Thinking Christian Comments
Original Post: Does God Explain Anything At All? (Part 1)
Tom,
I don't demand that the things being explained be material.
For example, if I note several examples of pairs of odd numbers with even sums, my explanation can be found in a theorem that says that the sum of any two odd numbers is even. None of this is material, and yet I can have predictive explanations.
Another example: suppose I see several acts that I feel to be evil, and in each case, I notice that the "evil-doers" always get their comeuppance (e.g., murderers always get bitten by rattlesnakes, felled by grand pianos, etc). I might then propose that there is a predictive law that "evil-doers always get their comeuppance." Yet, moral attributes of good and evil need not be material, and I don't think the law of universal justice would necessarily be material either.
So when I say "measure" or "observe", I really mean "experience" in general.
Q: Can something unpredictable be considered an explanation for some event?
Just to be sure I didn't miss something, dl--I don't think I wrote that the thing being explained needed to be material. I wrote that the proposed explanation, and the proposed experiments (or some such) to confirm it, needed to be measureable and material. So, with this comment are you correcting something I wrote, or are you extending what I wrote?
I want to make sure I haven't misunderstood.
Tom, I applaud you on your approach to this topic, it's a classic conflict resolution move, but one that is difficult for an opponent to take (which is why mediators are so helpful). It's a very hard thing to set your opinions aside (albeit temporarily) and try to understand your opponent. But that is the only way to truly learn.
SteveK,
Q: Can something unpredictable be considered an explanation for some event?It depends whether you mean individually unpredictable or statistically unpredictable. The former is acceptable, the latter is not.
A statistical prediction is still a prediction. If your theory predicts a probability distribution (even a flat one), that is still a prediction, and still qualifies the theory as a candidate explanation. IOW, to be explanatory, it is not necessary to predict individual chance events, as long as you can at least predict a statistical distribution.
Tom,
I don't think I wrote that the thing being explained needed to be material. I wrote that the proposed explanation, and the proposed experiments (or some such) to confirm it, needed to be measureable and material. So, with this comment are you correcting something I wrote, or are you extending what I wrote?I'm not sure I follow. Are you saying that any controlled experiment implies materiality, even if what is being measured is non-physical?
How about prayer studies? Your claim here suggests that controlled demonstrations of the effectiveness of prayer might render God material.
No, I'm not saying that.
(One reason, among others: a controlled study of prayer is strictly impossible, for reasons we've discussed before. God is not a blind participant in the study. Human psychological studies often conceal the true intent of the study until it's over. You can't do that with God. And you can't control the other variables, like the pray-ers' motives, which count in answered prayer; the possibility of other people praying, especially in medically-related prayer studies; the pray-ers' beliefs about God; and so on.
That's old territory that I didn't really want to open up again except as a brief outline for the sake of those who did not see it the first time around.)
Anyway, if you're not saying that measurable should be equated with material, then I don't want to put those words in your mouth. I'm trying to get this into a form that says what you would want it to say. How about this formulation:
- A) An explanation must be more than a restatement of observations
- B) An explanation must have substance beyond mere words
- C) Vague references to God do not qualify under (B)
- D) Predictiveness does qualify, and in fact is the only thing that does qualify
- E) Predictiveness is understood to be in terms of measurable results (possibly, correlating to a measurable explanans)
Does that cover it? I added a clause to (D) this time; it was in the general text of the blog post but I had forgotten to include it in the summation. Would you add anything, or change anything? And is a measurable explanans required?
Thanks for working with me on this.
Tom,
I think B is a little too vague. It could be interpreted in a lot of ways. (For example, the Jabberwocky has substance beyond mere words, but not in the fashion I think is relevant here.)
Perhaps we can say that B should be something like this:
B) An explanation must be more than a reference to an explanation that we don't yet have, but hope to have in the future.
I think (E) is a bit of a problem as stated because it looks like it comes out of nowhere. (E) doesn't come out of nowhere. I think that if one accepted (A)-(D), one would be led to discoveries in logic, Bayesian statistics and human nature, and these discoveries would tell us that one doesn't really have a prediction if there's no possible way to make the predicted observations.
Prayer is a good example of this. We agree that prayers are not answered every time. In that case, the claim that prayer is effective is a statistical prediction. However, if one then argues that prayer experiments are impossible to control, then the statistical distribution is impossible to see. Wouldn't you agree that predictions that prayers will be answered are empty if they are both statistical and impossible to see statistically?
So perhaps (E) should be:
E) Prediction has the implication that some future observations will raise your confidence in the explanation, and other future observations will lower your confidence in the application.
I think this says that a statistical prediction has to be visible, in principle, with a statistical study. Throw in the fact that humans are only statistically reasonable and competent, and we're led to the conclusion that every prediction has to be supported with statistical evidence. (Logic and Mathematics are certain in the sense that their statistical uncertainty can be lowered to arbitrarily low levels, so there's no conflict with the conclusion that statistics should be applied everywhere).
