Generalization (Learning)

Generalization, in learning, is the concept that humans, other animals and artificial neural networks apply what they learned in past situations to a present situation of learning when the conditions of the two are regarded as similar. It allows more efficient handling of new situations by recognizing similarities to situations already experienced, though the resulting assumption is not always accurate: a person who has an allergic reaction to apples might conclude they are allergic to all fruit and decline a banana on that basis, a generalization that sometimes holds and sometimes does not. Educational research documents both beneficial and harmful outcomes from generalization in learning, and it is typically contrasted with discrimination learning, in which a learner comes to distinguish differences between similar situations rather than treating them as equivalent.