Main Performance Training Page
Written by Chris Biro
Training Defined
Learning is behavior that has been modified due to an animal’s experience. Thus learning is the environment influencing an animal’s behavior. Learning is not an option, but is occurring every moment of every day. Training is intentionally altering the environment to bring about specific changes in behavior.
Whether an animal’s life with humans is joyful or painful, depends completely on our skills at training them to peacefully coexist with us. The success of an animal to live separated from their natural environment, depends entirely on mans ability to influence their natural behaviors to better fit into the unnatural conditions we have placed them. In most cases, the choice to live with humans was not theirs but was ours; and their success or failure in doing so is our responsibility, not theirs.
Communication & Motivation
Clear communication and hi levels of motivation are the two main elements of effective training. At any given time the status of these two areas will result in one of four training conditions in the trainee: Unable & Unwilling (C-, M-), Unable & Willing (C-, M+), Able & Unwilling (C+, M-), and Able & Willing (C+, M+). Recognizing if the animal’s poor performance is based on lack of motivation and/or lack of understanding can identify where to focus the emphasis of the training, on motivation or on communication.
|
|
Does Not Understand The Task |
Does Understand The Task |
Not Motivated to Participate |
Unable & Unwilling |
Able & Unwilling |
|
Is Motivated to Participate |
Unable & Willing |
Able & Willing |
If a trainer is encountering problems with his training, it is sometimes helpful to stop and consider which of the following four conditions applies to the student. As an example, if the animal is able to perform a task but unwilling to perform when requested, the trainer needs to focus his efforts on building better motivation – continued to attempts to communicate the criteria of the behavior would not be very effective.
The typical training sequence goes as follows:
1. At first the main emphasis is on stimulating the trainee’s attention and participation, i.e. on communicating information relating to motivation.
2. Once participation is achieved the main emphasis shifts to communicating the criteria of the desired behavior.
a. Motivation to participate, at this point, may continue from its own inherent momentum if not repressed.
b. If repressed the focus will need to occasionally revert from the criteria, back to participation (motivation, step A).
3. Once the criteria of the behavior are learned, the main emphasis shifts again to motivation, to insure continued proper performance of the learned behavior.
4. And once the animal understands all the criteria of the behavior AND is willingly, properly performing the behavior, the focus will shift to only mild attention on both criteria and motivation for maintenance of the behavior.
Motivation comes from an animal’s desire to pursue a condition or to avoid a condition. Repeatedly being presented with conditions that the animal wishes to avoid can result in poor motivation levels. Likewise, repeatedly being presented with conditions that the animal wishes to pursue can increase motivation levels. Motivation can be effected by what happens both before and after a behavior.
Antecedent = the observable stimulus/conditions leading up to a behavior.
Behavior = the animal’s observable response / action to the antecedents.
Consequence = the physical/observable result of the behavior.
There may be several antecedents and several consequences for each single behavior. Note: the previous consequence may be part of the antecedents of a following behavior.
As trainers we can only influence the behavior by altering the antecedents and/or the consequences. To alter an antecedent we must either add or remove some condition that actually contributes to the performance of the behavior. To alter the consequence we can add or remove some condition that immediately follows the behavior. Altering a consequence is actually altering an antecedent for the next repetition of the behavior. The function of a consequence is to affect the conditioning of an antecedent for the next repetition (if there is one).
It is first required to stimulate performances of behavior before the opportunity exists to affect a consequence. While teaching the criteria of a new behavior much of the trainers focus will be on the antecedents as it is attempted to find ways to trigger a form of the desired behavior. Once antecedents are found that can stimulate the performance of forms of the behavior, the trainer’s focus will begin to shift toward the consequence side of the equation to better control motivation levels.
Reinforcement = Reinforcement is any event that increases the probability that a response will occur again in a situation like that in which the reinforcing condition originally occurred. Generally, any condition strengthening learning.
1. Unlearned (Instinctive / Unconditioned / Primary) reinforcement is inherently desirable; most often referred to as “unconditioned reinforcer”.
2. Learned (Conditioned / Secondary) reinforcement is learned through associated pairings with some unlearned reinforcer or another learned reinforcer; most often referred to as “conditioned reinforcer”. The strength of a conditioned reinforcer will depend on the strength of the paired unconditioned or conditioned reinforcer and how strongly associated it is with the other reinforcer.
Punishment = is presentation of an aversive stimulus, or the removal of the opportunity to obtain reinforcement, which decreases the probability that the response leading to the behavior will persist. Conditioned punishers are learned through association with unconditioned and other conditioned punishers.
