Discrimination Learning, Training Methods

The term discrimination learning—learning to respond differentially to different stimuli—encompasses a diverse range of different types of learning, such as perceptual learning, concept learning, and language learning. The practical goal of discrimination training often involves complex discriminations, such as when pilots learn to recognize potentially hazardous patterns on complicated instrument panels. The greatest challenge in discrimination training—aside from the initial problem of perceiving differences between categories—is to accomplish long-term learning that transfers across situations and stimuli. Training techniques that produce good performance can be misleading; difficult training is sometimes more effective than easier training in the long term.

Discrimination learning occurs at many levels, such as perceptual learning, higher level concept and category learning, and language learning. A baby [Page 251]experiences all three types of learning: She has to learn basic perceptual categories (e.g., discriminating between ba vs. ga), higher level concepts (e.g., friends and relatives vs. strangers) and linguistic labels for these categories and concepts (e.g., dog refers to furry pets but not cats).

Learning to discriminate often involves a process of differentiation and unitization. Differentiation involves learning to separate similar stimuli into different categories (e.g., ravens vs. crows); unitization involves learning to group stimuli into larger units and to discriminate based on the unit (e.g., seeing a whole face instead of separate facial features). Unitization and differentiation may appear to be opposite processes, but both involve learning to perceive information at the appropriate level of representation.