the wrong can behind her back. As soon as the dog sniffs the right oil, give it the jackpot!
Repeat this until the dog sits spontaneously by the peanut oil (or marks in any other appropriate way), and ignores the sunflower oil. The most frequent reason for problems is progress that is too fast.
Step 4
Aim at increasing the dog's self confidence in picking the right oil. The task now includes three cans, of which one contains peanut oil and the two others different oils. Once again you ignore any interest towards the "wrong" oils, and click (maybe Jackpot!) when the dog goes to the right one.
NB: you have to interchange the positions of the cans for each try the dog has at them - the dog is smarter than you believe!
Repeat until the dog easily leaves the two wrong oils and picks the peanut oil, including correct marking behaviour. The second most frequent reason for problems is progress which is too slow.
Step 5
The aim is for the dog to search six cans and pick the right oil. Gradually increase the number of cans, like in step four. Most systems where dogs are working similarly have six to eight stations for the dogs to check. The cans may be in a row, or in a group or circle - find out what suits you and your dog best. Repeat until the dog easily and happily picks out the one positive from all the six (or eight if you prefer) cans.
Step 6
The aim is that the dog is to fulfill a search where no peanut oil is present, without any false markings but rather report to you that "There is nothing here!".
Present initially only one or two cans with other oil, and click when the dog leaves the last can! Gradually increase the number of cans to six or eight. Ignore false markings, but reduce the number of cans to make it easier for the dog. This is a stage where timing of the click may be critical. Only click immediately after the dog has left the cans. Little by little delay the click further, until the dog comes all the way back to you when no peanut oil is present. Repeat until the dog makes no mistakes!
Step 7
Aim at establishing a certain amount of endurance. You must decide how long you want your dog to work before a break is given. Do not demand too much, and increase the duration very slowly. Do it so that it is not only a gradual increase, but let some periods be shorter as well. -It must not appear to become increasingly worse for the dog! Be aware that every dog will have a different limitation here. Repeat until you have found a satisfying duration for a search period.
Step 8
Generalize the search work. You probably started the training in a peaceful place with little disturbance. Now is the time to remember that your dog must be able to search at other places and with other people, sounds and other disturbances present. Imagine a real life situation: peanut oil in your food will occur in restaurants and in food stores. Strive for a gradual approach to such environments, maybe you can arrange something with the owner of a restaurant or the manager of a supermarket. When your dog is comfortable about working in one environment, change to a new one.
Repeat until your dog works undisturbed in the environments that are relevant to you. Whenever training in a new environment, lower your expectations!
Step 9
The aim is that the dog will start working at a given cue. You may add the cue at an earlier stage, but you must never add the cue unless you know for sure that the dog will perform satisfactorily. My little test for whether to add the cue or not, is to prepare a training session, and when the dog performs three times in a row I give the cue the 4th time.
I really want to avoid the situation where I give the cue and the dog does not obey it - every time I call "come" and the dog does not come, the meaning of the cue loses value. That is why I am so careful about this. Give the cue in a well known environment the first few times, and go through step eight the way I described: thje first three times without cue, then the fourth time with cue, in each new environment. Repeat until the dog always responds satisfactorily to this cue. Good luck!
Step 10
The aim is for you to trust your dog. You need one or more helpers to prepare new and unknown tasks for you. The weakness so far (as you may have noticed) has of course been that you have always known the correct answer! When it comes to reality, you will never know the answer, since that is why we started to train dogs for this in the first place.
Initially, it may be a good idea that your helper has the clicker and you give the treat, but very soon you should handle the clicker yourself. The difficulty may be that you have (unconsciously) established one way or other to show your dog where the right one is, either by slowing down, drawing or holding your breath, fiddling with the leash. Repeat until your helper(s) confirm that you have learnt to trust your dog!
Further on down the road
Even though you have completed this programme successfully, you are only at the beginning of the work involved to get a fully trained peanut oil dog. Peanut oil occurs in food in various forms and concentrations. Further training will be to take the dog through these same steps, using food that contains peanut oil and food that is guaranteed to be free of it. One day you work on pastry containing peanut oil and other oils, another day pastry that is baked on plates greased with the same different oils, etc. Figure out all the different variations the oil may be presented in: boiled, fried, cooled, fresh... also bear in mind that meat or fish fried in oil may be irresistible temptations for your dog! To keep up the motivation for the search work, the reward must be as good as or better than what the dog is supposed to search for and leave without eating.
And you are never allowed to yell at or punish your dog for any mistakes. Mistakes only occur when your preparation has not been good enough. Besides, you will never really know whether your dog is right or not in its marking, will you?
Some final advice
The most common trap to fall into is to give attention to the dog when it is sniffing the wrong/negative oils. Any attention, positive or negative, is likely to be interpreted by your dog as a reward, and you end up teaching your dog to mark falsely. Attention may be pulling at the leash, saying "no", laughing, sighing. Instead of giving a negative reaction to wrong behaviour, you have to eliminate the chances of getting anything out of doing wrong, like letting the helper pull the can with the wrong oil away from your dog.
With independent work like this, you need a dog that is very confident. Both in Angola and South-Africa during my work there, I saw several good mine detection dogs being severely set back or even completely ruined, by being punished for false markings.
Accurate and correct timing of the click will be decisive for what the dog learns. If you click for approaching something, the dog's interest for this will increase. But if you click when the dog is about to leave it, the dog will learn to leave this alone. Watch your dog carefully!
Write a training diary. I do not know how many times my pupils (in many countries) and I have discovered how stereotyped our training actually is until the day we start taking exact notes about what we do. We think that we shall do it differently, but we tend to act in the same old patterns!
Many have also discovered that the same problems often occur at the same time of the year, and this too was not clear before they started to write their diary. Good luck with your dog!
If any of you readers are interested in further advice, or discussing the programme with me, I am available at my e-mail address
al-kvam@online.no- (c) Anne Lill Kvam, Troll hundeskole, Norway