Reinforcements tend to increase motivation levels and punishments tend to decrease motivation levels. Reinforcements tend to strengthen/increase/maintain the behavior while punishments tend to weaken/decrease/eliminate the behavior.
Punishment not only tends to decrease motivation, but also has many other side effects that interfere with learning and thus training. The use of punishment should be kept to a minimum, if used at all. When punishments are used, additional reinforcement opportunities may be needed to counter the effects of the punishment on motivation levels. For best results, the ratio between use of reinforcement and use of punishments should be kept hugely in favor of reinforcements.
Note: Reinforcement and Punishment are defined by the observable outcome and not by the animal’s or the trainer’s inner thoughts or opinions of the consequences nor from the trainer’s expectations or assumptions.
To train a desired behavior we must first have some behavior to reinforce. Rarely does the animal already perform our end goal behavior. We can adjust the antecedents to help set the animal up for success by using one or several in combination of the following common techniques.
1. Capturing is reinforcing a preexisting behavior. Waiting for the event to happen, then reinforcing when it happens - often trains potty training in the desired place or manner.
2. Modeling is (the trainer or some other animal) physically showing the animal how the target behavior should be performed - often used to train parrots to speak specific words.
3. Physical assisting is physically moving the animal through the behavior or into the positions of the behavior. Dogs are often taught to site by gently pushing downward on the rump and simultaneously causing the rear legs to bend, resulting in the animal placing its rump on the ground in the sitting position.
4. Luring is causing the animal to follow some desirable object. Placing a piece of meat inside an animal carrier will often set the occasion for a cat to enter the carrier.
5. Shaping is building the target behavior through reinforcement of tiny increments of improvement to the existing level of behavior – often referred to as shaping successive approximations.
Combining some or all of the above is the most common approach to training most behaviors. Some trainers develop one or more of these individual methods into incredibly powerful tools.
The performances of each behavior are rarely identical. Some will naturally tend to be “better” or more like what we want and others will be “worse” or less like what we want. To get started it is often necessary to reinforce performances of the behavior that are only remotely similar to the end goal behavior. As that level of behavior becomes learned, we can then “up the criteria” and reinforce those better performance that slightly better resembles our end goal behavior. Choosing to reinforce those that are “better” and not those that are “worse” will eventually result in producing our desired behavior if continued long enough.
Upping the criteria in too large of steps will often result in fewer or no opportunities to reinforce the animal – this can have a punishing effect on the animal since this denies opportunity to earn reinforcement (reinforcements are effectively withheld). If continued long enough, the previously learned form of the behavior will entirely cease to be offered by the animal, and the behavior will “extinguish”. Prior to the behavior becoming extinct the animal will tend to attempt to “make it work” by putting greater effort into doing what it has learned gets it reinforcements. This is called an “extinction burst”. This tendency is both to our advantage and our disadvantage. As the animal increases its level of effort, the variations in the offered behavior will tend to be greater, offering opportunity to reinforce larger steps of improvement. However, “extinction bursts” typically result from raised levels of frustration in the animal, which can have a negative effect on overall motivation levels.
Upping the criteria in tiny steps that are within the animal’s current “natural variations” of the behavior will avoid the frustration encountered with the extinction burst and will result in much faster improvements to the behavior and much higher levels of motivation. Ideally criteria should be raised only enough to result in performances by the animal being reinforced about 8 out of 10 repetitions. Scientists in laboratories and when doing field research, keep accurate logs and journals to help meet this schedule while many trainers just “wing it”. With practice, eventually some trainers develop a feel for how much is appropriate to raise the criteria without frustrating the animal. In this sense training becomes an art. However, use of a journal is recommended for obtaining the best results and the least amount of animal frustration.
Both reinforcers and punishers are conditions added to the consequence side of the training equation and thus are each delivered AFTER the behavior occurs. For the greatest effect, the delivery should be made immediately after the behavior occurs. The force of the effect of either will decline proportionately to the amount of time elapsing between the behavior and the time of delivery – the longer the delay, the less the effect of the reinforcement. With repeated immediate delivery of the consequences, the animal can begin to associate the behavior with that particular consequence. In this manner the criteria of a desired behavior can be communicated to the animal. However, immediate delivery is not always possible. An alternate and perhaps more effective method of marking the desired aspect of behavior and linking it to the corresponding consequence is called “bridging”.
Bridging = 1) marking a behavior by a specific signal that is followed by 2) delivery of a predictable consequence to the behavior.
