In this paper, I intend to contribute to the Islamic analytic theological engagement with modern science, focusing on the problem of scientific methodology. In the first part, I lay out what I take to be some basic commitments involved in the ideal Muslim relation to God and creation, which pertain to the conception and practice of science. Second, I critique Nidhal Guessoum’s procedure of defining methodological naturalism, understood as the distinguishing feature of modern science, on the basis of an unexamined distinction between the ‘natural’ and ‘supernatural.’ Third, I suggest that an alternative—exemplified by a proposal from Scott Tanona, of defining the ‘natural,’ in this context, in methodological terms—is preferable, both in terms of clarity and in maintaining the distinction between methodological and ontological naturalism.
Trusting in God and tying your camel
We all know the tradition of the Prophet ﷺ about trusting in Allah and tying your camel. I want to take this as my starting point because it so eloquently represents the ideal Muslim relationship between God and creation. In order to explicate some of the philosophical implications latent in this ideal attitude, it is useful to reconsider a challenge that has been perennially raised against it.
Let us imagine that someone puts her trust in God and ties her camel, and then returns to find her camel still in place. The skeptical challenger might say the following: Either God kept the camel in place or the rope did. If the rope kept the camel in place, then God did not, and so trust in Him was misplaced. If God kept the camel in place, then the rope did not, and so there was no sense in tying it. So either there was no sense in tying your camel, or your trust in God was misplaced. We, of course, are committed to the position both that there is sense in tying your camel, and that trust in God is well placed. So our response must be that the challenge rests on a false dilemma. But it is one thing to assert this, and another to show it. Showing it also involves a process of self-clarification. Where does the challenger’s argument fail?
First, we most certainly must maintain that God kept the camel in place. But secondly, we must also maintain that there is rational sense in tying it. More specifically, there is rational sense in tying it, as opposed to other possible courses of action. The Prophet ﷺ did not say, “Trust in God and cross your fingers,” or “Trust in God and eat your vegetables,” for example. This is because there is a meaningful sense in which tying your camel keeps it in place, whereas crossing your fingers or eating your vegetables does not. Therefore, we cannot concede that, since God keeps the camel in place, the rope does not; and then try to make sense of tying the camel in some other way. We are, in some sense, committed to the position that tying the camel makes sense because, all things being equal, it does keep the camel in place.
Moreover, the fact that tying your camel keeps it in place is not something known through revelation. The purpose of this hadeeth is not to inform us of the general effectiveness of tying camels. We already know that empirically. The hadeeth enjoins us to rely on our empirical knowledge of the natural world in our dealings with it, and informs us that this is not contrary to, but indeed connected to, trust in God. So taking seriously this maxim to ‘trust in God and tie your camel,’ means acknowledging that tying your camel is effective in keeping it in place, and that this can be known through means that are natural, by which we mean independent of revelation (though not, of course independent of God). It means, trust in the Creator and take recourse to creation in the way which you have natural reason to believe will be effective in accomplishing your goal. Thus, it asserts a proper scope for what may be called a kind of ‘methodological naturalism.’ The question is, what is its scope and what, precisely, makes it ‘naturalistic’?
Some of the parameters regarding the scope of a methodological naturalism emerge from reflecting on two further postulates, which I assume we all take to be axiomatic to the Muslim viewpoint. First, if God does not keep the camel in place, the rope will not. Second, if God does keep the camel in place, it will remain in place with or without the rope. Reflection on this pair of commitments brings to light that, for the believer, the proposition that God kept the camel in place is unfalsifiable. If she leaves her camel untied, and then returns to find that it strayed off, she will say that God made it stray, and that if it had stayed in place it would have been God who kept it there. There is, then, no possible outcome that could count against the proposition that whatever happens with the camel is God’s will. We are quite happy to agree that the proposition is not falsifiable, and therefore not scientific in the sense in which falsifiability is taken as a necessary condition for being ‘scientific.’
But we must, of course, oppose the challenger’s predictable assertion that, in case the camel does stray, trust in God was misplaced. This is simply because the purpose of trusting in God is not, in fact, to keep the camel in place. Therefore, whether the camel stays or strays is not an appropriate test of the proposition that trust in God is well placed. Trusting in God has an altogether different purpose from tying the camel, independent of whether the camel stays or strays. And so we are also happy to concede that the proposition, that trust in God is well placed, is not falsifiable and therefore, unscientific. By contrast, the proposition that (all things being equal) tying your camel will keep it in place, is falsifiable, and therefore an appropriate subject for scientific experimentation. The rope can be tested, but not God.
Finally, our challenger will likely raise this question. Given our assertion that, if God does not keep the camel in place then the rope will not; and that it will stray if God makes it stray, with or without the rope, then how do we claim that tying it will keep it in place, such that there is any rational sense in doing so? We have effectively claimed that God’s willing it is both necessary and sufficient for the camel staying in place. What, then, is left for the rope to contribute to the situation?
