calhais: How can you tell when you've presupposed something and then gathered--not `built'--evidence about it, and how can you tell when you've gone the other way around?

If you cannot tell when you have built/gathered evidence to support your presuppositions, then perhaps you should go to college and take a few classes in the sciences. Better yet, take a class that focuses only on the Scientific Method.

rmfr

The limits of the scientific method is that it cannot demonstrate the existence of things that have no effect.

Atheists, what are the epistemological limitations of the scientific method? What are its strengths? How do you know, and how do you know that your argument is reliable?

To the extent one speaks of the real world one cannot be certain; to the extent that one can be certain one cannot speak of the real world! (Here, I assume that our collective senses do reliably detect an objective reality of atoms and energy outside of ourselves even though we may often misinterpret that data. If that is not the case, then we have nothing left to talk about!)

In chess one may state with certainty that a player with only a king and a rook, and the move, can with accurate play checkmate an opponent who only has a lone king. Certainty follows from the rules that define chess. A simple method that constrains the possibilities does the job. In systems of pure logic, chess being an example, one begins with axioms that cannot be questioned. They cannot be questioned because, if coherent and non-contradictive, they define the system. As givens, they are neither true nor false. If you play chess you move the pieces one way; if you play checkers you move your men another way. It's not a question of one way being correct.

"Flat," Euclidian geometry, for example, and the geometries of curved space have variations of a key axiom. It is not a case of one geometry "really" being right. They are all valid geometries whether they apply to your room or not.

Thus, certainty can be had in systems of pure logic. The drawback is that these systems have no connection with the real world. If their axioms are neither true nor false, then they say nothing about the real world. There is no reference in such systems to the real word. If a version of mathematics proves useful it is because it was chosen to be useful. That is, we looked at the real world first then developed a math that proved useful. It may be that reality must follow, at least to a good approximation, certain geometries and other mathematical ideas. But how could we conclude that without actually looking at reality?

If axioms are chosen to model the real world then, of course, they would normally have true/false values. However, certainty is lost because of the problem of determining whether those axioms are perfectly matched to the real world. I am always amused by people like William Lane Craig who go to great lengths to present an impeccably valid argument (i.e., there are no errors in the logic itself) when the real argument centers about the axioms. Such formalism seems like an unnecessary distraction.

In the real world (defined above) our analysis must begin with our collective sensory data, our only demonstrative input, which must be a shared data if an observation is to be part of the community experience. We must have some shared idea about heavy rocks, the hardness of iron, rain, and the warmth of the sun in order to meaningfully speak about such things. Private experiences cannot be seen or experienced by the group.

Public knowledge then, rather than private knowledge, is the foundation that gives meaning to the idea that a statement is universally true. Truth rises above a private experience and can be shared. One might be able to convince the community that one is receiving messages from God, but it's a matter of public trust and not public knowledge. Simple trust is a poor foundation for evaluating the truthfulness of statements once one discovers that different prophets contradict one another!

As we integrate our sensory data about the real world we begin to construct models (interpretations of that data). Babies begin to model the space around them along with their concept of time. They learn what perspective is. They learn how things work, including the rules of their own tribe and family. (Some of that knowledge may have a genetic component.) That is, they form useful models that have some relevance to the real world and will, therefore, advance their chances for survival. Cultural evolution commonly adds communal stories about how the world began, about gods that send rain and lightening, and other such matters.

The difference between a "story" and a "model" is that the former is an entertaining narrative that usually explains how something came to be. A story is not particularly concerned with documentation. A model is a kind of physical picture of reality that attempts to explain some of its features. We enjoy fanciful stories about how a god or goddess created the sky. We may also be fascinated by the Babylonian model of the sky as a solid dome. Science sticks with numerical models because they are subject to exacting tests.

The great contribution of science is to confine itself to ideas that yield specific, testable predictions (falsifiable claims) and, therefore, claims that can be tested. Along the way the practical machinery for achieving that goal was refined to an exquisite level. It's loosely called the "scientific method" although it is something of a patchwork of good ideas rather than a rigid formula.

Thus, our elaborate system whereby potentially useful scientific work is shared widely in refereed journals, keeping scientists on the cutting edge and encouraging additional work to independently test those ideas. This is not a circle-the-wagons society! You get your Nobel Prize by overturning orthodoxy.

Hence, science is more than a philosophical debating society. It has avoided the stagnation of theology, and a general consensus is not only possible but virtually certain once data and analysis pass a critical threshold. True, there is a lot of rear guard action when an orthodoxy begins to fail, but eventually those old guys die out in favor of a new generation more open to the evidence. In all fairness, we should note that orthodoxy has in some sense earned its stripes and is often supported by some solid arguments. When plate tectonics was first proposed, it went nowhere because geologists were understandably hung up on the idea that moving a continent over solid rock was flatly impossible. They never looked further to the remarkable fossil and rock formation data that strongly pointed to just such a movement. Only later, in the light of extraordinary evidence did the paradigm change.

