I have, previously, discussed how absolute ontological certainly is precluded by the structure and nature of knowledge. I also recently engaged in a public debate that touched on the nature of the null hypothesis and reasonable beliefs. My opponent from that debate recently asked me to clarify my position on the issue of falsifiability. Specifically, the request was:
[W]hat is ‘falsifiability’? You seem to take it to mean that a proposition yields predictions which can then be compared to empirical evidence and shown to be false. Is that correct?
My response was that that statement was “Pretty much” accurate, but that I’d need to write up a blog post “since the answer is somewhat lengthy.” Dr. Chenvi’s followup question was:
[H]ow that definition is applied to beliefs like “There are no transcendental truths” or “We cannot have real knowledge of deep reality” or “We should only accept claims based on evidence”, or “the external universe actually exists”, etc… None of these claims are falsifiable, on the basis of your definition. So whatever statements you make about whether or not we should believe ‘unfalsifiable’ claims should wrestle with statements like these.
- P -> Q
- ∴ P
In simple English, the fallacy can be put it as “If it rains, the street will be wet. The street is wet, therefore it rained.” But, of course, as there are other possibilities (e.g. a fire hydrant was opened), you cannot conclude P from Q. You can, however, fail to prove the falsity of a claim if you don’t look for evidence of it. The only way to negate P is to change the proof to:
- P -> Q
- ¬ Q
- ∴ ¬ P
That is, mere conditionality does not prove a necessary truth: only a biconditional can do that. Absent a biconditional, one may only state that P is a sufficient cause of Q. And when P is merely one potential, a lone disjunctive premise merely leaves us with possibilities. All that simply touches on validity. Soundness remains absent without the ability to confirm that P –> Q is valid in the first place; in order for a proof to be sound, we must be able to both confirm that it is valid, and that all of its premises are true. Absent the ability to verify its truth value, a proof can only have indeterminate soundness. That conclusion is a statement about the language game of logic, and is true-within-the-system-of-deductive-reasoning: it’s no more a metaphysical statement than “you can’t use your hands in a game of soccer unless you’re the goalie.”
Falsifiability Via Utility, As Evinced By Objective Reality: The discussion ultimately rests on utility, and the structure and limitations of inductive and deductive reasoning. Utility relies on objective changes in reality, and while there are endless brain-in-a-jar and living-in-the-matrix objections to this, the ultimate fact of the matter is that our survival, in the context of this reality, rests upon the assumption that reality is ‘as real as it gets’. Various logically consistent ‘other realities’ can be posited, in point of fact an infinite number may be posited – but without a winnowing criteria, there is no way to know which one is true. A lone inclusive disjunction cannot validly prove any of its terms. For example, the following proof is invalid.
- A ∨ B ∨ C ∨ D
In a nutshell, it is invalid to conclude C unless one can also prove ¬ A, ¬ B, and ¬ D.
We can construct cogent inductive proofs to argue for C, but inductive proofs are not, and can not be, truth preserving. Denying falsifiability in deductive logic leaves one with the unenviable position of arguing for an epistemic system that can’t differentiate between fact and logical possibility. To put a finer point on it, without falsification criteria, it isn’t possible to know if A, B, C, or D are even ontologically possible, let alone existential potentialities.
This is not a statement about metaphysics any more than “a home run scores one point in baseball”; which is to say, they are both deductively sound, provided that one accepts their axioms. If, then, we accept that an epistemic system is to be judged by its utility, the rest falls into place neatly – it’s a fairly easy metric to use. But it must be stressed, again, that this is not a statement on the ultimate nature of reality; it is a maximally fit epistemic system predicated upon reliably and repeatably producing changes in objective reality in conformity with will.
Empirical Evidence: If we rely on utility, the ultimate standard becomes objective reality; the difference between a viper and a stick; the difference between food and poison; the difference between a person and a corpse. Again, one may choose to abandon this standard. But if we do not, then we must necessarily privilege any epistemic system that evinces the greatest ability to provide us with the highest degrees confidence in the truth value of claims, repeatably and reliably. Otherwise, we’re left with a sea of disjunctive claims where the premises may be logically but not ontologically possible, let alone actually true.
Transcendent Truths And Knowledge of Deep Reality: There may exist transcendent truths, but as demonstrated in the first link in this post, they are impossible to know, provided that one accepts one basic axiom. Rather than a metaphysical statement, this is a simple deductive statement which must be taken as true, if one accepts that everything we know about the universe is not wrong.
If everything we know isn’t wrong, then there is a limited processing capacity in the human nervous system, and hard limitations inherent in physics and biology, that prevent perfect knowledge. All the facts we have, from Uncertainty to the inability to perceive anything that hasn’t had its light reach us, prevent perfect knowledge. With cogent inductive support, we can use probabilistic reasoning to verify the strength of the premises in a proof. But since induction is not truth preserving, all premises supported by induction must be probabilistic, and are therefore precluded from being justified beliefs. When faced with disjunctions which cannot be winnowed down to actual truth – we have no way to know if 9/11 Troofers are correct, if global climate change is a massive hoax, if evolution is just an atheistic conspiracy, etc… Without knowing if something is even ontologically possible, let alone true, no objective truth claim can be reasonably held.
To deny this is to go beyond the limits of both inductive and deductive logic.