How Human Trusters Assess Trustworthiness in Quasi-virtual Contexts

Abstract

A virtual encounter is one in which an agent has only symbolic information about her coactors. In a quasi-virtual one she has a smidgeon of non-symbolic, sensory information. Internet encounters are virtual; and quasi-virtual ones are normal in laboratory experiments. In these experiments we want to limit the information the decision maker gets in order to have control (if you let people meet there’s much too much going on to tell afterwards what drove the behavior). But it is hard to achieve complete virtuality because people see the tops of each other’s heads and have other fleeting sensory contact. Sometimes we want the smidgeon because it ‘engages’ the subject’s human reactions, and these may be the object of study.

Prof. Michael Bacharach has been invited speaker at the AAMAS-02 workshop on “Deception, Fraud and Trust in Agent Societies” held in Bologna on July 15, 2002. The following paper is the contribution he sent us for the workshop proceedings. The presentation and ideas shown in Michael’s talk were, in fact, really more complete and explanatory than these notes. Nevertheless, it is possible to identify also in this blueprint some relevant and interesting ideas that the genius of Prof. Bacharach was able to produce. Michael’s sudden disappearance on August 12, 2002 has left us without an outstanding personality in this field.