Internationale Reddingshonden Groep

Downloads

RESEARCH ARTICLES

This research was presented on the international symposia of Boulder 1995 and Vienna1997.

SCENT DISCRIMINATION TESTS IN THE TRAINING OF SEARCH-DOGS.

JAN BOGERS,

International Rescue dog Group (IRG)

5311AD, Gameren, The Netherlands.

ABSTRACT

In the years that we have trained and examined search dog for searching human victims we have noted that scent training is a weak point. Only few handlers actually know and care about the scent-complex their dog is actually working on. No scent discrimination tests are included in the training of the dogs. In training search dogs for our organization we developed a training scheme that is based on an exact definition of the work the dog has to perform when it is fully trained. Consequently we make a choice of the scent-complex the dog has to work on. By a scent-complex we mean in this context a collection of scents. As it turned out to be a big problem to isolate some and certainly one scent from this scent complex, we choose for the opposite way of approaching the problem. We call it the exclusion principle!After a dog has been trained to do a particular job, for example, searching for living victims, we do exclusion tests to be sure that he is working on those parts of the scent-complex we want him to work on. During these exclusion tests the dog is tested on scents coming from dead animals, dead persons, clothes, food etc. Often we do these tests during the exercises in the rubble, but regularly we do those tests in separated rooms where we can see sharply if the dog gives an alert on a certain scent or not. These tests also have the advantage that we can see if the dog is influenced by the handler. During these tests the dog has to make a decision by himself if he is going to give an alert or not.

1 INTRODUCTION

Holland is a country with about 16 million inhabitants. It has a population density of about 460 inhabitants on a square kilometre. It has 12 provinces, 5.5 million cars, 6 television stations, at least 13 political parties, some 10 rescue-dog groups, at least 3 police organisations having search-dogs, about a 150 rescue-dog handlers all being an institute in themselves. However, there are only about 15 operations each year requiring search dogs of all kinds, water-search dogs, surface search dogs and disaster search dogs.!

All these handlers have a different opinion about the training and the ability of their dog to go to an operation. And moreover these rescue dog handlers are extrovert: they all want to tell you their opinion explicitly! Especially if there is a call for search dogs in an operation they all want to go without asking themselves if they have a dog that is actually trained to do the job!

In the years that we have trained and examined search dogs for searching human victims in Holland, we have noted that at least, scent training is a weak point.

Only few handlers actually know and care about the scent-complex their dog is actually working on. No scent discrimination tests are included in the training of the dogs. In this paper we want to present a method to incorporate scent discrimination tests in the training schedule. It is written as a follow up of the symposium paper we wrote for the 1995 symposium in Boulder Colorado [1]. We ended the Boulder paper with the following conclusions.

1.        It has strong preference to train a SAR dog for a specific task, with its specific scent or scent complex, and not for a large scale of tasks! This will avoid unnecessary confusion and problems.

2.        If a SAR dog is trained for more disciplines with a possibility of confusion then we have to pay much attention to this problem in the training of this dog!

3.        SAR dogs trained to search for living victims should be trained with living victims only! Furthermore they should be tested regularly if they do not give false alerts on object carrying human scents or on dead victims.

4.        SAR dogs trained to search for dead victims should be trained on human tissue only! Also these dogs should be tested if they do not give false alerts on object carrying human scents.

In the time that followed the 1995 symposium we have spend a lot of time and energy to build a training scheme that was according these conclusions. Special attention was paid to the conclusions 2, 3, 4. For details see Bogers [2].

We will restrict ourselves to the work of search dogs which are trained to search for humans scents. The training of all these dogs has one thing in common:

They learn to work on a collection of human scents also called a human scentcomplex.

However we all know that human scents may come from more things than from a victim!!

And we all know that in a disaster situation there are other biological scents that can be misleading for the dog. Here we present a method to cope with possible problems.

The method presented is not a final stage but a reasonable framework we use in training dogs for search work to discriminate properly between several scents and to choose only that scent our scent-complex that is relevant for their work!!

In order to discuss the subject we choose in this paper for the following approach.

