Trust as an Energy-Saving Shortcut: Two Open-Data Tests of Mindsponge Theory Under Time Pressure
Description
Mindsponge Theory (MT) holds that minds filter incoming information through subjective cost-benefit judgments, and that trust saves energy by letting information in without full evaluation. If this is right, making evaluation costlier should shift a mind’s reliance away from evaluating new evidence and toward what it already trusts. We derived two predictions from this claim and tested each on open data, with the decision rule written into a script, checked by simulation against rival accounts, fingerprinted, and run once. The first test used a four-armed bandit task in which time pressure was varied within each of 99 people (72,958 choices). Under time pressure, people became less likely to abandon an option after a bad outcome (from 0.556 to 0.440) and did not become more likely to abandon it after a good one (from 0.273 to 0.225). The interaction of pressure with a bad outcome was −0.21 logit (95% CI −0.32 to −0.11), and the prediction was supported. An exploratory check with a second definition of a bad outcome gave the same pattern. This pattern rules out the possibility that time pressure merely adds noise or merely adds a habit of repeating choices. The second test used three truth-judgment experiments with 317 people, in which time pressure was varied between participants. MT, on one reading, predicts that the boost that repetition gives to belief should grow under time pressure, because repeated information resembles what the mind already holds and can pass with less evaluation. It did not grow: the repetition effect was 0.32 logit, and pressure changed it by −0.02 (95% CI −0.16 to 0.12). The prediction was not supported. Taken together, the results suggest that the energy-saving shortcut operates on what a person has come to trust through experience, not on mere familiarity, which is consistent with the book’s statement that a value must first be accepted before it can serve as a reference for trust. The positive result does not separate MT from resource-rational accounts of cognition. We set out the tests that could do so and the data they require.