Learning Functional Relations Based on Experience WithInput-Output Pairs by Humans and Artificial Neural Networks

Before making any serious decision, we normally try to anticipate how the effects of our action will vary depending on the action taken. Forexample, before an anaesthetist can decide the amount of anaestheticto administer to a patient, she needs to predict how the analgesic effectwill vary as a function of the amount injected. Before a father can decidethe amount of money to invest in his son’s college education, he needsto predict how the return will vary as a function of the size of theinvestment. The point is that prediction is essential to decision making.