REFERENCES: AFI 31-202 USAF Military Working Dog Program. PART I
LESSON OBJECTIVE: The objective of this lesson is to enable
security police dog handlers to become thoroughly familiar
with the principles of conditioning.
DESIRED LEARNING OUTCOMES: Each student will:
1. Identify the basic needs of a dog.
2. Discuss the dogs sensing system.
3. Describe reward training.
4. Identify and discuss reward schedules.
5. Describe avoidance training.
6. Describe escape training.
7. Describe learning acquisition.
8. Discuss stimulus control.
9. Discuss stimulus generalization.
10. Identify and discuss conditioning techniques.
11. Discuss the utilization of learning.
Lecture/Discussion. Discuss the
principles of conditioning after having outlined the basic needs, senses, and many types of rewards that exist in the animal world. Draw the students into the discussion by giving them some hypothetical situations and having them decide the appropriate reward or correction necessary. A test will be
administered at the end of this class. After testing, critique the test and entertain any question the student may
have.
DEVELOPMENT
1. BASIC NEEDS. A dog responds to its environment in order to satisfy its physical and social needs. In order for a dog to be able to learn, or perform a task, it is necessary to have these needs met. To gain its complete attention, the handler must help insure the dog has all basic needs met, which in turn, builds rapport between dog and handler. Several of the basic needs are used to reward performance.
a. OXYGEN. Breathing to obtain oxygen is perhaps the strongest drive. Exercise or excitement creates an
increased oxygen requirement which causes panting and hinders the olfactory (sense of smell) tasks.
b. WATER. This must be provided to the dog in adequate quantities to prevent thirst from interfering with
learning or task performance. Water cannot be used as reward in military dog training.
c. PREY KILL. This need incorporates chase with the dog manipulating a decoy by biting and tearing and can be used as a reward for task accomplishment through chase/bite agitation training.
d. FOOD. This must be supplied in adequate quantities to prevent hunger from interfering with task performance. Food sometimes is used as reward to teach dogs to perform drug or explosive detector tasks.
e. SOCIAL NEEDS. A period of socialization between dog and handler is required in order for vocal and petting praise to acquire reward value to the dog. This socialization must continue for praise to effectively maintain the desired behavior from the dog after training.
f. PAIN AVOIDANCE. The need to minimize pain enables a trainer to teach the dog not to make specific responses. The use of physical correction does not, however, teach the dog the correct response to any specific cue.
2. SENSING SYSTEM. A dog perceives its environment through the use of its basic senses. The dog senses its
environment much differently than humans. A lack of knowledge of the dog's sensing a system that causes much of the misunderstanding in the dog's training and use. The following is a discussion of the companion working dog's sensing capabilities and limitations.
a. JUST NOTICEABLE DIFFERENCE (JND). Before we can accurately describe the sensing system of the military working dog, we must first explain JND. JND is the animal's ability to detect slight changes or differences pertaining to a specific sense. ALL sensing systems are included in this definition. Examples of JND include: the smelling of a weak odor as opposed to a stronger odor; the hearing of low
volume noises as opposed to louder volumes; the feeling of light pressure as opposed to progressively heavier pressure,and so on. Again, JND applies to all senses of the animal.
(1) A JND in odor concentration is the input a dog uses to determine the strongest odor, such as a human hiding
or a substance the dog has learned to search for. All dogs have limitations. There first must be certain amount of odor concentration for the dog to detect at all. If inadequate concentration is available to the dog it will be unable to locate the source(s). This minimal level is referred to as the dog's absolute threshold, which is a just noticeable difference between no detectable odor and enough for the dog to recognize. The absolute threshold is a fixed amount of odor that is necessary for the dog to recognize it. The absolute threshold of one dog can be different than that of another; thus, one dog may detect an odor that another dog missed regardless of how well the other dog is trained. In other words, a dog's absolute threshold is the point of recognition of an odor the dog has learned to detect and respond to. Other odors are also detected but will be ignored. NOTE: There is an exception: if the odor is very similar to the one the dog learned to respond to, the dog may "stimulus generalize" and make a like response to a similar odor.
(2) The dog learns in order to obtain a reward, it must find the next higher JND and then the next and the next until it reaches the source of the odor, or the highest concentration. The dog will move about, sniffing,
until a difference is determined. When the difference is lower, the dog will change direction and continue to search for a JND which is higher.
(3) When the dog locates a higher JND it will search for one still higher and so on until it reaches the
source or the point of saturation. There may be more odor present, but since there are no more JND's the dog is
literally at the end of the line. To ask more of the dog would be asking it to perform beyond its capabilities. If you continue, you will be extinguishing the behavior the dog originally learned to perform.
(4) The dog must receive its reward when it has gone as far as it is capable of going. Likewise, if the dog
reaches a physical barrier, such as a door that can't be opened, the reward should be given. Once the dog reaches
the extent of its capabilities, it has completed the task and must be rewarded!
(5) In the first sentence of this portion of the lesson we said that "Just Noticeable Difference" applies to all the senses. The same is true of the "spectrum". It can be used as a visual, sound (volume or pressure) or any other spectrum. You can go through the entire list of basic senses and apply the concept of absolute threshold, JNDs and the point of saturation. By omitting all references to odor and replacing them with terms pertaining to another basic sense you have an explanation of how this concept applies to all basic senses.
