Moral Emotions for Robots

摘要

As robotics moves toward ubiquity in our society, there has been only passing concern for the consequences of this proliferation (Sharkey, 2008). Robotic systems are close to being pervasive, with applications involving human-robot relationships already in place or soon to occur, involving warfare, childcare, eldercare, and personal and potentially intimate relationships. Without sounding alarmist, it is important to understand the nature and consequences of this new technology on human-robot relationships. To ensure societal expectations are met, this requires an interdisciplinary scientific endeavor to model and incorporate ethical behavior into these intelligent artifacts from the onset, not as a post hoc activity. We must not lose sight of the fundamental rights human beings possess as we create a society that is more and more automated. One of the components of such moral behavior, we firmly believe, involves the use of moral emotions. Haidt (2003) enumerates a set of moral emotions, divided into four major classes: Other- condemning (Contempt, Anger, Disgust); Self-conscious (Shame, Embarrassment, Guilt); Other-Suffering (Compassion); Other-Praising (Gratitude, Elevation). Allen et al (2006) assert that in order for an autonomous agent to be truly ethical, emotions may be required at some level: While the Stoic view of ethics sees emotions as irrelevant and dangerous to making ethically correct decisions, the more recent literature on emotional intelligence suggests that emotional input is essential to rational behavior. These emotions guide our intuitions in determining ethical judgments, although this is not universally agreed upon (Hauser, 2006). From a neuroscientific perspective, Gazzaniga (2005) states: 'Abstract moral reasoning, brain imaging is showing us, uses many brain systems', where he identifies the locus of moral emotions as being located in the brainstem and limbic system.