158 BRAIN MECHANISMS AND LEARNING 



and thus cause a diffuse interference; and (3) utilizing penetrating im- 

 planted electrodes, we are in a better position to get beyond the neocortex. 

 Furthermore, we use a discrimination reversal experiment, to be 

 described below, so that after preliminary training the animal can learn 

 and re-learn the same problem day after day, eliminating errors each day 

 in roughly the same number of trials. This methodology allows us to test 

 animals one day on rate of learning under the electric 'lesion' and another 

 day to test the same animals on control ; and we may repeat the process as 

 often as we choose. 



METHODS 



To date, 150 rats with implanted electrodes have been tested in a 

 problem box where they had to learn over again each day which pathway 

 and lever would activate a feeding mechanism. The problem box is shown 

 in Fig. 6. A red plastic lever on the left activated the food magazine one 

 day, and a white metal lever on the right activated it the next. Each day the 

 rat had to learn anew which lever worked. This took a small number of 

 runs, after which the animal would run only to the correct level. After 

 each pellet was discharged, the animal had to go to the magazine and eat 

 (thus breaking the photo-beam) befoi'e the magazine would work a second 

 time. Thus animals could not stay at the pedal discharging several pellets 

 and then go to the magazine to eat them all, but were forced to run the 

 maze both ways for each pellet. Ten consecutive responses on the correct 

 lever was the criterion of learning. All animals were put through extensive 

 preliminary training (running every day for more than a month) before 

 being used in experiments. By the end of this training period, animals 

 would always begin each day by distributing runs randomly to right and 

 left; then they would stabilize on the lever that activated the food maga- 

 zine. Depending on the animal, this would take from three to twenty-five 

 runs each day. Animals were not used in experiments until progressive 

 improvement in this daily score had ceased. During training and tests, 

 animals were maintained on a feeding schedule with 23 hours of starvation 

 and I hour of feeding each day. They were tested after 22 hours of 

 deprivation. 



The data were plotted in terms of cumulative response curves (see Fig. 7). 

 Solid lines were used to indicate correct responses, dotted lines, to indicate 

 errors. The slope of the line indicated the speed of responding (on the 

 correct or error pedal). When the dotted line flattened out, this indicated 



