MT Testing Program: Four Predictions
Authors/Creators
Description
This program derives testable predictions of Mindsponge theory (MT) from the original book (Vuong, 2023) alone. Every assumption is tied to a quoted passage. Two modeling choices are ours and are flagged as such: a linear mixture of the trust and evaluation routes to value, and a standard observation model that links a perceived value to a probability of acceptance. From four claims in the book we derive four predictions. P1: when evaluation becomes costlier, reliance shifts from evaluation to trust, which lowers sensitivity to new evidence and moves all behavioral shifts along a single direction. P2: the weight given to content is greatest at indifference and falls toward zero at both ends of the trust gradient. P3: information held in the buffer zone can be accepted later, but heavily distrusted information is discharged and does not return. P4: a single cost-benefit threshold law holds at every level of organization. Each prediction is set against a named rival, with a proposition, a decision rule, and a simulation in which the rule recovers MT and every rival at least eight times in ten. One part of P1 has been tested and supported on open data, but the test does not yet separate MT from resource-rational accounts. The other predictions await suitable data.