Degrees of unfalsifiability
As short post before I nod off, but something I’ve been wanting to blog about for a while now. Put simply, the falsifiability of a hypothesis or theory is one of the criteria that makes a hypothesis or theory a scientific one. A hypothesis needs to be able to be subject to testing, so that if false, its falsity will force a reversion to the null hypothesis.
Not all theories are testable. A general theory of God/first intelligence (deism) isn’t testable in that one can devise an experiment and from the results say that either an intelligence does or doesn’t exist. Existing empirical knowledge can be interpreted either way. God, generally, isn’t a scientific hypothesis. Specifically, God is only a scientific hypothesis when it makes testable claims.
For example:
“God is a visible milkshake sitting beside me as I write this, a God that didn’t create the Universe, isn’t intelligent and doesn’t do anything other than what a milkshake does”, is testable. I look beside me and there is no milkshake-God sitting next to me. There is no milkshake-God. The theory is false. It is the fact that it can be tested for this false state that makes it a scientific theory.
A theory that can be tested for falsity, yet isn’t found to be false (e.g. drug trials – in a scientific test, either a medicine works or it doesn’t) is an inductively strong scientific theory. The kind that science corrigibly lends its support to.
Of the kinds of theories that don’t meet the falsifiability criteria, I think there are three degrees.
- Perpetually unfalsifiable – The is where form of argument in a theory precludes testing. The Flying Spaghetti Monster (FSM) is a classic example of this, given that the FSM hypothesis claims that the FSM tampers with all forms of tests for its existence. A FSM proponent (not that there are any – I hope) can always claim in response to a negative “oh, the FSM made that happen”. It doesn’t matter what test results you get, it can always be interpreted as a positive for the FSM.
- Strongly unfalsifiable – This is where a scientifically known barrier to scientific endeavor prevents testing. For example, there is a theoretical upper limit (a bit of a blurred limit) to the amount of energy one can provide to a particle accelerator, so it follows that hypothesis requiring these higher energies (i.e. simulating the earliest minute moments following the Big Bang) to be tested, can not be tested. It is important to note that science, as a function of falsifiability, is corrigible about known barriers (e.g. the sound barrier, the lack of an atmosphere preventing space flight), so what is strongly unfalsifiable now, may not be in the future.
- Weakly unfalsifiable – This is where no scientifically known barrier to scientific endeavor exists, but rather just practical barriers exist. Perhaps CERN can not be provided with enough of the grid’s electricity to perform an experiment, but it may do later. Perhaps NASA can’t get the funds necessary to build an array of telescopes capable of sufficiently resolving distant objects. The reason I mention this category, one that may otherwise seem worthy of oversight, is because I see theories (in some but not all cases) descending from strongly to weakly unfalsifiable. I see Bertrand Russell’s “Celestial Teapot” falling into this category (there is now no reason to assume that it is beyond testing, it would just be very, very difficult to test), where during his own time (i.e. before space exploration) it was safer to assume it was strongly unfalsifiable.
That’s it. Post ends. 
~ Bruce
Bruce,
An offhand comment to an offhand post…
I think you’re playing with semantics here.
As I understand it, a scientific theory is basically a testable explanation for something. If it’s not falsifiable, it’s not a scientific theory.
Also, I think you’re conflating falsifiability and testability.
Here’s a hierarchy: theory, hypothesis, conjecture, guess.
Finally, FSM is a parody of monotheism, not a theory
Um, and yet again I reread my comment and wince.
Try again:
As I understand it, a scientific theory is basically a falsifiable (at least in principle, as opposed to testable) explanation for something. If it’s not falsifiable, it’s not a scientific theory.
Ah, but who adopts religion as a result of being rationally convinced that some god or other exists?
I think some of the Christians that I’ve met would say that they’ve had religious ‘moments’, instances of personal revelation, and that there is therefore some vaguely empirical basis for their belief. Not a testable basis, of course. But surely nobody ever came to religion because they were convinced by that Amway of arguments that comes under the rubric ‘ontological’?
I think you’re playing with semantics here.
I’m deadly serious. Corrigibility is a important criteria in science, if only as a consequence of falsifiability, but also as a function of repeatability and a bulwark against dogma. By making a distinction, I’m showing how science needs to be corrigible about what is or isn’t falsifiable.
Indeed, given that Bertrand Russell’s Celestial Teapot has moved along this scale, whereas the FSM by virtue of its nature can’t, and considering that the Celestial Teapot is still used in argument to this day (when the possibility of its falsifiability is near), I think the distinction is important.
Further to the above, I think that what I’ve called “weakly unfalsifiable” overlaps with “untested , but testable scientific theory”. This provides a pathway from metaphysics to the accepted scientific component of normative philosophy (with modification along the way). Sort of like reclaiming some of the nutrients from the cadaver of what Nietzsche proclaimed dead.
Also, I think it’s important in practice; a good criteria for determining what level of support a proposal needs. String theory, is strongly unfalsifiable, but through its inspiration of science proper (much like the atomic theory of Democritus), it warrants philosophical investigation.
Analogues of FSM are seriously taught in Universities, which seriously begs the question of their academic merit.
Also, I think it interesting (in a Kuhnian sense) that in it’s original paradigm, the Celestial Teapot is not falsifiable, but the exact same question posed in a modern paradigm isn’t necessarily so. Says something about the changing of terms across paradigms.
But now I’m waffling. I wrote this post to get back to it later when I needed it.
Also, I think you’re conflating falsifiability and testability.
You are perhaps right. Testability can be purely a search for purely inductive support ala confirmation bias, or just statistical justification for suspicion by way of correlation (i.e. to arrive at inductively weak answers). A test under falsificationism requires deductive reasoning to arrive at an inductively strong or unsupported (falsified) hypothesis.
I’m being a bit casual with my language, although I am relying somewhat on my readers being able to pick up what I’m getting at. I’m writing a blog here, not a journal!
Finally, FSM is a parody of monotheism, not a theory
I think this is a bifurcation. I think the FSM is a parody-theory.
Just think, if I made a categorical syllogism as devil’s advocate, is it no longer a categorical syllogism? Similarly, I don’t think FSM loses its theory status (albeit not a scientific theory) because nobody takes it seriously. Still, others take analogues seriously, so FSM as perpetually unfalsifiable makes a nice reductio, in my view more so that just calling it unfalsifiable.
As I understand it, a scientific theory is basically a falsifiable (at least in principle, as opposed to testable) explanation for something. If it’s not falsifiable, it’s not a scientific theory.
Yes, falsifiability and the problem of demarcation. I never intended to say otherwise, although I think some of the weakly unfalsifiable, overlapping with the untested falsifiable, actually does constitute science.
Ah, but who adopts religion as a result of being rationally convinced that some god or other exists?
Aww bum. I didn’t want this to be a theological discussion. :O
Great!