(6) It is important to note threshold shifts occur when the sensory system is activated. This means that once
the dog's sense (remember it applies to all of them), has been activated a certain recovery period is required before the dog will be able to detect a smaller amount of odor, light, or sound, etc. The greater the amount of odor, light, or sound, etc., that the dog is exposed to, the longer the recovery period before smaller amounts can be detected. An example of this effect is demonstrated by going from bright sunlight into a dark room. After a period of time you can see objects in the room you could not see when
you first entered.
NOTE: JND SUMMARY: JND applies to all of the senses shown.
Your knowledge of JND in relation to these individual senses should explain many of the misconceptions you may have had in the past.
b. SIGHT (VISION). Through conditioning, the dog learns to respond or ignore visual cues in its environment.
Cues such as brightness, shape, size, and motion provide information to the dog enables it to learn required tasks.
c. HEARING (AUDITION). Loudness, frequency, direction, and distance provide information about the sound
source. Dogs must be trained to adapt to loud noise such as gunfire.
d. SMELL (OLFACTION). The dog can discriminate one odor from another. The concentration of an odor provides the information about the source of an odor and its direction.
e. PRESSURE. Specialized tactile hairs on the dog provide the capability to detect small changes in air flow
and wind direction. Pressure sensors also are located all over the animal's body which allow the animal to adjust it's behavior to physical correction, physical reward and escape training.
f. BODY POSITION AND MOVEMENT PROPRIOCEPTION (Propri-o-cep-tion). Receptors located in the muscles, tendons, ligaments and joints send information to the dog's brain about its body positions and body movements. A reward that follows a particular body movement or body position is effective because this sense allows the dog to learn which movements or positions gain a reward when given a cue.
g. PAIN. The dog can detect different levels of physical and emotional pain. The other basic senses are used to alert the dog of the impending onset of pain, which enables it to avoid pain. This is the method used to control
the dog by use of vocal correction in avoidance training and the pressure of the choke chain in escape training. By combining physical correction with vocal correction, the vocal corrections become emotionally painful to the dog.
h. TEMPERATURE. The dog can detect hot and cold temperature. Sudden or extreme changes in temperature
inhibit learning during initial training or in subsequent task performance.
i.TASTE. The dog's sense of taste is important if food is used as a reward. The dog uses this sense to indicate one food preference over another.
j. EQUILIBRIUM (BALANCE). The dog's sense of balance maintains the dog's body in an upright position. The dog learns to ignore this sensory input in order to accomplish certain tasks, such as rolling over.
k. VIBRATION. The dog can detect vibrations be way of its vibratory sense. Gradual exposure to an environment with vibration, slowly conditions the dog to ignore it.
l. INTERNAL RECEPTORS. Internal receptors tell the dog when it is hungry or thirsty. This sensory system also informs the dog when it needs to eliminate. It is necessary to insure the dog is not hungry or thirsty during training or task performance. The dog should be provided ample opportunity to eliminate when it needs to. This allows the dog the best chance to learn without distractions from its internal receptors. At times it may be beneficial for a dog's internal receptors to register hunger. This is true when food deprivation is used in the training of food reward detector dogs.
2. TRAINING METHODS. There are three methods to train dogs to perform tasks upon command.
3.
a. REWARD TRAINING. Through this method the dog learns how to obtain a reward by performing a task. Learning is a change in behavior due to practice and is not associated with fatigue or maturation. Vocal and petting reward must be of value to the dog in order for reward training to be effective. Socialization of the dog handler in initial training is designed to gain reward value of both vocal and petting praise. We use the conditioning model "stimulusresponse- reinforcement" (SD-R-Rf). The dog learns that, given a certain stimulus, it must make a certain response to obtain a reward. This process is explained in greater detail in para 7b. Reward training is one way of allowing the dog to learn correct versus incorrect behavior. When the dog makes a correct, or near-correct response, it obtains a reward. When the dog does not respond or makes an incorrect
response, it does not get a reward. The dog always decides the value of the reward.
(1) REWARD SCHEDULES. There are six types of reward schedules:
(a) CONTINUOUS REWARD SCHEDULE (CRS). A reward is given immediately when the dog makes a correct or near correct response. Assisting the dog to assume a particular position is permissible, but remember to reward the dog while it is in position. Dogs that initially make near-correct responses are rewarded as they make subsequent response nearer and nearer the correct response. Shaping the dog to make the correct response will require a slight delay in giving the dog a reward. An excessive delay of a reward does not help the dog to learn a task and leads to frustration. Be very careful with the delay-reward period.
(b) EXTINCTION SCHEDULE (EXT). No reward is ever given by the dog handler when the dog makes no response or an incorrect response. The use of the Extinction Schedule in conjunction with the Continuous Reward Schedule in initial training, provides information to the dog, teaching it the correct response from the incorrect response. If the reward is of sufficient value to the dog, it will select the
right response in order to obtain the reward. NOTE: The dog must perform each task with 100 percent
accuracy before proceeding to the Fixed Ratio and Variable
Ratio Reward Schedules.
(c) FIXED RATIO REWARD SCHEDULE (FR). A reward is given to the dog after it makes two or more correct responses. The two or more responses do not necessarily have to be the same response. To start a dog on this schedule, every second correct response is rewarded. When the dog consistently makes two responses to obtain a reward, three responses can be required. By increasing the number of responses, one at a time, and allowing the dog to perform at each level with 100 percent proficiency, a high Fixed Ratio
Reward Schedule can be attained. If the proficiency is inadequate at any time, decrease the number of responses
required by the dog to obtain a reward. Then proceed as before, adding one response at a time, and require the dog to perform at each level 100 percent accuracy. For example:
(1) Two correct responses = 1 reward.
