Discussion groups are a promising tool for bridging the divide between former conflict antagonists. However, such groups do not always produce the desired outcome of improved attitudes, even when they meet the conditions generally seen as favoring positive interaction. In this article, we examine specific discussion protocols that mitigate polarization risks while fostering reconciliation. Using a randomized, controlled design, we formed a pool of 429 ex‐combatants and members of conflict‐affected communities in Colombia. Participants were asked to join heterogeneous groups and discuss their proposals for the future of Colombia. Overall, community members improved their attitudes towards ex‐combatants significantly, while ex‐combatants’ attitudes toward community members do not tend to polarize. Those participants who were randomly assigned to a perspective‐giving treatment protocol (where they were asked to refer to their personal experience and perspective) consistently improved their intergroup attitudes towards ex‐combatants, and by a proportionally higher percentage than those taking part under argumentation and no‐treatment control conditions.