Everyone knows that when the late Russian physiologist Ivan Petrovich Pavlov rang a bell, his dogs started salivating in anticipation of food. But about a century later, scientists still aren’t exactly sure why.
The scientists conducted a cellular experiment to test the mechanisms involved and showed how brain cells associate a brief command with a delayed reward.
“When you teach a dog the command sit, you first tell it to sit and then a while later, long after the sound waves have left the room and after the brain has stopped responding to the sound, you give the dog a cookie,” Shouval said.
The prevalent theory, according to Shouval, has been that once an animal experiences a stimulus, a process called a synaptic eligibility trace is initiated. This eligibility trace lasts awhile and if a reward appears before the trace decays, this triggers a change in networks of neurons in the brain.
Until recently, this theory has not been proven.
In their paper, Shouval and scientists from Johns Hopkins University and the University of California, Davis show that such eligibility traces indeed exist and can form a link between the stimulus and the reward. The study was conducted in a mouse model.
“But this paper has shown more; it has shown that two eligibility traces exist, one to signal a strengthening of connections between neurons and the other to signal a potential weakening of connections,” said Shouval, who is also on the faculty of The University of Texas Graduate School of Biomedical Sciences at Houston.
In addition, Shouval explains why two different opposing traces are necessary. When a dog learns that it will get a cookie when it sits after a command, it actually knows it will get one cookie, not two, or three. In order to learn this, the animal has to stop learning once it has reached the correct result. The two opposing eligibility traces allow the animal to stop learning because they compete with each other and once the correct result is reached, this competition balances out and learning stops.
This research has shown that these two such eligibility traces not only bridge the gap but they are also set to compete with each other, such that once correct learning is obtained they cancel each other out, he said. These two traces therefore allow dogs to learn they will get a cookie if they sit when instructed, but also prevent them from constantly being disappointed.