Maryland Metacognition Seminar -- Folk Intuitions about Selves and Responsibility: First computational steps.

Fri Jul 27, 2012 12:00 PM

Folk Intuitions about Selves and Responsibility: First computational steps

Paul Bello
Office of Naval Research, Arlington, VA

We are ultimately interested in the question of whether or not it is possible to build moral machines. This question comes along with obvious philosophical and computational baggage, much of which has been put into focus by researchers in the nascent field of "machine ethics." There has been much hand-wringing about the tractability of algorithms meant to generate ethically *correct* decisions; even leading some machine ethicists toward a thoroughgoing moral behaviorism and the subsequent renouncement of folk-psychological concepts as useful for moral evaluation. For various reasons mentioned in the talk, we find these worries unwarranted at the moment, and the behavioristic alternative to be deeply problematic. Moreover, we find this sort of reasoning symptomatic of a broader strand of thinking within machine ethics that is obsessed with building globally-rational moral machines. In this talk, we briefly argue to the contrary, and provide a positive proposal for how to move forward by first understanding the information-processing architecture of human moral cognition. To do so involves computationally modeling the folk concepts associated with our moral intuitions, and exploring how they might be dependent upon the nature of human cognitive architecture.
In this talk, we computationally explore a case study of human intuitions about the self and responsibility. We hypothesize that a significant portion of the variance in reported intuitions for this case might be explained by appeal to an interplay between the human ability to mindread (i.e. to ascribe mental states to both self and other) and the way that knowledge is organized conceptually in the cognitive system. In the talk, we build on a pre-existing computational model of mindreading by adding constraints related to psychological distance, a well-established psychological theory of conceptual organization. Our initial results suggest that studies of folk concepts involved in moral intuitions lead us to an enriched understanding of cognitive architecture and a more systematic method for interpreting the data generated by studies of human folk concepts.


Paul Bello (RPI, 2005) is a program officer at the Office of Naval Research in the Warfighter Performance and Protections Department. His portfolio spans a wide range of basic research in computational cognition and artificial intelligence. Specific foci include technologies for rich inference over multi-representational data, natural language dialogue, socially-guided learning in human-robot teams, and the cognitive science of moral judgment. His intellectual interests lie in the human capacity to detach from the real world in order to consider the past, possible futures, pretenses, hypotheticals, counterfactual alternatives, and the contents of other minds. Most of his research is focused on building computational accounts of mental-state ascription that are consistent with developmental, behavioral, and neuropsychological data.

Read more about the Maryland Metacognition Seminars.