We propose a policy specification approach that combines social sciences and trust theory to facilitate ad-hoc interactions of self-interested parties in open environments. Our socio-cognitive approach allows us to reason about uncertainty and risk involved in a transaction, and automatically calculate the minimum trust threshold needed to mitigate the vulnerabilities. The trust threshold comprises the core of security policies that govern the interactions. The threshold calculation is based on balancing objective and subjective trust components, which together predict that a transaction will result in an acceptable outcome. We propose to apply the Prospect Theory [20] to specify policies that determine a set of acceptable outcomes. We present the trust threshold negotiation primitives.