(2) Three correct responses = 1 reward.
(3) Four correct responses = 1 reward.
NOTE: The dog must perform each step at the 100 percent
accuracy level before proceeding to the next step. If a dog
does not perform correctly on a four or a three response
requirement, return to a two or one response requirement.
(d) VARIABLE RATIO REWARD SCHEDULE (VR).
Once the dog has learned to perform on a high Fixed Ratio Reward Schedule, the Variable Ratio Reward Schedule is used. The maximum number of responses by this schedule must have been learned by a dog on a Fixed Ratio Schedule. Select a range (Example: 5 to 10 responses) of responses required and reward the dog on a random basis, within this range. (Example: The dog has learned to respond correctly 15 times
on a Fixed Ratio Reward Schedule. Reward the dog somewhere between 5 to 10 correct responses on a random basis. The dog will learn that it must correctly respond at least 5 times,but will not be required to respond correctly more than 10 times in order to obtain a desired reward. This reward schedule, once attained, gives you greater control of our dog's behavior than the Continuous Fixed Ratio Reward Schedule).
(e) FIXED INTERVAL REWARD SCHEDULE (FI). A reward is given when the dog is required to respond for a fixed period of time. In initial training, a short period of time must be selected. If the dog does not respond
correctly, a shorter period of time must be selected until the dog responds correctly to obtain a reward. As in the
Fixed Ratio Reward Schedule, short periods of time (Example: 5 seconds) are added to the interval and the dog is required to attain 100 percent accuracy at each interval. If the dog fails to respond correctly for the required length of time, readjust the time requirement of a response to lower time requirement.
NOTE: In initial training you must adhere to a Fixed Interval Reward Schedule in conjunction with the Continuous Reward Schedule, a Fixed Ratio Reward Schedule, or a Variable Ratio Reward Schedule. You are always on more than one schedule at the time. (Example: When you ask the dog to "sit" and he responds by sitting, you reward the dog with vocal and/or petting praise). In this case, you are using a Continuous Reward Schedule. The reward is given immediately as the dog responds. As you successively delay the reward, the dog learns that the request to "sit" also has a time requirement. The reward given satisfies the Continuous Reward and the Fixed Interval requirements. A minimum of 10 trials on each response must be obtained on a CRS at the 100% performance level prior to proceeding to the FR and VR schedules.
(f) VARIABLE INTERVAL REWARD SCHEDULE (VI).
Once the dog has learned to perform any task for a specified period of time on a fixed interval reward schedule
("Stay" in the "Sit" or "Down" position for 3 minutes) the variable interval reward schedule is used. Select a time range (1 to 2 minutes) and reward the dog on a random basis within this time period. Example: The dog has learned to respond correctly for 3 minutes on a fixed interval reward schedule. Reward the dog somewhere between the 1 and 2 minutes time requirement on a random basis. The dog will learn that it must respond for at least a minute and will not be required to respond more than 2 minutes to obtain a desired reward. This reward schedule, once attained, gives you greater control of your dog's behavior than the fixed interval reward schedule.
b. AVOIDANCE TRAINING. This is used in conjunction with the reward and escape training. The dog learns not to make a wrong response in order to avoid a vocal and/or physical correction. A Continuous Reward Schedule is followed in avoidance training. That is, every time the dog makes an undesired response, it receives one correction. Vocal correction must be used initially to determine if it will prohibit undesirable behavior. If vocal correction does not control undesirable behavior, vocal and physical correction must be given simultaneously for each incorrect response. Pairing vocal and physical correction will, after a number of trials, allow the dog to learn how to avoid making an incorrect response, thereby avoiding the vocal and/or physical correction. The dog itself determines the severity of vocal and physical corrections. When vocal and
physical corrections are first applied, start with a low vocal and a low physical correction level. Increase the
vocal and physical correction levels gradually when the dog continues to make wrong responses. The dog's behavior determines the level of correction required for you to control its behavior. Vocal and physical corrections are given only during the time the dog is in the process of making an incorrect response. Corrections at any other time do not provide the dog efficient learning cues. More than one correction per incorrect response by the dog is punishment. Punishment does not allow the dog to learn efficiently. Improper application of avoidance training degrades the value of the vocal and petting praise used as a
reward in reward training.
c. ESCAPE TRAINING. This is the third type of training used. The dog learns to make the correct response to
terminate pain or pressure. (Example: While walking the dog, the handler applies pressure to the dog's neck by
holding the chain next to his/her left leg, so that the dog increases the pressure of the chain if it moves ahead of or legs behind the handler. After a relatively short time, the dog will find the point when the pressure terminates, and will learn to remain in that position - the "HEEL" position). This method also is used to teach the dog to SIT and go into the DOWN position. This training method additionally should be used to teach the dog to stay in the HEEL position while the handler varies his speed while walking, running or marching.
4. LEARNING ACQUISITION. The final response for every task the dog will be required to learn should be defined. This includes what, when, where, how long, etc. (Example: The dog will be required to "Stay" in the "Sit" position "3 feet" from the handler for "3 minutes." By successive additions of time and distance, the dog will be shaped to respond for the required time period and distance from the
handler).