In Section 2 we will present the scent complexes that belong to an important example: The work on dead and living human victims. Next, in Section 3, we discuss some aspects of scent complex that is chosen by the dogs when his training is in a beginning phase.

In Section 4 we introduce what we call the exclusion principle by discussing scent discrimination tests. Finally, in Section 5 we will discuss some common problems that might occur.

2 THE SCENT-COMPLEXES

In this section we define the most important concepts used in the sequel. Many misunderstanding is caused by bad definition of the words used. It is therefore of major importance to focus the discussion on essential points and not to let it get stuck in a discussion about the meaning of words.

At first we will tell what we mean in this context by scent and give a definition. Then we will define the important concept of scent complex and give tables of the most important scent complexes met by a SAR-dog. Some of the definitions given in this Section may be formal but for a good understanding of the discrimination tests which will be dealt with in the sequel it is absolutely necessary to know these facts. See for details Bogers [1] and [2]. We start with the definition of scent.

           Scent is a mixture of substances in a gaseous state (vapours).

As all substances this mixture consists of molecules, which are the smallest parts of a substance still having the physical and chemical properties of the particular substance. Because we are talking about a mixture it will be obvious that there are several kinds of molecules in scent.

Very obvious examples of scents are for instance the smell of sweat, perfume, soap, the smell of rubber (rain) clothing etc.

We all know of course that not all substances are producing equal amounts of scent. Several substances we can smell very easy whereas other substances we can hardly smell.

The ability of smelling a scent depends not only on the amount of scent produced but of course also on the nose capacity of the creature smelling. Making for instance a comparison between man and a dog we see that a dog has 225,000,000 nerve cells on a surface of 150 cm², whereas a man has 5-7,000,000 nerve cells on 5 cm². From these figures it should be obvious that a dog has a far greater capability in smelling scents than man. For more details see De Bruin [3] and Schoon [5]. We now give a definition of a scent complex.

            A scent complex is the collection of scents on a particular location.

This definition contains a restriction regarding the location. This is for the training of the search dog of great importance as we will see in the sequel. An example of a scent complex is for instance the collection of scents met by a rescue dog trained for rubble search.

In table 1 we give a general scheme of a human scent complex. This scheme includes the scent complex of the most common types of work done by rescue dogs, i.e., rubble search (living and dead), area search (living and dead). For all the search work done by dogs it can be advisable to make a table of the scent-complex a dog meets when he is doing his job. It can be a great help in solving problems!

Table 1: THE SCENT-COMPLEX

The complex given in table 1 is quite general. The distribution of the categories I-IV over the scent complex varies a lot depending on all kind of circumstances. We give some examples.

-The complex belonging to a child under 10 years is quite different from an adult (no apocrine and no sebum).

-The complex belonging to a person from china will be different to a European (hardly any apocrine).

-The complex belonging to a dead person will be different to the complex belonging to a living person. For details see De Bruin [3], Schoon [5] and Stoddard [6].

Also the surroundings where the dog has to perform its work is of great importance! One has to think about the differences between the scent complex belonging to an area search and a rubble search!

Hence it is necessary to make a subdivision of the complex presented in table 1 depending on the specific work the dog has to perform. As an important example we give the subdivision between dead and living victims.

Before we give the tables with the scent complexes, we will start by stating some facts about a living victim. Substances which on the body of a living victim are heated depending on the temperature of the body. The temperature on the outside of a living body is constant and " 27⁰ C. Due to this body temperature the victim will constantly generate a particular biological scent. At this temperature certain substances on the body will vaporize and others will not. Of course volatile substances vaporize sooner that non volatile substances. Also belonging to this temperature there is a particular bacterial metabolism on the skin of the victim. Last but not least a living person is breathing and sweating. For more details see Schoon [5] and Stoddard [6].

Table 2 summarizes these facts. We only give the categories I and II as these are of major importance and categories III and IV are the same for both living and dead victims under the condition that they are located in a similar situation.

