Trust operates as Bayesian belief updating grounded in cooperation experience.

Evidence

My capacity is constrained by the weight of past experience with all of the reassessment and revision of my views that this experience has stimulated. In a Bayesian account of knowledge, for example, I make a rough estimate of the truth of some claim—such as that you will be trustworthy under certain conditions—and then I correct my estimate, or “update,” as I obtain new evidence on you. If I take the risk of cooperating with you, I soon have some evidence on whether you are trustworthy in that single context. I might test further and further, updating until I have a good sense of your degree of trustworthiness in various contexts. I might do this—indeed, typically would do it—not necessarily to test you but rather to benefit from cooperating in new ways. Hence trust—the belief in another’s trustworthiness—has to be learned, just as any other kind of knowledge must be learned. — Russell Hardin, Trust and Trustworthiness, 526

Categorization

• DEFINE: Formally characterizes trust as iteratively updated belief about another’s trustworthiness, modelable in Bayesian terms.
• XREF: Directly relevant to integrating with game-theoretic accounts of repeated interaction and to why Saramago’s blindness breaks the evidence-gathering loop that makes trusting rational. — connects to: Bayes’ rule is the unique rational way to update belief — Cailin O’Connor and James Owen Weatherall, The Misinformation Age: How False…, 108