One of the major issues in conditioning is selective attention. How does a person or animal select out, from the multitude of stimuli it encounters, the ones that will be predictive of important positive or negative consequences in the future, and learn to act on such stimuli appropriately? The solutions to this problem in a neural network context have broad implications not only for the psychologist but also for the engineer who seeks to build goal direction or planning into intelligent devices. A collection of articles in progress (Levine & Leven, in press) focuses on the interconnected issues of motivation, emotion, and goal direction.