Peter Kreeft – Professor of Philosophy at Boston College – writes (Angels and Demons, 1995, pp. 32-3):
“Isn’t the supernatural unscientific?
Yes. Science can’t prove it.
But it’s not antiscientific, because science can’t disprove it either. All the reasonings about all the observations of all the events within the system of nature can’t disprove something outside of nature [. …] If angels can cause changes in nature – if angels can stop a speeding car in two seconds or appear and disappear in bodies – then the human senses can detect their effects […] But the cause is invisible, since angels are pure spirits. Science cannot observe angels. They don’t reflect light.” (original italics)
This isn’t quite right. Science can only detect effects. It then postulates causes based on the effects. So, angels wouldn’t be unscientific in this regard.
So, what makes angels seem unscientific? Is it that they might exist outside of space and time, and so are non-natural in a sense? I think that’s getting ahead of ourselves: there is another reason why many people consider postulating angels to be unscientific.
First, the phenomena that Kreeft mentions above are erratic – like many natural phenomena, such as meteorites falling to earth, it is difficult to replicate them in a laboratory. This just means that until one gets very good reason to believe it occurs, it’s easy to disregard reports or postulate closer to hand explanations, if not doing so requires a (hypothetically) large metaphysical shift from the normal kinds of explanations.
An analog to this would be bug detecting in coding. Sometimes, bugs are highly replicable (and therefore usually easy to solve). Other times, though, one gets an “erratic bug.” It seems to occur at unusual times, without any seeming reason. Often, one can’t even replicate the supposed bug – it’s rather a customer that is claiming to have one. One could jump to the conclusion that the bug isn’t real, or that the bug is not part of the code (but rather a problem with, say, the customer’s operating system). Yet, often that’s not the right conclusion to jump to. Rather, the first culprit is usually something going on within the code. Even so, though, figuring out the cause of the erratic effect may be difficult.
Similarly, most scientists probably think of angelic phenomena like Kreeft mentions above in a similar way: it’s difficult to say what, exactly, is going on. For the time being, goes the line of thought, let’s assume it’s something occurring within the established natural framework of causes and effects, and save more radical hypotheses about it for once we’ve eliminated the closer to hand possible causes.
(This is combined with the question: does it make sense to focus on trying to explain these phenomena right now? Similarly, one might not focus on an erratic bug, because it seems more likely one will make more progress by focusing on replicable bugs instead. Perhaps it will be solved along with solving something else, or perhaps at a later point one can return to it, but for now, there are other things to do, goes the thinking.)
Second, the phenomena Kreeft doesn’t mention but seem more replicable – angels guiding humans (bringing them messages) – also seems explainable by something else: namely, part of the brain (or so the presumption goes). (Neurotheology investigates effects like this.)
In both cases, there’s nothing unscientific about postulating angels per se, except that it seems that there are easier explanations at hand, given the picture of the world that science already has. Both this and the first consideration are instances of Occam’s Razor. Perhaps, though, upon close inspection of the phenomena, there doesn’t seem to be any good conventional explanations at hand, and something like an angelic being outside of space and time does start to seem reasonable – just as sometimes, when learning about a bug, it starts to seem likely that it is something outside the code causing it: perhaps it’s in the customer’s operating system, or hardware, or some other program running at the same time. In this case, it becomes reasonable, and scientific, to start postulating other causes. There’s no general rule here, but rather an informed sense of what could be going on.
Having said all this, Kreeft is on to something right here. Warranted belief for a ‘supernatural’ cause is not possible in science in a sense, because “to be physical” = “to be in the physical causal network,” where the way that something is judged to be in the physical causal network is if it has effects on other things in the physical causal network. So, if one has good evidence of effects in the physical causal network, then the postulated causes become linked to the physical causal network, i.e., they become ‘physical’ also. As far as something becoming part of the physical causal network becomes ‘natural’, then causal evidence for ‘supernatural’ processes is impossible.
(Consider the idea that the universe was ‘birthed’ from another universe, and that there are many universes in existence. Is the original universe ‘supernatural’? That’s not my intuition – it’s natural, and physical, while existing outside of this universe.)
The corollary of this conclusion is the fact that the definition of what is physical is changing to fit whatever can be posited to cause effects in the physical causal network. So, electromagnetic causal processes would not have been considered ‘natural’ or ‘physical’ at some point in the past – but now they are paradigmatic.
Also see here.