5.STIMULUS CONTROL. The ability to control the dog's behavior is a result of gaining stimulus control. The dog has learned that a request or a certain command allows it to perform a certain task in order to obtain a reward. Initial socialization training of the dog and handler is designed to increase the value of vocal and petting praise. The dog must accept vocal and petting praise as being high in incentive value for reward training to be effective. The ball, other play objects, or food (detection task) provides greater reward strength, which assists in gaining greater stimulus control. Failure to use avoidance training correctly, reduces the value of vocal and petting praise. Failure to use reward schedules correctly also reduces the value of vocal and petting praise. The Variable ratio and Variable interval reward schedule provide greater stimulus control than the remaining reward schedules.
6. STIMULUS GENERALIZATION. When a dog learns to respond to a stimulus, a similar stimulus may cause the dog to make the same response. We use stimulation generalization in some portions of training such as attack, when we want the dog to attack any decoy or suspect and use different people (varying size, location, sex, etc.). Drug or explosive odors that smell similar to, but not exactly like, the odors that the dog has been trained to detect, may cause the dog to respond to untrained odors.
7. CONDITIONING TECHNIQUES.
a. BASIC DOG TRAINING TECHNIQUES (KNOWLEDGE, PATIENCE, PRACTICE). First, knowledge of the principles of conditioning is essential to effectively conduct training. Secondly, the handler must have patience. Training at a rate faster than a dog can learn will result in an untrained animal. The dog determines when to proceed in training. Losing one's temper is a typical situation where the dog is pushed too fast. Thirdly, the team must practice and it must be intelligent practice. The handler constantly must be aware of what he or she is doing.
b. THE CONDITIONING MODEL (SD - R - Rf). This is the model used to training a dog to learn a task. There are three symbols used in the model. The first symbol "SD" is known as the discriminative stimulus and is referred to as the stimulus. The second symbol "R" is known as the response and the third symbol "Rf" is known as the reinforcement or salary. For this model to work for you, the "SD" and "R" must be paired and the "R" and "Rf" must be paired. Before giving an example of how this works, remember the command (stimulus) that a handler proposes to use to obtain a response means nothing to an untrained dog. Therefore, the
handler must provide cues that guide the dog to the specific response and pair the "will be" stimulus with the
response. Now the example: The final response is "Sit" at the handler's left side. The handler begins by providing the cue with pressure (escape training) of the choke chain and leash, to draw the dog near the "heel position." By applying upward pressure (escape training) on the dog's neck with the choke chain and downward pressure (escape training) on the dog's hindquarters, the dog sits. You begin saying the word "sit" and verbally reinforce the response. Eventually, after pairing the response with the reward, the dog will learn that the stimulus is "sit" and soon the stimulus will cause the response without providing additional cues. Remember that although you control the SD, the dog decides whether it will listen or pay attention to that cue. The dog has to learn to make the desired response on its own volition. You also control the reward in most cases (sometimes the response itself is rewarding), but the dog decides the value of the reward.
c. POSITIVE/NEGATIVE ASPECTS OF THE CONDITIONING MODEL.
The dog makes the response and the response can be positive or negative. The reinforcement can also be positive or negative. Normally, the handler controls the reinforcement.
d. REWARD TRAINING. In initial reward training, the reward is simultaneous, or follows the response as possible, but when the dog learns to respond to cues you gradually delay the reward on a FI schedule so that you can eventually put the dog on a VI schedule. In avoidance training, physical and vocal correction must be paired with each other. For maximum learning to occur, the time between the onset of one and the onset of the other should not exceed 0.5 seconds. "NO" should be paired with each physical correction so that "NO" does not become an unconditional stimulus. IMPORTANT: A period of at least a second should separate "NO" and any other command after it has become a conditioned stimulus. Example: "NO HEEL," "NO OUT HEEL," "NO
SIT," "NO DOWN," AND "NO STAY."
8. UTILIZATION OF LEARNING. Learning can have a positive or negative transfer.
a. POSITIVE TRANSFER. When learning has positive transfer, it contributes to future learning like learning to
walk, then to run. Learning to SIT to a vocal and physical cue transfers positively to learning to sit at the
location of an odor in drug and explosive detection.
b. NEGATIVE TRANSFER. When learning has negative transfer, it doesn't contribute to future learning - it has
a detrimental affect. (Example: A dog learning to attack and then attempting to learn a standoff). Learning
building search transfers negatively to learning to search for drugs or explosives in buildings and it can cause a delay in training due to the extension of adaptation training.
c. OVER-LEARNING VS TRANSIENT TRAINING. Over-learning means that the dog has had sufficient practice to allow it to learn a task to a very high degree of accuracy. This type of training allows a dog to recall this type of training after long periods without practice. Transient training (like familiarization training) allows the dog to lose proficiency rapidly without repeated practice.
C. CONCLUSION
SUMMARY: We have covered a dog's basic needs, senses and the methods used to teach the dog to perform properly. Maximum efficiency can be attained if the handler and the supervisor work together. The handler must exercise patience, practice and apply principles of conditioning to the training of his dog.
RE-MOTIVATION (OPTIONAL): If your dog has a training problem, don't be the type of handler who ignores the problem of hides it; tell someone about it and work to solve the problem. Remember that your life could depend on the proficiency of the dog.
