In his attempt to define the boundaries of Islam al-Ghazâlî singles out a limited number of teachings that in his opinion overstep the borders. In a separate book, The Decisive Criterion for Distinguishing Islam from Clandestine Unbelief (Faysal al-tafriqa bayna l-Islâm wa-l-zandaqa) he clarifies that only teachings that violate certain “fundamental doctrines” (usûl al-‘aqâ’id) should be deemed unbelief and apostasy. These doctrines are limited to three: monotheism, Muhammad's prophecy, and the Qur’anic descriptions of life after death (al-Ghazâlî 1961, 195 = 2002, 112). He stresses that all other teachings, including those that are erroneous or even regarded as “religious innovations” (singl. bid’a), should be tolerated. Again other teachings may be correct, al-Ghazâlî adds, and despite their philosophical background, for instance, should be accepted by the Muslim community. Each teaching must be judged by itself, and if found sound and in accordance with revelation, should be adopted (al-Ghazâlî 1959a, 25–27 = 2000b, 67–70). This attitude leads to a widespread application of Aristotelian teachings in al-Ghazâlî's works on Muslim theology and ethics.

Al-Ghazâlî's refutations of the falâsifa and the Ismâ’îlites have a distinctly political component. In both cases he fears that the followers of these movements as well as people with only a cursory understanding of them might believe that they can disregard the religious law (sharî’a). In the case of the Ismâ’îlites there was an additional theological motive. In their religious propaganda the Ismâ’îlites openly challenged the authority of Sunni theology, claiming its religious speculation and its interpretation of scripture is arbitrary. The Sunni theologians submit God's word to judgments that appear to be reasonable, the Ismâ’îlites said, yet they are purely capricious, a fact evident from the many disputes among Sunni theologians. No rational argument is more convincing than any of its opposing rational arguments, the Ismâ’îlites claimed, since all rational proofs are mutually equivalent (takâfu’ al-adilla). Only the divinely guided word of the Shiite Imam conveys certainty (al-Ghazâlî 1964b, 76, 80 = 2000b, 189, 191). In response to this criticism al-Ghazâlî introduces the Aristotelian notion of demonstration (burhân). Sunni theologians argue among each other, he says, because they are largely unfamiliar with the technique of demonstration. For al-Ghazâlî, reason (‘aql) is executed most purely and precisely by formulating arguments that are demonstrative and reach a level where their conclusions are beyond doubt. The results of true demonstrations cannot conflict with revelation, al-Ghazâlî says, since neither reason nor revelation can be considered false (Heer 1993, 186–88). If demonstration proves something that violates the literal meaning of revelation, the scholar must apply interpretation (ta’wîl) to the outward text and read it as a symbol of a deeper truth. There are, for instance, valid demonstrative arguments proving that God cannot have a “hand” or sit on a “throne.” These prompt the Muslim scholar to interpret the Qur’anic passages where these words appear as symbols (al-Ghazâlî 1961, 175–89 = 2002, 96–103). The interpretation of passages in revelation, however, whose outward meaning is not disproved by a valid demonstration, is not allowed (Griffel 2000, 332–35; 2009, 111–16).

Al-Ghazâlî's rule for reconciling apparent conflicts between reason and the literal meaning of revelation was widely accepted by almost all later Muslim theologians, particularly those with rationalist tendencies. Ibn Taymiyya (1263–1328), however, criticized al-Ghazâlî's rule from an scriptualist angle. Ibn Taymiyya (1980, 1:86–87) rejected al-Ghazâlî's implication that in cases of conflict between reason and the revealed text, priority should be given to the former over the latter. He also remarked that al-Ghazâlî's own arguments denying the possibility that God sits on a “throne” (Qur’an 2.255), for instance, fail to be demonstrative. Ibn Taymiyya flatly denied the possibility of a conflict between reason and revelation and maintained that the perception of such a disagreement results from subjecting revelation to premises that revelation itself does not accept (Heer 1993, 188–92).

On the falâsifa's side Averroes accepted al-Ghazâlî's rule for reconciling conflicts between reason and the outward meaning of revelation but he did not agree with his findings on what can and cannot be demonstrated (Griffel 2000, 437–61). Averroes composed a refutation of al-Ghazâlî's Incoherence, which he called The Incoherence of the [Book of the] Incoherence (Tahâfut al-tahâfut). This work was translated twice into Latin in 1328 and 1526, the later one on the basis of an earlier Hebrew translation of the text (Steinschneider 1893, 1:330–38). The two Latin translations both have the title Destructio destructionum (the later one is edited in Averroes 1961). They were printed numerous times during the 16th century and made al-Ghazâlî's criticism of Aristotelianism known among the Averroists of the Renaissance. The Italian Agostino Nifo (c.1473– after 1538), for instance, wrote a Latin commentary to Averroes' book. While accepting the principle that only a valid demonstration allows interpreting the Qur’an symbolically, Averroes maintained that Aristotle had already demonstrated the pre-eternity of the world, which would elevate it, according to al-Ghazâlî's rules, to a philosophical as well as religious doctrine. Averroes also remarked that there is no passage in the Qur’an that unambiguously states the creation of the world in time (Averroes 2001, 16). Al-Ghazâlî was clearly aware of this but assumed that this tenet is established through the consensus (ijmâ’) of Muslim theologians (Griffel 2000, 278, 429–30; 2002, 58). While al-Ghazâlî condemns the pre-eternity of the world at the end of his Incoherence of the Philosophers, the subject of the world's pre-eternity is no longer raised in his later more systematic work on the boundaries of Islam, The Decisive Criterion for Distinguishing Islam from Clandestine Unbelief.