Abstract
<p>Discussions can end in a shared view, in persistent disagreement, or in little change. Such differences are usually explained by participants' initial opinions, by who interacts with whom, and by how one opinion influences another. This study examines a further, largely overlooked dimension: how much of a speaker's view is transmitted in a single contribution, which we call utterance scope. Views are not lists of independent opinions but webs of judgments about which considerations support or conflict with one another, and a contribution may transmit one such judgment or many linked judgments at once. Studying this dimension in a computational model of discussion, we find that its consequences differ across levels. In a single interaction, what matters is not how much is transmitted but how many of the receiver's judgments would actually change, and which party holds the more internally coherent view: coherent views resist revision, so an asymmetry resembling biased assimilation arises without motivated reasoning. In repeated discussion, utterance scope reshapes collective outcomes. Narrow contributions leave participants differentiated; broad contributions consolidate change into a shared direction and align remaining differences along a single axis; and new combinations of existing judgments are most abundant at intermediate scope. These differences are not explained by the total amount of change; they emerge as taken-up judgments circulate through later contributions. How much of their views people convey at a time is thus a distinct lever shaping where discussions go.</p>