Must reason be preferred when it conflicts with revelation?

When a rational proof and a scriptural proof appear to conflict in matters of belief, which is preferred? This is the question of the "universal law" on which Ibn Taymiyya wrote Dar' Ta'arud al-'Aql wa'l-Naql; its editor Muhammad Rashad Salim gives in his introduction the texts of the theologians and philosophers who held the law. ^{1 2}

As Ibn Taymiyya quotes it at the start of the book: if scriptural and rational proofs conflict, either both are accepted, which is impossible since it combines contradictories; or both are rejected; or scripture is put first, which is impossible since reason is the root of transmission and putting scripture first would impugn its root; so reason must be put first, and the text is then either interpreted or its meaning left to God. ³ He calls this a "universal law" of al-Razi and his followers, anticipated by al-Ghazali, the judge Abu Bakr Ibn al-Arabi, al-Juwayni and al-Baqillani. ¹

• Al-Razi: in Asas al-Taqdis he divides the case where certain rational proofs establish something and transmitted proofs seem to say otherwise into four options, rejects three, and concludes that one must hold to what reason requires and judge the texts either inauthentic or not meant in their apparent sense; then, if interpretation is allowed, one interprets, and if not, one leaves their knowledge to God – "the universal law to which one refers in all ambiguous texts". ^{4 5} In Muhassal Afkar al-Mutaqaddimin wa'l-Muta'akhkhirin he says that verbal proof gives certainty only when ten conditions are met, one being the absence of a rational objection. ⁶
• Al-Ghazali: in Qanun al-Ta'wil he counsels never to give the lie to rational demonstration, since it is by reason that we came to know revelation, gives examples of what must be interpreted, and then counsels refraining from fixing an interpretation when possibilities compete, holding suspension safer. ^{7 8}
• Al-Juwayni: in al-Irshad he divides the foundations of belief into what is known by reason only, what is known by revelation only, and what may be known by both. ⁸

Ibn Taymiyya says these theologians are less extreme than the philosophers who claim that the prophets' texts do not match reality and were addressed to the common people in terms suited to their minds, such as the law Ibn Sina set out in al-Risala al-Adhawiyya. ⁴ The editor quotes Ibn Sina's view that pure tawhid, put to the masses as it is, would have met obstinate refusal, so the revealed laws addressed them through likeness and image; and Ibn Rushd in Manahij al-Adilla, that the apparent sense is the duty of the masses and the interpreted sense the duty of scholars, and in Fasl al-Maqal, that where demonstration conflicts with the apparent sense of revelation, that sense admits of interpretation according to the Arabic rules of interpretation. ^{9 10 2}

After a general answer, Ibn Taymiyya says that asserting a conflict between rational and scriptural proof and insisting on preferring the rational is "known to be corrupt by necessity", then gives his detailed answer in aspects. The first: if the two conflicting proofs are certain, their conflict is not conceded as possible; if they are probable, the stronger is preferred absolutely; if one is certain, the certain one is preferred, and if the rational one is the certain one it is preferred for being certain, not for being rational. ^{11 12} The second: the division into four options is not exhaustive, since the rational may be preferred at one time and the scriptural at another according to certainty and strength. ¹² The third: the claim that "reason is the root of transmission", if it means that reason is the root of revelation's truth in itself, is something no rational person says, for what the Messenger reported is true in itself whether or not we know it; reason is only the root of our knowledge of its truth. ^{12 13} He takes the aspects to forty-four. ²

The texts of al-Razi, al-Ghazali, al-Juwayni, Ibn Sina and Ibn Rushd here are as Muhammad Rashad Salim quotes them in his introduction, not from editions of their books; Ash'ari replies to Ibn Taymiyya and academic studies of the question have not yet been added.