An idea is tested when its major predictions are checked against reality. Most major ideas are not perfectly conceived at their start, so allowances must be made concerning contradictions with reality. Will reasonable changes in the model skirt the problems or have too many bandages already been applied? There is an understandable reluctance to throw away a model that has enjoyed a great deal of success. Perhaps it's close to the truth and just needs some tweaking. It ultimately comes down to mature, scientific judgment. There is no magic formula that tells us when to change paradigms.

Deductive reasoning (the machinery that takes you from the model to its logical predictions) is great for testing a model and bringing out unsuspected relationships. If you apply deductive reasoning you can learn a lot of interesting facts about parabolas and ellipses, and a lot of other things. But that knowledge was only hidden in the axioms; it's not really virgin knowledge.

To understand the real world means to leave the certainty of deductive reasoning and rely on inductive reasoning. We begin with a pile of collected facts on which there is no serious disagreement. We try to explain those facts by constructing models. Models go hand in hand with mathematical calculations, and that would be an impossible task if those models attempted to consider all the messy, real-world details. Thus, a model is usually the core concept, the elegant essentials, the really important ideas that should allow a meaningful test of the hypothesis. A model gains a great deal of credibility when it goes beyond the facts that gave rise to it and predicts some really cool stuff, predictions that check out and allow further investigation and insight.

Since an infinite number of logical explanations (models) apply to any collection of facts, 100% certainty cannot be had. How does one rule out, in practice, an infinite number of logically possible alternatives? The nice thing is that we don't need certainty! Our airplanes fly quite well based on conclusions that are "merely" highly credible. Credibility, then, is the currency of statements about the real world--not certainty. Therefore, every model will have loopholes. Those loopholes may be microscopic or huge, farfetched or serious. Scientists deal with a reality that generally supports only a limited number of credible models, allowing them to bypass the infinite number of models that are totally farfetched. Certainty is thus lost, but meaningful knowledge is thereby gained, knowledge measured in terms of credibility.

The spherical shape of the earth is not 100% certain, even as a good approximation, because we can't rule out the logical possibility of a mass hallucination. However, the lack of 100% certainty doesn't make all models equally good. By any reasonable standard that we can come up with, the spherical shape of the earth warrants a huge confidence rating.

Thus, we come to the crux of the matter. The chief epistemological limitation of the scientific method is a limitation shared by any attempt to probe the real world. You have no demonstrable claim for certainty. The chief strength of the scientific method is that it has refined the intellectual (and physical!) tools to maximize a credible investigation of the real world. And, we don't have to take someone's word for it. We have landed men on the moon and probed Pluto, explored the inner workings of atoms (discovering some fundamental limitations in our knowledge!), and have even come to understand a great deal about our universe which can now be traced back to a Big Bang event some 13 billion years ago. All of these findings fit together into a tapestry of interlocking, known facts even as they open vistas to new frontiers. That adds a large degree of credibility to them.

In short, the scientific method is the best approach by far (with demonstrable results) for learning about the real world. The currency of the real world is the degree of credibility, not certainty. When it's high enough we tend to call something a fact. It's a fact that the sun will rise tomorrow. Evolution is a fact of life. When scientists admit to a lack of certainty they are just being good philosophers. They are not admitting to a defect that makes the scientific method inferior to others.

A high confidence level means that a model fits nicely into the greater, woven fabric of accepted (tested), interlocking knowledge, that it accounts not only for known data but predicts unexpected results that are confirmed, that it opens new vistas to meaningful exploration, and that it suffers no serious contradictions with observation. A high confidence level (interchangeable with "credibility") in a model means that within its framework lots of facts fit together that would otherwise have no good explanation. A high confidence level means that a model has no serious competition, and that there is no immediate prospect of such. (In a newly evolving science we might expect big surprises at any moment.) The test is not whether we can be lawyers and poke some logical holes in an hypothesis. There will always be loopholes to any hypothesis about the real world.

CONCLUSION

Atheists, what are the epistemological limitations of the scientific method? What are its strengths? How do you know, and how do you know that your argument is reliable?

The scientific method is restricted to investigating only falsifiable ideas about the real world. Therefore, science says nothing about gods and goddesses and other non-falsifiable ideas including a supernatural world. While this may seem to be a limitation it is actually a major strength. Non-falsifiable ideas about the real world are beyond any meaningful kind of verification. There is no point in arguing whether they are true or not since any such argument is no more resolvable than the question of how many angels can fit on the head of a pin.

Short of total nihilism, which is in the awkward position of claiming to know something about what we can know, we may reasonably assign confidence levels to models. When a model's credibility is high enough it is reasonable to say that we know something. However, we don't claim certainty as we must be open to new evidence or reasoning. Truth, then, is provisional but it does work for us. The reliability of scientific models is just another word for our confidence levels assigned to those models. Again, certainty is not in the cards but neither are we clueless as indicated by the ways a model gains credibility. Not being certain does not make all claims equal. With enough confidence in a model, we may provisionally treat it as reliable knowledge in the same sense that a Rolls Royce is a reliable car. That makes a lot more sense than pretending that we don't really know anything about the real world.

When Jesus is deciding who to burn, I doubt somehow that he'll be interested in epistemological arguments.