LESSON OBJECTIVE: The objective of this lesson is to enable
security police dog handlers to become thoroughly familiar
with the principles of conditioning.
DESIRED LEARNING OUTCOMES: Each student will:
1. Identify the basic needs of a dog.
2. Discuss the dogs sensing system.
3. Describe reward training.
4. Identify and discuss reward schedules.
5. Describe avoidance training.
6. Describe escape training.
7. Describe learning acquisition.
8. Discuss stimulus control.
9. Discuss stimulus generalization.
10. Identify and discuss conditioning techniques.
11. Discuss the utilization of learning.
Lecture/Discussion. Discuss the
principles of conditioning after having outlined the basic needs, senses, and many types of rewards that exist in the animal world. Draw the students into the discussion by giving them some hypothetical situations and having them decide the appropriate reward or correction necessary. A test will be
administered at the end of this class. After testing, critique the test and entertain any question the student may
have.
DEVELOPMENT
1. BASIC NEEDS. A dog responds to its environment in order to satisfy its physical and social needs. In order for a dog to be able to learn, or perform a task, it is necessary to have these needs met. To gain its complete attention, the handler must help insure the dog has all basic needs met, which in turn, builds rapport between dog and handler. Several of the basic needs are used to reward performance.
a. OXYGEN. Breathing to obtain oxygen is perhaps the strongest drive. Exercise or excitement creates an
increased oxygen requirement which causes panting and hinders the olfactory (sense of smell) tasks.
b. WATER. This must be provided to the dog in adequate quantities to prevent thirst from interfering with
learning or task performance. Water cannot be used as reward in military dog training.
c. PREY KILL. This need incorporates chase with the dog manipulating a decoy by biting and tearing and can be used as a reward for task accomplishment through chase/bite agitation training.
d. FOOD. This must be supplied in adequate quantities to prevent hunger from interfering with task performance. Food sometimes is used as reward to teach dogs to perform drug or explosive detector tasks.
e. SOCIAL NEEDS. A period of socialization between dog and handler is required in order for vocal and petting praise to acquire reward value to the dog. This socialization must continue for praise to effectively maintain the desired behavior from the dog after training.
f. PAIN AVOIDANCE. The need to minimize pain enables a trainer to teach the dog not to make specific responses. The use of physical correction does not, however, teach the dog the correct response to any specific cue.
2. SENSING SYSTEM. A dog perceives its environment through the use of its basic senses. The dog senses its
environment much differently than humans. A lack of knowledge of the dog's sensing a system that causes much of the misunderstanding in the dog's training and use. The following is a discussion of the companion working dog's sensing capabilities and limitations.
a. JUST NOTICEABLE DIFFERENCE (JND). Before we can accurately describe the sensing system of the military working dog, we must first explain JND. JND is the animal's ability to detect slight changes or differences pertaining to a specific sense. ALL sensing systems are included in this definition. Examples of JND include: the smelling of a weak odor as opposed to a stronger odor; the hearing of low
volume noises as opposed to louder volumes; the feeling of light pressure as opposed to progressively heavier pressure,and so on. Again, JND applies to all senses of the animal.
(1) A JND in odor concentration is the input a dog uses to determine the strongest odor, such as a human hiding
or a substance the dog has learned to search for. All dogs have limitations. There first must be certain amount of odor concentration for the dog to detect at all. If inadequate concentration is available to the dog it will be unable to locate the source(s). This minimal level is referred to as the dog's absolute threshold, which is a just noticeable difference between no detectable odor and enough for the dog to recognize. The absolute threshold is a fixed amount of odor that is necessary for the dog to recognize it. The absolute threshold of one dog can be different than that of another; thus, one dog may detect an odor that another dog missed regardless of how well the other dog is trained. In other words, a dog's absolute threshold is the point of recognition of an odor the dog has learned to detect and respond to. Other odors are also detected but will be ignored. NOTE: There is an exception: if the odor is very similar to the one the dog learned to respond to, the dog may "stimulus generalize" and make a like response to a similar odor.
(2) The dog learns in order to obtain a reward, it must find the next higher JND and then the next and the next until it reaches the source of the odor, or the highest concentration. The dog will move about, sniffing,
until a difference is determined. When the difference is lower, the dog will change direction and continue to search for a JND which is higher.
(3) When the dog locates a higher JND it will search for one still higher and so on until it reaches the
source or the point of saturation. There may be more odor present, but since there are no more JND's the dog is
literally at the end of the line. To ask more of the dog would be asking it to perform beyond its capabilities. If you continue, you will be extinguishing the behavior the dog originally learned to perform.
(4) The dog must receive its reward when it has gone as far as it is capable of going. Likewise, if the dog
reaches a physical barrier, such as a door that can't be opened, the reward should be given. Once the dog reaches
the extent of its capabilities, it has completed the task and must be rewarded!
(5) In the first sentence of this portion of the lesson we said that "Just Noticeable Difference" applies to all the senses. The same is true of the "spectrum". It can be used as a visual, sound (volume or pressure) or any other spectrum. You can go through the entire list of basic senses and apply the concept of absolute threshold, JNDs and the point of saturation. By omitting all references to odor and replacing them with terms pertaining to another basic sense you have an explanation of how this concept applies to all basic senses.