This is not, of course, a new question. In asserting that there is a rational sense in tying the camel, we are committed to treating the rope as ‘secondary cause’ in the classical sense in which this is distinguished from the ‘primary cause,’ which in every case is God. This distinction has been made for a long time, and is sufficient, perhaps, for maintaining our theological position. But assuming that we want to ‘tie our camels,’ so to speak, as systematically and scientifically as possible, there is good reason to examine carefully what it is to be a secondary cause; not only to clarify the distinction between secondary and primary causation, but to distinguish between secondary causation and other, non-causal creaturely relations.
In the 17th discussion in the Incoherence of the Philosophers, Al-Ghazali famously argued that ‘existence with’ a thing does not constitute ‘existence by’ it, thus distinguishing primary causation, as ‘existence by,’ from secondary causation, as ‘existence with,’ relegating the former to God alone, and creation to the latter. So far so good, for the theological position, taken by itself. But doing science effectively—‘tying your camel’ intelligently—requires recognizing that every ‘existence with’ is not equal in this respect. Mere correlation is not causation, neither in the primary nor the secondary sense. Perhaps all secondary causation is ultimately some form of correlation, but science cannot get very far on the notion that all correlation is causation. A viable Islamic philosophy of science calls for not only a distinction between primary and secondary causation, but for an analysis of secondary causation that clarifies the difference between secondary causation and non-causal forms of correlation (assuming for the moment that secondary causation is some form of correlation).
So far, two tasks have been proposed. One is defining the precise meaning and scope of methodological naturalism. The other is giving a scientifically workable analysis of secondary causation. And of course, we would like to do both in a manner compatible with Islamic theological commitments. The two topics are closely related, but in what follows I will limit my focus to the first.
Modern science and the ‘nature’ of methodological naturalism
With regard to the question of methodological naturalism, I would like to start with some critical comments on the paper Nidhal Guessoum wrote for Kalam Research and Media, “Kalam’s Necessary Engagement with Modern Science.” Guessoum views methodological naturalism as the fundamental characteristic of modern science. Methodological naturalism, he writes, “insists that science only admit explanations of natural phenomena that rely solely on natural causes and leave out entirely any appeal to supernatural agents, be they spirits, angels, demons, or indeed God.” Guessoum takes care to distinguish methodological naturalism from what he calls philosophical naturalism—what I will call here ontological naturalism—which is the substantive denial of the existence of any supernatural entities, including God. Methodological naturalism, by contrast, is only a provisional, hypothetical assumption of ontological naturalism, for the purpose of scientific research. In spite of this, he proceeds to raise some problems between Islamic religious belief and modern science, which could only emerge by way of equivocating on this distinction.
One such problem, according to Guessoum, is “how to reconcile a naturalistic study and explanation of the world/nature and the belief in a present/personal God. Does He act in the world, and if so does this conflict with modern science; or does He not act (at least not physically) in the world?” Guessoum writes that he personally advocates the ‘spiritual action only’ position (whatever that means). But so long as modern science really does only commit us to a provisional assumption of ontological naturalism for limited methodological purposes, and not to a substantive metaphysical position, then this problem should not arise. It can only arise when we adopt the position that the methods of natural science are appropriate to determine whether or not (and how) God acts in the world, and that constitutes a substantive metaphysical position. It is an attempt to tie down God, and the world as a whole, with a rope that is only suitable for camels.
This raises the question, precisely what we are assuming, when we provisionally assume ontological naturalism, and for what purpose? So far, the answer to the first question is just that we are assuming the nonexistence of the supernatural. So what, precisely, is the difference between that and the natural? Guessoum says nothing other than to mention “spirits, angels, demons, or God.” But what do these things have in common? Could we throw in Santa Claus, elves, Bigfoot, and phlogiston? The point is that no principled distinction seems to be operating here, and without that, it remains unclear what methodological naturalism is. And if that is taken as the defining feature of modern science, it remains impossible to enter any real engagement with it. We can only either blindly submit or blindly resist its epistemic hegemony.
So let us engage. If methodological naturalism is the provisional assumption of an ontology, then we should presumably draw the distinction between natural and supernatural in ontological terms. From a classical Muslim theological perspective, for instance, one might identify the natural with the creation, and the supernatural as the Creator. A more philosophically informed classical view might be that the natural is spatio-temporally bound and subject to change, while the supernatural is not. These two positions may or may not be equivalent, depending on one’s metaphysical commitments. Neither, however, clarifies what Guessoum has in mind, because demons, spirits, and angels are all created, and at least ‘demons’ (or jinn) are spatio-temporally bound—assuming that we are taking Islamic ontology seriously in our engagement with modern science.
