Implicit Bias, Reinforcement Learning, and Scaffolded Moral Cognition

Abstract

Abstract Many morally significant decisions depend on affectively valenced reactions. But these reactions are not simply performance errors or malfunctions of a moral system. They are the predictable result of attuning to social environments with reinforcement and calibrational learning mechanisms. This chapter shows how a better understanding of these mechanisms can help to explain why “snap judgments” often fall out of synch with reflectively held ideals. This approach also explains why attempts to regulate implicit bias by adopting better reflective attitudes are likely to fail; and it suggests ways of using counterfactual imagination to modulate existing associations. Finally, since the dynamic nature of social learning is likely to trigger backsliding into previously held and problematic attitudes, it shows that we should strive to eliminate implicit bias by engaging in a form of social niche construction, and making our world represent different and better things for us.