(6) It is important to note threshold shifts occur when the sensory system is activated. This means that once
the dog's sense (remember it applies to all of them), has been activated a certain recovery period is required before the dog will be able to detect a smaller amount of odor, light, or sound, etc. The greater the amount of odor, light, or sound, etc., that the dog is exposed to, the longer the recovery period before smaller amounts can be detected. An example of this effect is demonstrated by going from bright sunlight into a dark room. After a period of time you can see objects in the room you could not see when
you first entered.
NOTE: JND SUMMARY: JND applies to all of the senses shown.
Your knowledge of JND in relation to these individual senses should explain many of the misconceptions you may have had in the past.
b. SIGHT (VISION). Through conditioning, the dog learns to respond or ignore visual cues in its environment.
Cues such as brightness, shape, size, and motion provide information to the dog enables it to learn required tasks.
c. HEARING (AUDITION). Loudness, frequency, direction, and distance provide information about the sound
source. Dogs must be trained to adapt to loud noise such as gunfire.
d. SMELL (OLFACTION). The dog can discriminate one odor from another. The concentration of an odor provides the information about the source of an odor and its direction.
e. PRESSURE. Specialized tactile hairs on the dog provide the capability to detect small changes in air flow
and wind direction. Pressure sensors also are located all over the animal's body which allow the animal to adjust it's behavior to physical correction, physical reward and escape training.
f. BODY POSITION AND MOVEMENT PROPRIOCEPTION (Propri-o-cep-tion). Receptors located in the muscles, tendons, ligaments and joints send information to the dog's brain about its body positions and body movements. A reward that follows a particular body movement or body position is effective because this sense allows the dog to learn which movements or positions gain a reward when given a cue.
g. PAIN. The dog can detect different levels of physical and emotional pain. The other basic senses are used to alert the dog of the impending onset of pain, which enables it to avoid pain. This is the method used to control
the dog by use of vocal correction in avoidance training and the pressure of the choke chain in escape training. By combining physical correction with vocal correction, the vocal corrections become emotionally painful to the dog.
h. TEMPERATURE. The dog can detect hot and cold temperature. Sudden or extreme changes in temperature
inhibit learning during initial training or in subsequent task performance.
i.TASTE. The dog's sense of taste is important if food is used as a reward. The dog uses this sense to indicate one food preference over another.
j. EQUILIBRIUM (BALANCE). The dog's sense of balance maintains the dog's body in an upright position. The dog learns to ignore this sensory input in order to accomplish certain tasks, such as rolling over.
k. VIBRATION. The dog can detect vibrations be way of its vibratory sense. Gradual exposure to an environment with vibration, slowly conditions the dog to ignore it.
l. INTERNAL RECEPTORS. Internal receptors tell the dog when it is hungry or thirsty. This sensory system also informs the dog when it needs to eliminate. It is necessary to insure the dog is not hungry or thirsty during training or task performance. The dog should be provided ample opportunity to eliminate when it needs to. This allows the dog the best chance to learn without distractions from its internal receptors. At times it may be beneficial for a dog's internal receptors to register hunger. This is true when food deprivation is used in the training of food reward detector dogs.
2. TRAINING METHODS. There are three methods to train dogs to perform tasks upon command.
3.
a. REWARD TRAINING. Through this method the dog learns how to obtain a reward by performing a task. Learning is a change in behavior due to practice and is not associated with fatigue or maturation. Vocal and petting reward must be of value to the dog in order for reward training to be effective. Socialization of the dog handler in initial training is designed to gain reward value of both vocal and petting praise. We use the conditioning model "stimulusresponse- reinforcement" (SD-R-Rf). The dog learns that, given a certain stimulus, it must make a certain response to obtain a reward. This process is explained in greater detail in para 7b. Reward training is one way of allowing the dog to learn correct versus incorrect behavior. When the dog makes a correct, or near-correct response, it obtains a reward. When the dog does not respond or makes an incorrect
response, it does not get a reward. The dog always decides the value of the reward.
(1) REWARD SCHEDULES. There are six types of reward schedules:
(a) CONTINUOUS REWARD SCHEDULE (CRS). A reward is given immediately when the dog makes a correct or near correct response. Assisting the dog to assume a particular position is permissible, but remember to reward the dog while it is in position. Dogs that initially make near-correct responses are rewarded as they make subsequent response nearer and nearer the correct response. Shaping the dog to make the correct response will require a slight delay in giving the dog a reward. An excessive delay of a reward does not help the dog to learn a task and leads to frustration. Be very careful with the delay-reward period.
(b) EXTINCTION SCHEDULE (EXT). No reward is ever given by the dog handler when the dog makes no response or an incorrect response. The use of the Extinction Schedule in conjunction with the Continuous Reward Schedule in initial training, provides information to the dog, teaching it the correct response from the incorrect response. If the reward is of sufficient value to the dog, it will select the
right response in order to obtain the reward. NOTE: The dog must perform each task with 100 percent
accuracy before proceeding to the Fixed Ratio and Variable
Ratio Reward Schedules.
(c) FIXED RATIO REWARD SCHEDULE (FR). A reward is given to the dog after it makes two or more correct responses. The two or more responses do not necessarily have to be the same response. To start a dog on this schedule, every second correct response is rewarded. When the dog consistently makes two responses to obtain a reward, three responses can be required. By increasing the number of responses, one at a time, and allowing the dog to perform at each level with 100 percent proficiency, a high Fixed Ratio
Reward Schedule can be attained. If the proficiency is inadequate at any time, decrease the number of responses
required by the dog to obtain a reward. Then proceed as before, adding one response at a time, and require the dog to perform at each level 100 percent accuracy. For example:
(1) Two correct responses = 1 reward.
