Disagreement about the best policy is due to different preferences or beliefs, or both. Believing that different preferences cause the opinion differences discourages debate (no point arguing over taste after all), leads to polarisation and partisanship. For example, right-wingers may believe that left-wingers prefer to disincentivise entrepreneurs with high taxes, and left-wingers may believe that right-wingers prefer to harm the poor by reducing government transfers. To put it starkly: the other side just prefers evil policy by nature.
By contrast, believing that disagreement over what should be done is caused by differing beliefs assumes that the other side is good-hearted, but mistaken. For example, left-wingers may believe that right-wingers mistakenly believe that transfers to the poor disincentivise them from working or finance their addictions. Right-wingers may believe that left-wingers mistakenly believe that entrepreneurs are not discouraged by higher taxes – being entrepreneurial by nature, they start companies because it is interesting, not out of greed. Mistaken opponents’ opinions can be corrected using data and logic, patience and understanding.
Even if policy disagreement is interpreted as coming from divergent preferences, some such differences are interpreted as less evil than others. For example, impatience is perceived as better than selfishness. Many policies trade off non-simultaneous benefits and costs: invest in infrastructure now to use it after some years, mitigate climate change now to reduce harm to future generations. Paying a current cost for a future benefit may be optimal for patient people, but not for impatient, causing a policy disagreement. The same opinion difference may be due to altruistic people wanting to invest to help others (future users of the infrastructure or the environment), but selfish ones preferring to keep the money now. Believing the same disagreement to be due to selfishness polarises people more than perceiving unequal patience as the cause.