Jeffrey at Issues In Mormon Doctrine has had a series of good posts on Dennet's "Breaking the Spell of Religion". While most of my comments never quite manage to be in sync with the subtleties required, the novel threads I end up thinking about after the face must mean I am at least adjusting my track, or perhaps just getting successively more off base.
One issue that has taken me a while to come to grips with is the various things people mean when they say something is scientific. Many branches of study seem to cling to the coat tails of science in a way that is analogous to how religions cling to divine endorsement. Often it is used as a tool of authentication rather than a means of establishing rigorous absolutes.
Going through Voltaire's Bastards, one certainly gets presented with the idea that much of what we assume to be valid may in fact just be the finalizing of a self justifying and hence, logically consistent system. However, one assumes science is independent of this for a few reasons.
1. Science is weakly dependent on value judgments. Facts are able, in most part, to speak for themselves. If not, they can be explained in a way that minimizes subjective interpretation.
2. Scientific results are context independent. They specify the conditions under which they are valid. Eventually conclusions can be shown to be correct in any attainable situation.
3. As an extension of 2, context independence infers that reality is what is being described. Repeatability solidifies this assumption.
In the scientific pursuit one of the things sought are unimpeachable results. These should apply to any specified conditions, or at least be simplified approaches that apply to limited domains (Newtonian mechanics vs. quantum mechanics). They should be repeatable, hence reliable. In other words they should describe reality. Following this line, one can come to the conclusion that reality is anything that can be depended upon and is context independent. More specifically many people would say that it corresponds to something absolute. Science is a means to describe this absolute reality.
An absolute, science communicates in communal fashion. It's results are thing upon which every logical person can agree. Those that don't are factually and demonstrably incorrect. This means it is based on communal realities. Now of course it happens that shared realities are the ones most likely to represent physically reality. However our experience of reality can't occur without participation. While philosophers will probably call me a fool, I don't think we can logically talk about reality as an absolute without reference to our experiential contact with it. In other words, it is like trying to talk about the momentum of an electron without reference to it's position. While precise momentum measures are possible, we can't take this figure and extrapolate it back to a given position. Their distinguishabilities are inversely related. When we talk about science discovering absolutes, we can't talk about it independent of experience. Of course degree to which experience interacts with reality varies depending upon the object and, perhaps, the observer.
If I am a schizophrenic and hear voices of an imaginary person communicating with me, in what way is this less real than another person who has an actual person talking to them? It is less real because no one else can experience it. While this obviously isn't a very useful way of dealing with things, what happens if other people were to act as if the schizophrenic voice was real? Obviously we would say we are entering into a world of pretend. It falls apart because there is no way different people could experience that same thing. Each "voice" would be different, hence unreliable. If, however, people understood the imaginary voice as communicating the same thing in a repeatable fashion, is it still unreal? As the voice becomes more anthropomorphized, the answer mimics reality more and more. Questions like this really lead one to question the value in the difference between actual realities and functional realities. If they are indistinguishable, what is the difference? To some extent, is this what religion may actually be proposing?
One of the first problems is that we can no longer deal with one reality that we all experience. There are individual realities that have substantial overlap, especially in certain areas. In this realm, religion has a niche. The unification it facilitates may enable its fundamental promises. It implies that what one accepts, to some extent, is able to change the reality one experiences. Fundamentally this is what repentance is about. Real change is feasible, and paradigm shifts can enable things that otherwise would not be possible.
To me this leads to an interesting way to think about the three degrees of glory. Somehow people in one degrees are able to share a reality that is fundamentally different from those in another. It could be a similar idea of how god works and how to follow his example. In this way it would be analogous to Christ's proclamation that if you have seen him and what he has done, you have seen the Father and what he has done. While Jesus and the Father are not physically one, it's a good bet they have a similar take on how things work. I don't think it a religious stretch to say that their reality is quite a bit different that what many of us would accept, conceive, predict, or apply.
Similarly there are certain other shared realities that are possible. Some may involve no spiritual promptings at all. In this regard I find the cognitive sci approaches to religion interesting. They seem to imply that religious inclination is at least fairly widespread. The conclusion that the supernatural, while not actually real, is nonetheless an effective way to embody ethereal connections seems fairly analogous to the mormon conception of the telestial kingdom. Distinctions in the other kingdoms could also be based on the way one perceives, shares and experiences reality.