Table 2 The scent complex from a living victim

Turning to the discussion of the scent complex produced by a dead victim we see immediately that there is one very important difference between a dead and a living victim: A dead victim is not breathing nor sweating!

If the outside temperature is lower than 27⁰ then the temperature of the dead body is decreasing at a constant rate depending on the temperature of the surroundings. Due to this temperature change, the victim will generate a changing scent complex until a stable temperature is reached. This is the temperature of the surroundings. At this temperature fewer substances on the body will vaporize.

After two days the destruction process of a dead person starts. Again this depends strongly on the outside conditions such as temperature, humidity, etc. In this process bacteria and insects play an important role. There will be a different bacterial metabolism on the victim compared with the bacterial metabolism on a living victim. Anal gases are produced.

If the outside temperature is higher than 27⁰ then the process of destruction will start almost directly again causing the scent complex to change rapidly!

Of course during the process of destruction the scent complex does not remain the same. It will undergo some major changes in time. However to clarify our major point of discussion it is sufficient to restrict ourselves to the facts stated above. For more detailed discussion see N. Sharp [4] and Stoddard [6].

Table 3 summarizes these facts. Again we restrict ourselves to categories I and II.

Table 3. The scent complex coming from a victim that is dead for about three days.

It will be obvious that the scent complex follows a trajectory from table 2 to table 3 from the moment of death to about three days after death, but again this strongly depends on the conditions.

One thing is absolutely discriminating between living and dead victims, that is the fact that at moment of death breathing and sweat production stops!!!

Coming back now to the conclusions 3 and 4 we made in Boulder Colorado 1995 it will be obvious that in order to train a dog to search for living victims we should try to focus the scent complex on which this dog is working, to those aspects that are specific for this job.

One way to do it is to find a certain scent from the complex of a living victim and to train a dog on this scent (in the case this could be breath) .

This method is often used in the training of search dogs for other disciplines for example explosive-search dogs or drugs-search dogs. As we do not yet know for certain what scents a dog trained on human scent actually smells, this method contains a risk and may be in practice hard to realize. Imagine a victim breathing and not sweating or a victim not breathing but still be alive and sweating!

That is why we often choose another approach to this problem. After a basic training of the dog (in case of a dog searching for living victims only this will be on living victims), we start doing so called exclusion tests. By these tests we are excluding those scents from the scent complex the dog is working on, that could possibly give confusion!!

3 THE CHOICE OF A SCENT-COMPLEX

From the scent complexes introduced in section 2 it will be obvious that a dog has a large amount of substances he can choose from when his work is on human victims. After some years of doing the exclusion tests it turned out that the dogs actually makes a choice from the scents that are in the scent complex!!

In most cases (in training SAR-dogs) the basic training of a search-dog on human victims (dead our alive) starts with working on living victims.

Looking now at the scent complex belonging to a living victim (table 2) it is obvious that a dog trained with living victims can choose from the scent-complex those scents that are specific for a living person or he can choose scents that are both on a living and on a dead victim (compare tables 2 and 3).

After the basic training we do not know for sure what complex the dog is working on! The dog made his own selection of scents. The only way to find out is to do a test and see if the dog is alerting on scents that tables 2 and 3 have in common or if he is alerting on those scents that are in table 2 and not in table 3. However, as stated before it may be hard to isolate the scents.

We choose for the following approach. Dogs being trained to search for living victims are tested on those scents that possibly could cause confusion, like objects carrying human scents for example clothes, recently used beds, shoes etc. Also they are tested on scents coming from dead victims and those coming from dead animals.

Dogs being trained to search for dead victims are tested on those scents that possibly could cause confusion for this type of work, like objects carrying human scents, clothes, recently used beds, shoes etc. Of course they are also tested on scents coming from dead animals, like pigs, chicken etc.

In the beginning when we start doing these tests we regularly see dogs that give false alerts on misleading scents. Later on we see that the dogs get better and better focused.

Specially dogs that are trained for more than one job need special attention! For instance, tracking and searching for living victims often causes problems and searching for living and dead victims can be a cause of problems also!

