Unnecessarily divided: Civil conversations reduce attitude polarization more than people expect

People with opposing attitudes can learn from one another through civil discourse and debate. Yet, people routinely avoid discussing their differences of opinion, preferring instead to discuss their attitudes with like-minded others. We propose that people lack interest in discussing their differences of opinion, in part, because they expect such conversations are unlikely to change their own and others' attitudes. Importantly, we find these expectations are systematically miscalibrated: Civil conversations reduce attitude polarization more than people anticipate. Participants with opposing attitudes toward cats and dogs (Study 1 and Supplemental Study S1), cancel culture (Studies 2 and 4), and Joe Biden's performance as president (Study 5) underestimated how much their own and others' attitudes would depolarize in spoken conversations. Moreover, participants retained somewhat less polarized attitudes 1 week later. Participants underestimated attitude change, because they misunderstood why their attitudes differed: Whereas participants inferred their attitudes differed, because they fundamentally disagreed; their attitudes actually differed, because they were focused on different aspects of these topics (Study 3). As such, having a conversation surfaced unexpected areas of agreement (Studies 2, 4, and 5). Importantly, participants became more interested in discussing their differences of opinion, when they were informed that their own and others' attitudes might depolarize in a conversation (Study 6 and Supplemental Study S2). In total, the current work reveals that miscalibrated expectations can create an unnecessary barrier to civil discourse, leaving people with diverse points of view more divided, more polarized, and less informed than they otherwise could be. (PsycInfo Database Record (c) 2026 APA, all rights reserved).