(2) Three correct responses = 1 reward.
(3) Four correct responses = 1 reward.
NOTE: The dog must perform each step at the 100 percent
accuracy level before proceeding to the next step. If a dog
does not perform correctly on a four or a three response
requirement, return to a two or one response requirement.
(d) VARIABLE RATIO REWARD SCHEDULE (VR).
Once the dog has learned to perform on a high Fixed Ratio Reward Schedule, the Variable Ratio Reward Schedule is used. The maximum number of responses by this schedule must have been learned by a dog on a Fixed Ratio Schedule. Select a range (Example: 5 to 10 responses) of responses required and reward the dog on a random basis, within this range. (Example: The dog has learned to respond correctly 15 times
on a Fixed Ratio Reward Schedule. Reward the dog somewhere between 5 to 10 correct responses on a random basis. The dog will learn that it must correctly respond at least 5 times,but will not be required to respond correctly more than 10 times in order to obtain a desired reward. This reward schedule, once attained, gives you greater control of our dog's behavior than the Continuous Fixed Ratio Reward Schedule).
(e) FIXED INTERVAL REWARD SCHEDULE (FI). A reward is given when the dog is required to respond for a fixed period of time. In initial training, a short period of time must be selected. If the dog does not respond
correctly, a shorter period of time must be selected until the dog responds correctly to obtain a reward. As in the
Fixed Ratio Reward Schedule, short periods of time (Example: 5 seconds) are added to the interval and the dog is required to attain 100 percent accuracy at each interval. If the dog fails to respond correctly for the required length of time, readjust the time requirement of a response to lower time requirement.
NOTE: In initial training you must adhere to a Fixed Interval Reward Schedule in conjunction with the Continuous Reward Schedule, a Fixed Ratio Reward Schedule, or a Variable Ratio Reward Schedule. You are always on more than one schedule at the time. (Example: When you ask the dog to "sit" and he responds by sitting, you reward the dog with vocal and/or petting praise). In this case, you are using a Continuous Reward Schedule. The reward is given immediately as the dog responds. As you successively delay the reward, the dog learns that the request to "sit" also has a time requirement. The reward given satisfies the Continuous Reward and the Fixed Interval requirements. A minimum of 10 trials on each response must be obtained on a CRS at the 100% performance level prior to proceeding to the FR and VR schedules.
(f) VARIABLE INTERVAL REWARD SCHEDULE (VI).
Once the dog has learned to perform any task for a specified period of time on a fixed interval reward schedule
("Stay" in the "Sit" or "Down" position for 3 minutes) the variable interval reward schedule is used. Select a time range (1 to 2 minutes) and reward the dog on a random basis within this time period. Example: The dog has learned to respond correctly for 3 minutes on a fixed interval reward schedule. Reward the dog somewhere between the 1 and 2 minutes time requirement on a random basis. The dog will learn that it must respond for at least a minute and will not be required to respond more than 2 minutes to obtain a desired reward. This reward schedule, once attained, gives you greater control of your dog's behavior than the fixed interval reward schedule.
b. AVOIDANCE TRAINING. This is used in conjunction with the reward and escape training. The dog learns not to make a wrong response in order to avoid a vocal and/or physical correction. A Continuous Reward Schedule is followed in avoidance training. That is, every time the dog makes an undesired response, it receives one correction. Vocal correction must be used initially to determine if it will prohibit undesirable behavior. If vocal correction does not control undesirable behavior, vocal and physical correction must be given simultaneously for each incorrect response. Pairing vocal and physical correction will, after a number of trials, allow the dog to learn how to avoid making an incorrect response, thereby avoiding the vocal and/or physical correction. The dog itself determines the severity of vocal and physical corrections. When vocal and
physical corrections are first applied, start with a low vocal and a low physical correction level. Increase the
vocal and physical correction levels gradually when the dog continues to make wrong responses. The dog's behavior determines the level of correction required for you to control its behavior. Vocal and physical corrections are given only during the time the dog is in the process of making an incorrect response. Corrections at any other time do not provide the dog efficient learning cues. More than one correction per incorrect response by the dog is punishment. Punishment does not allow the dog to learn efficiently. Improper application of avoidance training degrades the value of the vocal and petting praise used as a
reward in reward training.
c. ESCAPE TRAINING. This is the third type of training used. The dog learns to make the correct response to
terminate pain or pressure. (Example: While walking the dog, the handler applies pressure to the dog's neck by
holding the chain next to his/her left leg, so that the dog increases the pressure of the chain if it moves ahead of or legs behind the handler. After a relatively short time, the dog will find the point when the pressure terminates, and will learn to remain in that position - the "HEEL" position). This method also is used to teach the dog to SIT and go into the DOWN position. This training method additionally should be used to teach the dog to stay in the HEEL position while the handler varies his speed while walking, running or marching.
4. LEARNING ACQUISITION. The final response for every task the dog will be required to learn should be defined. This includes what, when, where, how long, etc. (Example: The dog will be required to "Stay" in the "Sit" position "3 feet" from the handler for "3 minutes." By successive additions of time and distance, the dog will be shaped to respond for the required time period and distance from the
handler).
