In this paper, I will first briefly summarize some aspects of what we now know about the nature of biases in emotional processing. I will then turn to the main issue that I want to address-whether such biases are invariable or can be modified, and if so, how. I will present evidence that emotional biases are far from constant, vary according to state and context, and can be modified by treatment and by experimental manipulations. Experiments on modifying emotional biases demonstrate that they can cause emotional changes rather than being a secondary consequence of mood state. Finally, I will argue that these findings suggest a new direction for research-away from just demonstrating that biases exist, and towards research directed at questions of change and causality.