Some Effects of Punishment

My next big paper/presentation is on Humanism, so in doing some research I came across a really interesting paper written by the Psychologist Dr. Hans Eysenck. He describes two interesting experiments done with rats and dogs:
In the first one, a rat is put at the bottom of a t-maze, and so reaching the end can turn right or left. In all 3 conditions food is placed on the right end and want to teach him to turn right. One group is rewarded with food when they turn right, a second group is punished by shock when they turn left and rewarded when turn right. A third group is punished when they turn right in addition to getting the food.

You would think that if punishment deters learning, the third group should learn less quickly and 2nd group most quickly, but as it turns out the the 2nd and 3rd group both learn equally more quickly than group 1. It seems strange, but it doesn't matter if the reward and punishment are paired or paired as right or wrong. The point made against traditional forms of punishment like serving jail time for stealing is that it doesn't matter if they were successful at stealing or not, but if the thieves are punished it is reinforcing the behavior regardless.

In the next experiment dogs are put in compartment A, divided by a low fence from compartment B which the dogs can jump over. They are given a signal, and a few seconds later compartment A is electrified, making the dog jump to B, which they quicklylearn. Now to teach them not to jump to the signal. In group 1 the researchers give the signal but do not shock so if the dog hesitates it learns not to jump. Group 2 has no shock in compartment A but electrocutes B instead. If punishment for jumping into B is effective in making them learn not to jump, they should stop the habit more quickly than group 1, but the dogs punished for jumping continue to jump into compartment B and never learn it is safe to remain in A. They behave like recidivists who return to criminal activity each time they are punished for criminal behaviour.

The idea from both of these is that punishment can stamp in behaviour we are punishing rather than stamp it out. This in itself is pretty profound (and this research was done decades ago), however it seems that in the case of education, this could be used as an argument to punish children in school. Since memory works better when you punish the subject regardless of whether they were right or wrong, wouldn't regular random punishment make students smarter? There must be some reason they don't do this because they haven't allowed hitting in school for quite a long time now.