Trust cannot be modeled as rational choice or probabilistic prediction

Evidence

Trust is not a means that can be chosen for particular ends, much less an ends/means structure capable of being optimized. Nor is trust a prognosis; its correctness cannot be measured by the occurrence of the predicted event and after some experiences have been reduced to probability values. These types of techniques, which are significant within the framework of decision-making models, have, as does trust, the function of reducing complexity. They are functional equivalents for trust, but not acts of trust in the true sense. As far as they extend, trust is unnecessary. They can replace trust, just as, conversely, the need for trust as a complementary way of absorbing uncertainty is a result of the limited effectiveness of those decision-making techniques. Trust is, however, something other than a reasonable assumption on which to decide correctly, and for this reason models for calculating correct decisions miss the point of the question of trust. — Niklas Luhmann, Trust and Power, 307

Categorization

• DEFINE: Distinguishes trust from functionally equivalent uncertainty-reduction techniques like decision calculus — trust is irreducible to rational-choice or probability models.
• XREF: Directly relevant to Giddens’s distinction between trust in expert systems vs. calculative assessment; also connects to Stevens’s deference being non-calculative in Ishiguro. — connects to: Ontology cannot predict the content of any individual’s free acts