Chapter 6 Summary

By the early 1970s, research had uncovered several surprising properties of taste aversion learning. Writers began to question the generality of the laws of learning that had been discovered up to that point.
Many of the unusual properties of aversion learning were eventually also found in other examples of classical conditioning. These properties began to be explained by an expanding set of general learning principles.
Research on taste aversion learning produced many insights into the general aspects of learning. We now understand that learning serves a biological function and that not all cues are equally associated with various consequences. Function and evolution have a powerful influence on learning.
General learning processes may evolve because of exaptation: the concept that a learning mechanism adapted for solving one particular problem may work well enough at handling other problems. True adaptive specialization may require that a mechanism adapted to handle one problem be functionally incompatible with the solution of another problem.
At a descriptive level, conditioning principles often seem surprisingly general across species and conditioning preparations. For example, the relative validity effect—originally demonstrated in rats, rabbits, and pigeons—has now been demonstrated in human categorization and causal judgment. It has also been shown in classical conditioning in honeybees. The honeybee result is interesting because the bee’s brain evolved independently of the brains of rats, rabbits, pigeons, and humans.
There are many correspondences between conditioning in honeybees and vertebrates, but there are also possible differences. Behavioral outcomes that look similar at a descriptive level may sometimes result from different processes.
There are also many correspondences between classical conditioning in animals and categorization and causal learning in humans. Categorization and causal learning may obey conditioning principles like the ones in the Rescorla-Wagner model. Another idea is that they involve other processes, such as probabilistic contrast and the perception of causal power.
Research on associative learning in humans has uncovered new effects. One is backward blocking. Although backward blocking was not predicted by models of classical conditioning, those models are being extended to account for it.
Some investigators have argued that conditioning involves learning propositions about the relationships between events, like CSs and USs, in the world. The Perruchet effect, however, suggests that processes that control conditioned responses and verbal reports can be dissociated and can operate separately.
In the long run, research that has examined the scope and generality of learning laws developed in the animal learning laboratory has helped us develop better general laws.