Positive reinforcement training for laboratory mice
The mouse (Mus musculus) is the most commonly used mammal in biomedical research and can easily be restrained by both hand and by a device. Restraining animals causes stress that not only decreases the animal?s wellbeing, but may also influence the research results. The first aim of this study was to train mice to voluntarily stand still using positive reinforcement training. Another benefit of teaching a mouse to stand still is that you can avoid anaesthesia for some procedures. The second aim of this study was to investigate if postnatal handling could also facilitate learning.
That mice can be trained using positive reinforcement training (PRT), has been shown in numerous experiments as well as on video-sharing websites such as YouTube, yet most training of laboratory mice is pretty basic; focusing on habituation by calm, gentle handling and patience. In PRT, animals are being reinforced for performing a desired behaviour and will not be forced to cooperate. A reinforcer could be anything the animal likes, such as food or access to a favourable environment. Giving the animals control over their situation and a chance to work for their food are both factors that are associated with increased wellbeing.
Postnatal handling is a brief and daily separation of the pup from the mother for the first two to three weeks of the pups? life. This has been shown to have several different effects on the animals; mice that have been handled are more explorative, less anxious and have a lower response to stress response than unhandled mice. The effects of postnatal handling are thought to be mediated by a changed maternal behaviour towards the pups after handling and/or a result of the handling itself. Several experiments with mice have shown that pups receiving increased maternal care had a reduced stress response. There are many factors influencing the effect of postnatal handling; e.g. gender, strain and the time of day when being handled.
Housing factors have been shown to affect the brain development of laboratory animals. Mice are known to develop stereotypies when raised in standard laboratory cages. Stereotypies are repetitive behaviours that can indicate a suboptimal environment. An enriched environment has been shown to improve learning and memory in mice.
Our experiment included six pregnant NMRI females and their offspring. The females were handled prior to parturition and were presented with different rewards until the pups were weaned. This was to evaluate which one was most popular and make sure the pups were used to the smell of the reward from an early age. Half of the litters were randomly selected to be handled prior to weaning. 24 of the pups were randomly assigned to be part of the training group with equal distribution between genders and handling/non-handling. All mice were always lifted in a cupped hand by the researcher and never by the tail.
Training was done using a device with a sensor that triggered a reward-signal when the mouse was in the correct position. The reward was delivered through a metal nipple in the device. To further evaluate the effects of handling two behavioural tests were performed; an elevated plus maze and a handling test. A skilled animal technician, who fixated the mice by hand and scored them as compliant, hesitant or unwilling, performed the blind handling test. The mice had no prior experience of this person.
The most popular reward in the study was the strawberry ice cream. After spending 14 times training in the device, nine out of 24 mice had learnt to take the reward from the metal pipe. Five out of six unhandled females had learnt this task and only two males had learned to do so. The device did not function as planned and the reward-signal was not used since the refraction time was too long. Instead
reward was given to the mice with increasing intervals from 0 seconds up to 3.5 seconds by the end of the experiment.
In the elevated plus maze, the only difference was (when pooling both genders) that the non-handled animals exhibited a significantly higher number of stretch attend postures when compared to the handled animals (20,0 ± 5,8 vs. 27,6 ± 6,4 (P