Hmm - what about if we factor in the percieved value of trusting someone for a specific purpose; we then trust someone for some purpose if the probability of the their actually being trustworthy times the value of trusting is greater than the probability of their not being trustworthy times the cost of misplaced trust.