In the absence of any clear distinction between the natural and the supernatural, Guessoum’s definition of methodological naturalism is much less informative than his explanation of why it became a pillar of modern science. That is, he writes, because “the assumption of supernatural factors as explanations was quickly identified as a ‘science stopper,’ an end to the explanatory process and thus a non-productive (or even counter-productive) approach to progress (progress in finding further truths about nature and devising useful applications).” Shall we, then, not just define a ‘natural explanation’ as one that does contribute to finding further truths about nature and devising useful applications? That is, the distinction between natural and supernatural will not be ontological at all, but one defined by the requirements entailed in a specific objective, characterized, perhaps, as that of explaining what we do not understand in terms of that which we do, and can at least possibly manipulate, rather than explaining it in terms of something which we do not understand and therefore cannot manipulate. As Guessoum puts it, “if a doctor explains some mental disorder as the work of demons, s/he will not be able to understand the deeper brain processes at work, nor will any medication be found…”
But if we draw the distinction in this way, then Guessoum’s definition of methodological naturalism has it backwards. Whereas Guessoum bases his characterization of the modern scientific method on a fixed (but undefined) distinction between the ‘natural’ and the ‘supernatural,’ the present suggestion is that, in this context, ‘natural’ is defined by the requirements entailed by the specific aims of the modern scientific method. Thus, we need not assert that methodological naturalism is the provisional assumption that only a certain kind of thing exists, only to find ourselves unable to say what kind of thing that is. Instead, we can say it is the provisional assumption that all that exists meets the requirements entailed by the objectives of modern science. Consequently, this distinction between the ‘natural’ and the ‘supernatural’ is highly contingent. It will depend on the requirements of our objective, which depend, in turn, on the nature of the objective itself.
Conceiving the ‘natural’ in methodological terms
Imagine that jinn (or ‘demons’, as Guessoum calls them) are like germs. They are out there causing illness, but we have so far been unable to detect or observe them under controlled conditions. In this case, consideration of jinn as the cause of mental illness is not conducive to devising treatments (though it may be conducive to using treatments that we could not devise ourselves but have been informed of by other means). So it is a ‘supernatural’ explanation, in that respect. But if, in the future, we develop the means of discovering and studying them systematically, and find that some mental illness is demon possession, is there any remaining principled reason to call this a ‘supernatural’ explanation? It would be no more ‘supernatural’ than explaining a crime scene as the result of human agency. Elsewhere in his paper, Guessoum asks us to consider seriously the prospect of hitherto undiscovered non-human intelligent life, presumably of the extra-terrestrial sort. Would an explanation of disappearances as alien abductions, then, be less supernatural than that of mental illness as demonic possession?
Imagine then, that we scientifically discover that extra-terrestrial demons exist and have been causing mental illness by abducting people and possessing them. (As an aside: if they landed in a spaceship and introduced themselves, would they then count as having been ‘scientifically’ discovered?) At any rate, space-demons would be far less useful than germs, as an explanatory factor, in relation to the objective of curing illness, since they presumably would have a complex psychology, and as such their behavior would be at least as difficult to predict and manipulate as that of a human being. This, of course, is the reason questions have been raised as to whether the social sciences and economics can ever attain the status of the ‘hard sciences.’
But clearly, it is a mistake to judge the explanatory factors operative in these sciences by the same standards that we judge those of physics or biology, because the objectives of these sciences are quite different. As such, the kinds of phenomena that are epistemologically useful as explanatory factors are different in each field. Therefore, what is admissible as ‘natural’—in this sense of methodologically correct—will differ from discipline to discipline. The alternative would be, for example, to classify the social sciences as given over to ‘supernatural’ explanations and therefore ‘pre-modern,’ which is implausible.
On the other hand, it might be suggested that we define what counts as ‘natural’ specifically in terms of what meets the epistemological requirements entailed by the objectives of physics, biology, and the like. Then we would have reasonable grounds for not classifying the human and social sciences as ‘natural sciences’, in virtue of not operating under a naturalistic methodology so defined. But that would not also be grounds for labeling them as ‘supernatural.’ It would just mean that the ‘natural’ is not to be understood simply in a binary opposition to ‘supernatural.’ Not all that is not natural is ipso facto ‘supernatural.’ This would include human behavior and culture, as well as any hypothetical space-demon behavior and culture. It would also mean that identifying methodological naturalism as the essential characteristic of modern science would effectively exclude the social sciences from modernity.
Furthermore, the objectives within a given scientific discipline itself are not constant, but have undergone, and most likely will continue to undergo, historical development. If then, our conception of methodological naturalism, understood as the defining feature of modern science, is fixed to a specific set of epistemological requirements, then we will either have to deny the possibility of any further development in the objectives of our sciences as they currently are, or to acknowledge the possibility that science itself might outgrow ‘methodological naturalism,’ and therefore, modernity.
