One can argue that a central purpose of knowledge is prediction: Francis Bacon famously said "nature, to be commanded, must be obeyed." We can briefly illustrate the point. First and most evidently, we benefit from the ability to understand the mechanisms of the natural world well enough to predict their behavior. Prediction of seasons, tides, and other such phenomena played an important role in the development of civilization. Second, invention and design require the ability to predict the mechanical, electrical, or other properties of materials to ensure that a device will consistently perform as desired. Knowledge thus has, at minimum, a practical anthropic function through its role in prediction.
A prediction is a statement that has at its core a claim of a future fact. Usually, though not always, the statement also includes conditions that must be met for the future fact to obtain. Thus, "The sun will rise tomorrow" and "If our team loses tomorrow, they will be out of the playoffs" are examples of the form. A statement need not be true or correct to constitute a prediction. However, it is important that both the conditions and the future fact be possible to verify or falsify (for readability in the remainder of this discussion, verification shall refer to both positive and negative verification). If the statement is not possible to verify, it is better described as fiction or fantasy. This is not a conclusion but rather an analytic posit:
A statement predicts only if it can be verified.
Under what circumstances might a statement be impossible to verify? One important case is when the terms in the statement have no reasonably well-defined referent. It is difficult to draw a bright line for this requirement, but statements like "when my qi is low I will perform poorly" undoubtedly qualify. More arguable would be statements concerning qualia or events outside the relativistic light cone. The point is one of burden: one who purports to make a prediction is responsible for at least outlining how it is possible that it could be verified.
Let us now make a stronger and more interesting claim:
A statement reliably predicts only if it can be repeatedly verified.
To the original posit we have added the terms "reliably" and "repeatedly," but there is also a more subtle addition. The consequent holds not only if the statement itself can be repeatedly verified, but also if it can be repeatedly verified a fortiori via a more general statement from which it is deduced. Thus from "the sun rises every day," which can be repeatedly verified, we can deduce "the sun will rise tomorrow," thus meeting the requirement.
This claim does not speak to the sufficient conditions for a reliable prediction: there are many. It only claims that repeatable verification is necessary. The reason is simple: if the stated future fact, or the future fact in the applicable general case, will only happen once, then the basis of the prediction is necessarily underdetermined. Even if we have hypothesized a mechanism and have some empirical support for it, an abductive leap was required.
One might argue that it is possible to empirically rule out all other possible mechanisms without having tested the hypothesis at issue, and thereby rely on it in a single instance. While this might very well improve our confidence, particularly in situations where it is necessary to make such a prediction and take action accordingly, this does not make the prediction reliable. Until and unless we achieve a complete and analytically solvable model of all reality, ruling out a finite set of mechanisms leaves an indeterminate remainder.
Our claim has the further benefit of being itself repeatedly verifiable. Indeed, if we look at scientific experiments as well as the world in general, it is quite rare for predictions about singleton events to be accurate. Thus, unlike Logical Positivism, our statement of Predictive Positivism at worst is not self-refuting.
We can refer to statements that do not predict as "non-predictive," or pejoratively following Wolfgang Pauli as "not even wrong." We can refer to statements that do not reliably predict as "unreliable" or perhaps "speculative." What we cannot say is that that such statements are "meaningless," as was hoped by the Logical Positivists. Indeed such statements may be quite meaningful to the speaker. The term "meaningful" has a subjective, qualitative element that makes its verification essentially impossible. Our formulation, while not eliminating all possible sources of disagreement, avoids the subjective element yet solidifies the importance of empirical grounding.
A brief addendum arising out of subsequent discussion: the terminology used in this post strongly suggests a binary view of verification, i.e., verify (positive) or falsify (negative). This is too limiting and should really be expressed in more directional terms, e.g., supported or impugned. It does not seem that this point affects the overall conclusions here, but we should deifnitely interpret verification as a continuum of evidence ranging from strongly negative to strongly positive and everything in-between. (the case of an entirely neutral result seems at first problematic, but really it is just empirical confirmation as to the uncertainty of the statement).
ReplyDeleteSo, infinity valued logic, supported +, impugned -. Rather like "karma" and "dharma" :) I like your explanation of what I view the scientific method. Mainly, is it useful in the mechanistic, objective, universe, or useless. Metaphysics I think, deals with personal explanations of "Why" and so far as I know, is only predictive on a personal level, utilizing confirmation bias i'd guess.
ReplyDeleteThanks for the comment - I imagine, though don't know, that people have invented logical forms that incorporate continuous versus binary values; presumably each has its place. Yes, I am describing (part of) the scientific method, but in particular aiming to justify limits on its jurisdiction.
DeleteAnd yes, I would say that metaphysics might very well have a personal, subjective value to individuals, and sure, they might use it to explain phenomena as a way of reducing cognitive or emotional dissonance. It is wrong to say that it has no meaning because it might have a great deal of meaning to the individual. It is just not useful for successful prediction.
Another addendum relating to use of the term "metaphysical." It has some problematic connotations, arising from the era of Logical Positivism, as an epithet. Worse, it suggests that claims asserted as metaphysical are somehow inherently metaphysical. This is true in cases where the terms are not defined, but it is not uncommon for claims to have reasonably well-defined terms but simply not to make any verifiable claims, or predictions, that distinguish them from their competitors.
ReplyDeleteClaims of this sort may not be reliably verifiable only in the present state of knowledge. Thus "metaphysical" is really intended as an epistemic term. It means that with present knowledge, there is no known way to verify it, i.e., to make repeatedly testable predictions that demonstrate its superiority to alternative explanations.
There is, presumably, no way to verify or test a prediction that a claim can never be verified, as long as its terms are well-defined. Thus once again the burden of proof is on one who makes a claim, to provide an approach to verification. An opponent (not one who claims an alternative, equally unverifiable theory, but rather one who sees it as unverifiable) need offer only the barest of explanations as to why present knowledge does not enable verification, because it not really his responsibility.