That is why we came to the first conclusion in Boulder 1995 which stated:

"It has strong preference to train a SAR dog for a specific task, with its specific scent or scent complex, and not for a large scale of tasks! This will avoid unnecessary confusion and problems".

We have two goals by doing these tests, one is to see what scent-complex the dog is working on, and the other is to correct, if necessary, the scent-complex of the dogs own choice in the direction we want.

4 ABOUT THE TESTS

In this section we will give a description of the setup of the tests described in Section 3. As stated in Section 3 we start with a basic training.

Suppose we have a rubble-search dog to search for living victims only. As stated in Section 3 we give this dog a basic training on living victims. This training is done until we have a reliable alert.

After this phase in the training we give the dog a first test on two scent-complexes. Here we give as an example a test between a real living victim and Pseudo-Corpse scent formulation I.

Hence, one scent-complex is coming from Pseudo-corpse scent and one is coming from a real living victim.

For this purpose we create three or mostly four hiding places in four separate rooms. In one hiding-place there is Pseudo-corpse scent whereas in the other hiding place there is a living victim. Mostly we introduce one or two empty hiding place as well!

If the dog is giving an alert on the hiding place with wrong scent-complex (in this case the Pseudo-corpse) then we open up the hiding place, show the dog that there is nothing interesting in it and consequently there is no reward.

This dog is now being brought directly to the room with the victim and if he gives an alert on the victim he is very strongly stimulated and given a toy by the victim.

In this way the dog is taught to choose for the scent-complex coming from the victim by motivation (caused by the reward and stimulation).

During these tests the handler does not know in what hiding place the victim is. It is forbidden for him to give any signal at all to the dog before the alert (This often turns out to be a problem in itself). Also the handler has to wait until the dog smells the scent coming out of the hiding place. Consequently, the dog has to make the choice all by himself.

The toy that is carried by the victim is always in a scent-proof glass jar. In the table 4 below I have noted some of the results of this test.

Table 4. First test with PS-I.

From these results of the first exposure to the Pseudo Corpse formula-I scent-complex we see that there are some dogs that chose a scent-complex to work on which is to wide. They need extra focussing on living victims.

Also you see a fear reaction by my own dog Chaoot (The name is not chosen randomly). This is often described in reports and in the literature. Of course I hurry to tell you that at the second exposure to the Pseudo Scent-I this reaction was gone. Also in working with the "real thing" we had no fear reaction anymore. Still this shows the importance of these tests!!

Suppose now we train a rubble-search dog to search for dead victims. As stated in Section 3 we give this dog a basic training on the scents coming from a dead victim. Again this training is done until we have a reliable alert.

Again when this phase in the training is passed we give the dog a first test on two scent-complexes.

Here we give as an example a test between a scent-complex coming from clothes that have been worn and a scent-complex coming from a dead victim. We use the same setup as with living victims.

In one hiding-place there are scents coming from clothes whereas in the other hiding place there are scents coming from a dead victim. Again we introduce at least one empty hiding place. All these hiding places are in seperated rooms. If the dog is giving an alert on the hiding place with wrong scent-complex (in this case the clothes) then we open this hiding place, show the dog that there is no-body in it and of course give him no reward.

In this case the dog is being brought directly to the room with the hiding place containing the correct scent-complex (dead victim).

If he gives an alert on this scent then he is very strongly stimulated and given a toy which is in a scent-proof glass jar. In table 5 below we give some results of one of this tests.

Table 5. First test with clothes.

In this test there were two dogs, Yellow (No this is not a scary dog) and Bongo (This is not a drum-set but a dog), which had just finished their basic training. Whereas Aschwin is a more thouroughly trained search dog for dead victims. You can directly see from the results that Yellow and Bongo need some extra focussing.

After the basic tests with two scent-complexes we can introduce tests with more and different scent-complexes.

We in our organisation (IRG) mostly introduce one special aspect. During the first tests the handler needs to be absolutely silent in order to let the dog do his job by himself. The alert has to come from the dog. After the dog has reached more maturity and a good contact with the handler the handler starts actually showing with his hands where the dog has to check for scents. When the team is working good, which of course all our dogs do, the dog will only give alerts when he smells the correct scent-complex!! Consequently we accomplish by these tests actually two goals.

