The problem of defining intelligence independent of specific education and culture has been considered important for human intelligence testing since the beginning of the 1900s. Charles Spearman suggested in 1904 that there might be a general intelligence factor (usually called g) associated with all intellectual tasks. Its nature was never very clear, but it was thought that its value could be inferred from performance on puzzles involving numbers, words and pictures. By the 1980s, however, there was increasing emphasis on the idea that different types of human tasks require different types of intelligence. But throughout the 1900s psychologists occasionally tried to give general definitions of intelligence—initially usually in terms of learning or problem-solving capabilities; later more often in terms of adaptation to complex environments.
Particularly starting at the end of the 1800s there was great interest in whether animals other than humans could be considered intelligent. The most common general criterion used was the ability to show behavior based on conceptual or abstract thinking rather than just simple instincts. More specific criteria also included ability to use tools, plan actions, use language, solve logical problems and do arithmetic. But by the mid-1900s it became increasingly clear that it was very difficult to interpret actual observations—and that unrecognized cues could for example often account for the behavior seen.
When the field of artificial intelligence began in the mid-1900s there was some discussion of appropriate definitions of intelligence (see page 1099). Most focused on mathematical or other problem solving, though some—such as the Turing test—emphasized everyday conversation with humans.