Abstract
The author follows the distinction between transfer proper and transfer effect and utilizes the bar-pressing expt. involving a vertical and a horizontal bar in order to study transfer effect in a situation in which the principle of generalization (transfer) is present. The subjects were 640 O albino rats. Varying amts. of food-rewarded training were given with training on the horizontal bar always preceding the training on the vertical bar. Extinction of one or the other habit was obtained 22 hrs. after learning. There was no correlation between learning time and number of extinction responses. Measured by savings in learning time there was transfer effect from habit H to habit V which was negative for small amounts of preliminary training and positive for larger amounts. Measured by resistance to extinction there was transfer effect from habit H to habit V which was positive when habit V was based on a small amt. of training and negative when V was based on a large amt. In terms of resistance to extinction there was a positive transfer effect from habit V to habit H. The author concludes that present theory is inadequate to account for the apparent paradoxes in his results.