But at bottom, one might insist, certain fundamental objectives are universal. And this is correct. The objective of science is to ascertain the truth, and that of technology is to achieve our goals. The necessary relation between the two is clear. But these are universal to both the ‘modern’ and ‘pre-modern.’ Methodological correctness as a norm, in the most abstract sense, is not then sufficient to distinguish the modern. That would have to be based on the specific nature, of the objectives understood as ‘modern,’ and of our ideas about the correct methods of achieving them. But are these not subject to critique and development?
So long as we are defining methodological naturalism as a provisional acceptance of only ‘natural’ explanations, for certain methodological purposes, then the clearest way forward is to define ‘natural,’ in this context, as just the sort of explanatory factor that lends itself to those purposes. And yet, a degree of specificity in our conception of those purposes and the epistemological desiderata they entail is necessary if anything interesting and useful is to be inferred with respect to methodology. And yet again, it should be a conception that is stable and universal enough to render a general methodological prescription applicable to the range of scientific disciplines, and open to critical development with regard to their most basic objectives and presuppositions.
To that end, let us return to the basic ontology of Creator and creation, and the heuristic of tying the camel, as representative of our epistemic and practical relation to creation as such. In this case, our purpose is to understand creation in terms of itself, insofar as that is possible. From here, we recall the important point, which emerged in the classical encounter between Islamic kalam and falsafa, that if creation were not sufficiently ordered, it would be naturally unintelligible, and thus impossible to understand on its own terms, and to accomplish our goals therein. If there is simply no telling what God will bring about if I do or do not tie my camel, then action in the world is futile. The problem of induction may show us that we can never know how orderly things really are. If so, then it is a problem precisely because if nature is not orderly enough knowledge of nature is impossible. That is, knowledge about the world cannot be rationally pursued except under the assumption that things are sufficiently ordered. And this will be true irrespective of how our scientific objectives and basic suppositions might develop over time. Are there other specifications we can make?
Scott Tanona has recently proposed an interesting formulation of the natural, in terms of the epistemological desiderata of intersubjectivity and predictability. “A phenomenon is natural if it involves regularities between intersubjectively definable aspects of the world,” he writes, “where those regularities constrain the possible values of those intersubjective aspects to allow for predictions about them, including, if applicable, under interventions.” An ‘intersubjectively definable aspect’ is simply any aspect that we describe and measure in a way that is accessible to more than one subject (or observer). The supernatural, on the other hand, is “beyond any such intersubjective predictability,” such that it “is not merely contrary to currently-understood physical theory, but is beyond our ability to incorporate into any future theory of empirically testable physical regularity or law.”
Tanona agrees with Guessoum that methodological naturalism does not entail ontological naturalism. In fact, he makes it explicit that his own motivation is to defend the distinction from charges of unsustainability. He couches this distinction in terms of ‘internal’ and ‘external’ naturalism, where the former is the application of naturalism within the domain of science alone, while the latter refers to its universal application.
Internal methodological naturalism is based on the minimum requirement of any scientific account that it make specified contact with intersubjective data, and that with that contact it makes specified, predictive claims about other intersubjective data. It says neither that such accounts will be successful (never mind everywhere) nor that they must be exclusive.
This formulation of methodological naturalism, in other words, does not amount to the positive position that everything in the natural world is explicable in terms that lend themselves, in principle, to intersubjective predictability. It is simply the practice of limiting one’s considerations to factors that do so lend themselves, for the specific purpose of bringing as much of nature that does, in principle, to do so in actuality. This is perfectly consistent with the awareness that there are some things that, either contingently or in principle, do not so lend themselves. It is also perfectly consistent with the awareness that nothing of the order that has hitherto been discovered can eliminate the real possibility that in the next moment, it will be radically altered. So as the Prophet ﷺ advised, if you are planting a tree and the Day of Judgment arrives, then continue planting.
 Sunan al-Tirmidhī 2517
 See Popper, Karl. Logik der Forschung. Verlag von Julius Springer, Vienna. 1992.
 Ghazali, Abu Hamid. Incoherence of the Philosophers. Michael Marmara, trans. Brigham Young University Press, 1997.
 For more on this subject, see Moad, Edward. “Occasionalism and Contemporary Theories of Causation,” forthcoming.
 Guessoum, Nidhal, “Kalam’s Necessary Engagement with Modern Science,” Kalam Research and Media Monograph Series, No. 6. 2011, pp. 4.
 Guessoum 5
 Tanona, Scott, “The Pursuit of the Natural,” Philosophical Studies, Vol. 148, No. 1 (March 2010), pp. 82
 Ibid. 83
 Musnad Aḥmad 12491