Altered dopamine transmission is thought to influence the formation of persecutory delusions. However, despite extensive evidence from clinical studies there is little experimental evidence on how modulating the dopamine system changes social attributions related to paranoia, and the salience of beliefs more generally. 27 healthy male participants received 150mg L-DOPA, 3mg haloperidol, or placebo in a double blind, randomised, placebo-controlled study, over three within-subject sessions. Participants completed a multi-round Dictator Game modified to measure social attributions, and a measure of belief salience spanning themes of politics, religion, science, morality, and the paranormal. We preregistered predictions that altering dopamine function would affect i) attributions of harmful intent and ii) salience of paranormal beliefs. As predicted, haloperidol reduced attributions of harmful intent across all conditions compared to placebo. L-DOPA reduced attributions of harmful intent in fair conditions compared to placebo. Unexpectedly, haloperidol increased attributions of self-interest for opponents' decisions. There was no change in belief salience within any theme. These results could not be explained by scepticism or subjective mood. Our findings demonstrate the selective involvement of dopamine in social inferences related to paranoia in healthy individuals.