5.STIMULUS CONTROL. The ability to control the dog's behavior is a result of gaining stimulus control. The dog has learned that a request or a certain command allows it to perform a certain task in order to obtain a reward. Initial socialization training of the dog and handler is designed to increase the value of vocal and petting praise. The dog must accept vocal and petting praise as being high in incentive value for reward training to be effective. The ball, other play objects, or food (detection task) provides greater reward strength, which assists in gaining greater stimulus control. Failure to use avoidance training correctly, reduces the value of vocal and petting praise. Failure to use reward schedules correctly also reduces the value of vocal and petting praise. The Variable ratio and Variable interval reward schedule provide greater stimulus control than the remaining reward schedules.
6. STIMULUS GENERALIZATION. When a dog learns to respond to a stimulus, a similar stimulus may cause the dog to make the same response. We use stimulation generalization in some portions of training such as attack, when we want the dog to attack any decoy or suspect and use different people (varying size, location, sex, etc.). Drug or explosive odors that smell similar to, but not exactly like, the odors that the dog has been trained to detect, may cause the dog to respond to untrained odors.
7. CONDITIONING TECHNIQUES.
a. BASIC DOG TRAINING TECHNIQUES (KNOWLEDGE, PATIENCE, PRACTICE). First, knowledge of the principles of conditioning is essential to effectively conduct training. Secondly, the handler must have patience. Training at a rate faster than a dog can learn will result in an untrained animal. The dog determines when to proceed in training. Losing one's temper is a typical situation where the dog is pushed too fast. Thirdly, the team must practice and it must be intelligent practice. The handler constantly must be aware of what he or she is doing.
b. THE CONDITIONING MODEL (SD - R - Rf). This is the model used to training a dog to learn a task. There are three symbols used in the model. The first symbol "SD" is known as the discriminative stimulus and is referred to as the stimulus. The second symbol "R" is known as the response and the third symbol "Rf" is known as the reinforcement or salary. For this model to work for you, the "SD" and "R" must be paired and the "R" and "Rf" must be paired. Before giving an example of how this works, remember the command (stimulus) that a handler proposes to use to obtain a response means nothing to an untrained dog. Therefore, the
handler must provide cues that guide the dog to the specific response and pair the "will be" stimulus with the
response. Now the example: The final response is "Sit" at the handler's left side. The handler begins by providing the cue with pressure (escape training) of the choke chain and leash, to draw the dog near the "heel position." By applying upward pressure (escape training) on the dog's neck with the choke chain and downward pressure (escape training) on the dog's hindquarters, the dog sits. You begin saying the word "sit" and verbally reinforce the response. Eventually, after pairing the response with the reward, the dog will learn that the stimulus is "sit" and soon the stimulus will cause the response without providing additional cues. Remember that although you control the SD, the dog decides whether it will listen or pay attention to that cue. The dog has to learn to make the desired response on its own volition. You also control the reward in most cases (sometimes the response itself is rewarding), but the dog decides the value of the reward.
c. POSITIVE/NEGATIVE ASPECTS OF THE CONDITIONING MODEL.
The dog makes the response and the response can be positive or negative. The reinforcement can also be positive or negative. Normally, the handler controls the reinforcement.
d. REWARD TRAINING. In initial reward training, the reward is simultaneous, or follows the response as possible, but when the dog learns to respond to cues you gradually delay the reward on a FI schedule so that you can eventually put the dog on a VI schedule. In avoidance training, physical and vocal correction must be paired with each other. For maximum learning to occur, the time between the onset of one and the onset of the other should not exceed 0.5 seconds. "NO" should be paired with each physical correction so that "NO" does not become an unconditional stimulus. IMPORTANT: A period of at least a second should separate "NO" and any other command after it has become a conditioned stimulus. Example: "NO HEEL," "NO OUT HEEL," "NO
SIT," "NO DOWN," AND "NO STAY."
8. UTILIZATION OF LEARNING. Learning can have a positive or negative transfer.
a. POSITIVE TRANSFER. When learning has positive transfer, it contributes to future learning like learning to
walk, then to run. Learning to SIT to a vocal and physical cue transfers positively to learning to sit at the
location of an odor in drug and explosive detection.
b. NEGATIVE TRANSFER. When learning has negative transfer, it doesn't contribute to future learning - it has
a detrimental affect. (Example: A dog learning to attack and then attempting to learn a standoff). Learning
building search transfers negatively to learning to search for drugs or explosives in buildings and it can cause a delay in training due to the extension of adaptation training.
c. OVER-LEARNING VS TRANSIENT TRAINING. Over-learning means that the dog has had sufficient practice to allow it to learn a task to a very high degree of accuracy. This type of training allows a dog to recall this type of training after long periods without practice. Transient training (like familiarization training) allows the dog to lose proficiency rapidly without repeated practice.
C. CONCLUSION
SUMMARY: We have covered a dog's basic needs, senses and the methods used to teach the dog to perform properly. Maximum efficiency can be attained if the handler and the supervisor work together. The handler must exercise patience, practice and apply principles of conditioning to the training of his dog.
RE-MOTIVATION (OPTIONAL): If your dog has a training problem, don't be the type of handler who ignores the problem of hides it; tell someone about it and work to solve the problem. Remember that your life could depend on the proficiency of the dog.