THE RELATIVE EFFICACIES OF ESCALATION AND DEESCALATION FOR COMPLIANCE-GAINING IN TWO-PARTY CONFLICTS

Two independent experiments were conducted to investigate the determinants of target compliance to escalatory and deescalatory influence strategies during a two-person, mixed-motive conflict simulation. Both experiments involved the intermittent transmission of contingent threats of punishment with varying magnitudes from a robot to target subjects. In the first experiment, an escalatory (increasing) pattern of threats and punishments was employed and in the second, a deescalatory (decreasing) pattern was used. Escalation and deescalation were implemented either suddenly and noncontingently by robot, or gradually and contingent on target's response to previous influence exertion. Complementary cells of the two designs were nonparametrically compared to determine the relative compliance-gaining efficacies of escalation vs deescalation. Escalation of punishments was far more effective in eliciting compliance overall than was deescalation (p < 0.0005); these differences persisted within both the sudden (p < 0.0252) and gradual (p < 0.0054) conditions separately. Consistent with a utility analysis, however, sudden deescalation was not significantly less efficient in gaining compliance than gradual deescalation (p > 0.30). These results were discussed as contradictory of simple expected value models of compliance, and deescalation was seen as presenting a significant but limited alternative to escalation for cooperative conflict resolution.