Boolean relations, such as and, or, and not, are a fundamental way to create new concepts out of old. Classic psychological studies showed that such concepts differed in how difficult they were to learn, but did not explain the source of these differences. Recent theories have reinvigorated the field with explanations ranging from the complexity of minimal descriptions of a concept to the relative invariance of its different instances. We review these theories and argue that the simplest explanation – the number of mental models required to represent a concept – provides a powerful account. However, no existing theory explains the process in full, such as how individuals spontaneously describe concepts.