Harry Harlow (1949), presented monkeys with random S+/S- pairings for six trials at a time and studied the monkeys' abilities to solve the discrimination problems. The monkeys' performance improved with each subsequent trial from trial and error, stimulus-based responding to "insight" across novel problems. The animals learned new discriminations with fewer and fewer trials over time out of a series of hundreds of discrimination problems (i.e., they "learned to learn"), demonstrating one-trial learning in nonhuman primates. Harlow argued that this type of performance was not explained by reinforcement history, demonstrating that monkeys had developed a win-stay, lose-shift problem solving strategy and are capable of abstract thought. This belief was contrary to conventional learning theory and behaviorism at the time and suggested that a more cognitive approach needed to be incorporated.