CONCEPTS OF DRUG DETECTOR DOG TRAINING
STATEMENTS: These are assertions put forward to be measured for the truth.
1. Analytical. All of these statements are true by definition. Example: 2+2=4, 36/9=4.
2. Empirical. To make these statements you must at least have an interval level of measurement for the statements to be examined for the truth. Example: My dog can detect controlled drugs above an 85% level. (There are other variables that have to be controlled in order to assure you are measuring the dog’s accuracy rate without some other influence.)
3. Metaphysical. Beyond physical measurement. Example: my dog knows the drug is there.
4. Anthropomorphic. Attributing human characteristics to an animal. Example: my dog is happy.
NOTE: Neither of the last two types of statements can be measured for the truth.
You can operationally define happiness as a heart rate of between 90 and 140 beats per minute. You can accurately measure that heart rate under a variety of conditions. You cannot say you are measuring happiness.
SCHEDULES OF REINFORCEMENT:
Reinforcements are controlled by the handler and are given by the handler when the dog makes the defined final response to the target odor at a specific rate.
1. Continuous. The reinforcement is given every time the dog makes the correct response. (Rewarding a dog for making the defined final response on an operational search can lead to problems if that response was false and no drugs were found.) This is the most common schedule used with drug detector dogs.
2. Fixed ratio. The dog is rewarded after making a specific number of responses before receiving a reward.
3. Variable ratio. A dog can be trained to accept a reward on a random number of responses. (Dogs who beg at the dinner table make a random number of response before finally being rewarded, thus a difficult behavior to extinguish once it is has been developed across a number of trials.)
4. Fixed interval. The reward is given at a specific interval after the defined final response is made. (The American Kennel Club has a task that requires the dog to stay in the down position for 4 minutes without the handler.)
5. Variable interval. The reward is given on a random time basis after the response. Slot machines pay off on a random interval. This is the reason for people to continue playing them.
NOTE: Behavior learned on the variable ratio or variable interval schedules is very resistant to extinction. They can be used in dog training but one has to be well versed in how to use the reward schedules in order to be efficient.
A training procedure where you teach a dog to make a defined final response in or to the presence of a specific stimulus in order to obtain appetitive reinforcement.
The three elements are always present in this procedure, the stimulus, the response, and the reward. All drug dogs are trained by this procedure. Dog trainers all have their preferences as to how to train dogs but they have to follow the basics of the three elements and use the learning laws of continuity and the law of contiguity to be efficient. The direction of the presentation of the elements and the proximity of each is extremely important for learning to take place efficiently.
A procedure where two stimuli are presented to the animal in close temporal proximity. On of the stimuli has a reflex behavior or a previously acquired connection to certain response. When the two stimuli are repeatedly paired many times the second stimulus acquired the ability to evoke the response very similar to the one to the first stimulus. Example: Food causes salivation. When it is paired with a bell many times, the bell acquires the ability to evoke salivation. The clicker training is an example of where it can be used in dog training.
PRIMARY REINFORCEMENT: The primary reward that controls a specific dog response. The dog sets the value of the reward but the handler must control the reward in order to control the dog’s specific behavior.
SECONDARY REINFORCEMENT. A reward that occurs along with or is associated with the primary reward. Vocal and petting praise is paired with a primary reward such a ball, toy, or food.
EXTINCTION: An experimental procedure where you reduce the conditioned response by withholding the reinforcement. It is also a process where the conditioned response is eliminated when the dog responds to the wrong odor.
CHAINING BEHAVIOR: A procedure where the dog is provided several stimuli to induce a series of responses. Example: A dog is given several cues to get it to learn to search the exterior of a vehicle. If the procedure is handled correctly the dog learns a complete systematic search pattern. The METHOD OF SUCCESSIVE APPROXIMATION is used to teach parts of the final response on each of a series of successive trials. TIME of performance is a part of this type of training.
STIMULUS CONTROL: When a specific stimulus will elicit a specific defined final response by a dog at a predetermined rate of reliability, you have gained stimulus control of that behavior.
SHAPING BEHAVIORS: A task is broken down its parts. You train the first part then add the second part to the first part then you gradually add on all the other parts until the entire task has been accomplished. This too is the METHOD OF SUCCESSIVE APPROXIMATION. TIME is also measured as part of this behavior. The dog must learn at its own rate. If the dog fails anywhere along the way simplify the task and proceed at the dog’s rate of learning.
FADING: A search pattern is trained by repeatedly giving the dog cues to continue on this search pattern without interruption. Once the dog has learned the entire search pattern to a high degree of reliability the cues are systematically dropped out one by one. At that point you have a dog that will complete an entire search pattern on the initial cue at the start. This procedure is known as fading.
SPONTANEOUS RECOVERY: Once you have extinguished a previously learned behavior, that stimulus can cause the extinguished response to reoccur. Example: You extinguish the dog behavior of responding to dryer sheets. In the future that response can occur given the presence of dryer sheets. This is the reason to continue extinction trials on items the dog has made errors on before.
SIMULUS GENERALIZATION: The fact that a dog will respond to items very similar to the ones it has been conditioned to respond to. Example, when a dog is trained to detect lemons it will frequently respond to other citrus fruit.
NOTE: Just because you have trained a dog on an opiate does not justify the assumption the dog will respond to all drugs in that class. One still has to train and test under controlled conditions in order to rely on these measurements to make non-speculative statements about the dogs performance on derivative compounds.
TARGET ODOR DISSIPATION RATES: Odors dissipate across time otherwise we would still be smelling dinosaurs. Scientific studies have not been conducted to determine these rates. A single study on cocaine (methyl benzoate) has been done so far.
Law of Continuity: The fact that the three elements of operant conditioning (stimulus, response, reward) must be presented in that particular sequence in order to provide efficient learning.
Law of Contiguity: The three elements must be in close proximity to each other to provide efficient learning.
ESCAPE TRAINING: A form of learning where a response eliminates an unpleasant situation. Example: If you put enough pressure with a leash on a dog with a choke chain around its neck the dog will follow its neck to reduce the pressure. Pulling up on a leash and pressing down on a dog’s rump will induce sitting to avoid the pressure.
AVOIDANCE TRAINING: A noxious stimulus is administered only if the animal fails to make the prescribed avoiding response. Example: If you physically correct a dog for not staying in a STAY position. The dog will learn to STAY when cued.
NOTE: These two levels of measurement do not allow you to statistically compare member of a group or make probability statements about that group.
NOTE: These two levels provide you with mathematical data that can be statistically manipulated to provide one with probability statements. Comparison statements can be made with a high rate of reliability.
DAN J. CRAIG BS,