Experiments suggest that animals, just like humans, tend to prefer exaggerated, supernormal stimuli, and that a preference can rapidly propel itself to extreme levels (peak shift effect). In one experiment, through food rewards rats were conditioned to prefer squares to other geometric forms. In the next step, a non-square rectangle was introduced and associated with an even larger reward than the square. As expected, the rats learned to reliably prefer the rectangle. Less predictable was the third part of the experiment. The rats were offered the opportunity to choose between the rectangle they already knew and associated with large rewards and another rectangle, the proportions of which were even more different from those of a square. Interestingly, rats picked this novel variant, without undergoing any reward-based conditioning in favor of it. A possible explanation is thus that they chose the larger difference from the original square (i.e., the exaggeration of non-squareness).
In an experiment, after first establishing an ____(A)____ to squares, and then to non-square rectangles, rats were seen to pursue ____(B)____ rectangularity even without any additional reward.