Motivated reasoning has become the central framework for explaining attitude polarization between members of different social groups. For instance, in the US context, when Democrats and Republicans respond to the same information (e.g., economic indicators) in polarizing ways or a gulf emerges in surveys (e.g., support for Covid-19 vaccines) then these differences are usually attributed to the process of motivated reasoning. This theory posits that polarization is the result of an automatic, subconscious process fueled by hot cognition. Attitude polarization, however, can be the result of a variety of different processes that are not fueled by automatic biased information processing. It could emerge because members of different social groups update their beliefs from uncommon priors or because people consciously distort their opinions to align with their social identity.

This paper introduces original experimental manipulations that disentangle the mechanisms driving attitude polarization. Across multiple experiments with diverse applications to US politics (e.g., response to economic information, responses to politically violent rhetoric) and using different salient social groups (e.g., partisanship, race), the paper showcases how not all attitude polarization is created equal. While some divergence is attributable to motivated reasoning, results suggest that much of the polarization traditionally credited to this process is actually the result of conscious and deliberate attitude distortion as well as people evaluating the same information using different rational standards. Taken together, the paper showcases that research needs to be more precise when examining attitude polarization as well as the need to probe the mechanisms underlying such polarization.