Training procedures

Submitted by Sysop on Wed, 10/27/2021 - 01:58

Training procedures are necessary to establish cooperation with the animal. Training has to be adapted for a given animal species, since reaction paradigms differ between specises and animals react differently to reinforcements.

The most common elements to train are:

  • the animal to accept a reinforcement (usually preferred food items)
  • establish a favourable relationship between the trainer and the animal
  • establish a secondary reinforcement that does not interfere with the trials (for instance a whistle that has frequency that is not similiar to any tested frequency), this secondary reinforcement is used to decouple the signalling to the animal, whether the answer was correct or not from the actual reinforcement, which cannot always be given at the same time
  • establish a feeding schedule that incorporates the animals needs as well as the best feasible approach to train the animal, only well treated animals will give you a correct answer for your research question
  • train the animal in medical training
  • train the animal to touch a target
  • train the animal to station for at least some seconds, so that it is stationary during the trial
  • train the animal to couple a presumed well hearable, but not overly loud sound (for instance a click) to the correct reaction (touching a button/target etc) until the animal answers with a high correct detection rate
  • perform the trials

 

Training for the trials can be a science by itself and it is important to report how animals are trained and reinforced. The reinforcement scheme can vary drastically. If for instance correct rejections are inforced with double amount of food, the experimenter can train the animal to only report well hearable sounds. If correct reactions are enforced favourably for the animal, it will listen closely to the sound and rather accept a false detection (false positive), than reporting a false non-detection (false negative). This creates a bias that is called the "response bias".