Extending the Fuzzy-Trace Theory informed model of guilty plea decision making: Accounting for factual guilt and innocence.

OBJECTIVE: Understanding how sentence differentials influence plea decision making is vital to identifying when plea offers might lead to convictions that conflict with normative goals (e.g., avoiding wrongful conviction). We clarify previous findings by introducing and testing predictions of a Fuzzy-Trace Theory informed model relating plea discount to plea acceptance in mock guilty defendants and extend the work to mock innocent defendants. HYPOTHESES: We expected plea rates in guilty defendants to increase with increases in plea discount but to change more steeply when a change in plea discount created a categorically distinct change in sentence (vs. a less meaningful reduction). Innocent defendants were expected to be unresponsive to changes in plea discount until a breaking point at which the discount created a sufficient categorical difference between plea and trial sentences to compete with innocence and the values associated with it. METHOD: We conducted a vignette-based experiment with 3,646 participants. We manipulated potential trial sentence (9 years, 18 years), factual guilt (guilty, innocent), and discount (nine levels; 10%-90%) in a between-subject design and asked participants to accept or reject an offer. RESULTS: As predicted, mock guilty participants were generally responsive to changes in plea discount but showed increased sensitivity at meaningfully distinct points along the range of discounts. Mock innocent participants showed a more extreme pattern. As expected, plea rates in mock innocent participants were generally unaffected by incremental changes in plea discounts but showed jumps at "break points," particularly when perceived conviction probability was high. CONCLUSIONS: Results support our conceptual model of gist-based plea decision making and demonstrate important differences in cognition underlying decisions in those asked to assume guilt versus innocence. (PsycInfo Database Record (c) 2026 APA, all rights reserved).