-We can focus the scent complex on which the dog is working to that complex he needs to do his job properly.

-We get a more reliable and independent alert of the dog.

5 PROBLEMS

In this Section we will discuss some common problems in the work of SAR dogs. Every problem will be analyzed by using the scent-complexes presented in the tables above!

Problem 1. A dog that is trained to do tracking and searching for living victims at the same time is giving false alerts on for instance clothing or recently used beds.

Analysis. Obviously by the combination of these two jobs (tracking and searching for living victims) this dog has chosen those scents from the scent-complexes, that these two jobs have in common. One way to solve this problem is to focus this dog, by exclusion tests, in his work on living victims on the breath of the victim!

Next we discuss a problem that very often occurres in the work of rescue dogs when they work on both dead and living victims!

Problem 2. A dog that is trained to do water search (dead victims) by training him on the biological scents coming from hair and fatty acids, is giving false alerts in rubble search, on object that were touched recently by persons, clothing or recently used beds.

Analysis. Obviously this dog is trained on those biological scents that both table 2 and 3 have in common, i.e., dandruff, skin pieces, sweat from feet, fatty acids. It will be obvious that these biological scents also can be found on clothing and recently used beds etc. To solve this problem one should search for those biological scents that table 2 and 3 have not in common. In case the dog is trained to search for living victims this is not to difficult. You should focus the dog on breath!!

It will be obvious that this can cause serious problems. In our organisation (IRG) we never do these two jobs with the same dog at the same time!

Only in area search we allow a dog to do two jobs at the same time, living victims and dead victims because in this situation it is easy to verify the alert! However these dogs will never be used in a disaster situation because finding the survivers is item number one!! Only if one wants to find the dead victims we will use these dogs after the dogs for living victims have done their job!

Question. Is it possible that a dog that is trained to search for living persons gives alerts on dead victims whereas it is trained to give no alerts on clothes, beds and other objects carrying human scents?

Answer. In order to give an answer to this question we must look at tables 2 and 3 in order to find those scents coming from a living and a dead victim that are not on clothing and beds. We see that if the victim has died recently, then the bacterial metabolism has not yet radically changed. We already stated that the change in the scent complex follows a trajectory from table 2 to table 3 from the moment of death to some days after death. If the dog has picked the scent produced by the bacterial metabolism out of the complex itself or when it is trained to work on this scent, then it is possible that he gives alerts on recently died victims still producing this scent. Consequently training on sweat vapor (eccrine and apocrine) and breath does not exclude this possibility.

The only way to be absolutely certain what scents this dog is working on is to do exclusion tests on clothes and living victims and on clothes and recently died victims (Pseudo-Corpse-I).

6 CONCLUSIONS

-After a basic training a search-dog on human scent-complexes turns out to have made a personal choice from the scent-complex.

-The only way to find out what this personel choice is, is to do scent discrimination tests.

-By these tests it is possible to focus the scent complex for a specific job!

-Furthermore it is also possible to see if the dog gives good alerts without any influence of the handler!

REFERENCES

[1] Bogers J.,1995:Scent-Complexes in the training of SAR-dogs, Boulder Colorado USA, International Rescue-dog Symposium. USA.

[2] Bogers J.,1996:Scent-Detection of the SAR-dog. Reader of the International Rescue-dog Group (IRG), Gameren Netherlands.

[3] De Bruin J.C., 1988: De Speurhond en de Wetenschap. Rotterdam. Netherlands

[4] Sharp N.,1993: Research into Dead Body Detection. Notes presented on the fifth International Symposium on Rescue Dogs, Sweden 1993.

[5] Schoon A., 1997:The performance of dogs in identifying humans by scent. Doctoral dissertation University of Leiden. Leiden. Netherlands.

[6] Stoddard J., 1990: The Scented Ape. Cambridge University Press, Cambridge.UK.