Bridging is most often associated with the delivery of reinforcers and when used accordingly, the bridging signal itself will become a conditioned reinforcement – in theory bridging could be used as a conditioned punisher but trainers do not usually use such. Offering the signal and then immediately offering a specific consequence (repeating as required) will condition the bridge, i.e. the animal will learn that the signal means a specific consequence will soon follow the signal. To keep this conditioning strong, bridging should always follow a 1:1 ratio rule - one bridging signal equals one or more reinforcements. The bridging signal is the first consequence of the behavior but because it is a conditioned reinforcer it needs back up to retain its reinforcing properties – most trainers agree that the back up that is most effective for backing up their bridging signal is food.
“Bridging” the gap between the behavior and the final reinforcement, allows the reinforcement to be linked to the behavior even if several seconds elapse between the time of behavior and delivery of reinforcement. The proper bridging signal can usually more accurately mark the specific moment in time the desired behavior occurred, than can the simple delivery of most reinforcements. This increase in accuracy results in improved communication between trainer and trainee.
Different bridging signals will have different uses. If the bridging signal is brief it will better mark behaviors of short duration. If it is unique the animal will not be confused when that same signal is offered out of context to training. The signal may be audible, visual, tactile, etc. and each will have specific advantages and limitations. Visual signals require the animal to be looking at the source of the signal but may be observed at great distances regardless of noise levels. Sounds may be either too loud or too quiet for the training conditions but do not require the animal to be looking at the source of the signal. Tactile requires the animal to be physically with in reach of the source of the signal but can be delivered stealthily. Most trainers find a sharp audible sound to work for most general training conditions. Spoken words are sometimes used as bridging signals but said words are rarely used exclusively in the training environment and also are fairly long in duration. The crisp snapping sound of a cloths pin or harness hook (like found on the end of a dog leash) is unique and short duration but are both often difficult to handle and to make sound off with precision. Because of ease of use and unique and short duration sound, a version of the child’s toy “cricket”, commonly called a “clicker”, is currently the preferred bridging signal used by many animal trainers.
A single bridging with a clicker will not normally be enough to train the desired behavior. For the animal to understand the criteria of the desired behavior, the animal normally needs several repetitions. The less time that elapses between repetitions, the faster the learning will progress. The more repetitions achieved, the better the animal will understand what is expected and the more it will trust its ability to achieve the desired consequence (reinforcement) through performing that behavior. Thus fast and frequent repetitions produce better communication and higher motivation levels. Every effort in adjusting antecedents should be made to set the animal up to succeed. The highest level of learning will occur when there is the highest number of reinforcements being delivered. The faster the reinforcements are delivered, the faster the learning will occur.
It should also be noted that research shows that the animal does not actually need to “understand” the criteria for this training method to cause changes in the animal’s behavior. Often we alter our own behavior due to reinforcements in our own lives but never actually notice that we in fact learned from those reinforcements. In truth we have no way of knowing what the animal “understands” but we do know that clear communication is very good for motivation levels.
Have Theory, Need Starting Position.
Often both the trainer and the trainee enjoy performing their assortment of tricks for visitors and strangers. But usually of greater interest to the average animal owner are the other benefits realized from trick training. Training a few simple tricks is a very effective way of training both the animal and the animal trainer to better benefit from the practice of offering frequent reinforcement. By training a few simple tricks, the trainer usually learns much from practicing these techniques that are equally effective in modifying troublesome behaviors such as biting and screaming. The trainee also learns from this training and often many troublesome behaviors will vanish on their own due to the training of a few simple tricks.
Many animal trainers start with a behavior that has been selected due to its simplicity and due to its usefulness in the future. Often that task is called Targeting.
Targeting = touch the end of a target stick.
Targeting is especially useful as a first behavior because it initially consists of only two options, touch or not touch. There are no in-between phases to confuse the animal. And since it is prop behavior, the animal is prevented from using this behavior to beg for additional treats, thus helping teach one of the concepts of good stimulus control. Once the initial task of touch the stick is mastered, good stimulus control can be developed by then expanding targeting to include: move some distance to touch the stick; touch the stick on cue; only touch the stick on cue; do not touch the stick unless cued; and touch the stick quickly when cued. Targeting can be generalized by training additional variations such as touch the stick with a different body part, touch or move to a specific location (like inside a travel cage), or fly to me to touch the stick.
Parrots: More Than Pets, Friends For Life
Chris Biro
For more material about this topic written in a less formal writing style see Handbook, by Chris Biro.