DL:
A statistical prediction is still a prediction. If your theory predicts a probability distribution (even a flat one), that is still a prediction, and still qualifies the theory as a candidate explanation. IOW, to be explanatory, it is not necessary to predict individual chance events, as long as you can at least predict a statistical distribution.
I'm a little confused about this. Seems like there is a problem with circular referencing.
a) To get statistical data and form a predictive model you must observe events that have not been explained yet.
b) To explain events you predict it will fit the statistical data.
In a way it sounds like you are saying "I predict that what I see can be explained in terms of what I've seen before". That sounds like a restatement of observations, and you already said that isn't valid.
Please clear this up.
DL:
When we have competing explanations, which one(s) are acceptable, and why/why not? Here's an example:
a) Tom, putting the pot of water on the stove, explains the boiling water.
b) The desire to cook pasta explains the boiling water.
c) The chemical reactions in the brain explain the boiling water.
d) The heat of the flame explains the boiling water.
e) The temperature of the water explains the boiling water.
f) The laws of physics and chemistry explain the boiling water.
g) The chemical reaction expressed as [insert combustion reaction] explains the boiling water.
h) Tom's wife asking him to boil some water explains the boiling water.
i) God asking Tom to boil some water explains the boiling water.
Good questions, Steve. Thanks.
SteveK,
In a way it sounds like you are saying "I predict that what I see can be explained in terms of what I've seen before". That sounds like a restatement of observations, and you already said that isn't valid.I think the confusion arises because our minds often explain data implicitly.
Suppose I show you a graph of data points that seem to form a perfect straight line. If I propose a straight line "law" to explain the data points, you might be tempted to say that I'm just restating the observation that the data points lie on a straight line. However, that's your mind being a good theorist and instantly proposing a straight line explanation without any formalities.
In the restatement case we have:
My data points fall on the straight line y=ax+b.
In the explanation case we have:
My data points fall on the straight line y=ax+b, AND I predict that future measurements will lie on this same line (to within some level of precision).
The difference is critical.
Here's an example.
Suppose that you and I measure the amount of time it takes for a particular ball to fall from the top of the Tower of Pisa to the ground below. In four trials I get 3.3 seconds, 3.4 seconds, 3.3 seconds and 3.2 seconds. We take turns, and so you have also done trials, but have not yet told me your results.
Based on my data, I hypothesize that there is an explanatory "law" that the ball always takes the same amount of time to fall, and that this time is 3.3 +/- 0.1 seconds.
This is perhaps the simplest of cases, and the type that most resembles restatement. And yet, this is not a restatement of the data. It is not simply the claim that it took an average of 3.3 seconds during my trial (which would be a mere restatement). Rather, it predicts that redoing the experiment will get me the same results. It predicts that your trials will be in line with my own.
Computing statistical metrics is not explanation, but predicting that a statistical metric will hold in the future is.
Does that answer the question?
SteveK,
Explanations are what they predict. I am suggesting that when you specify your predictions in detail, you are placing your explanation in a canonical form. Two explanations that have the same canonical form are the same explanation.
You are citing explanations with the suggestion that they are all explaining the same thing. They are not.
a) Tom, putting the pot of water on the stove, explains the boiling water.
When this is used in common English, what is assumed? Stoves are for heating things, and are used by intelligent agents who can't heat things simply by wishing them heated. Tom is a human, and humans have the ability to place pots on stoves. Humans often do this to meet common goals, e.g., in food preparation.
IOW, this is an explanation of who boiled this particular pot of water, not the boiling of water in general. As an answer to this question, it seems quite predictive. We can go and ask Tom whether he placed the pot on the stove, and use established statistical facts about how people answer such questions to arrive at the conclusion he will say yes if he placed the pot there. We may find witnesses or fingerprints. It may be that Tom places a characteristic amount of water in the pot. Or perhaps he is slicing some broccoli on a nearby chopping board.
b) The desire to cook pasta explains the boiling water.
This is about how "desire to cook pasta" correlates with "boiling water." Sure, that's predictive, although it's not answering the same question as the explanation above. It's not about boiling water in general, but about correlations of boiling water with pasta cookery.
c) The chemical reactions in the brain explain the boiling water.
This answers a question not about water boiling in general, but about cases in which water boiling is initiated by intelligent agents with brains.
Predicts that neutralizing those chemical reactions in human brains stops them from boiling water.
d) The heat of the flame explains the boiling water.
It explains water boiling when in (mediated) contact with flames, not water boiling in general, or water boiling for pasta cookery.
That's very predictive. Remove the flame and the water should stop boiling.
e) The temperature of the water explains the boiling water.
This is about water boiling generally (at a fixed pressure).
Predicts that water at different temperatures doesn't boil.
f) The laws of physics and chemistry explain the boiling water.
This is troublesome. It works if you specify what those laws are. If it says the same thing as explanation (e), that's fine. But if you don't know what they are, you have no specific predictions and no explanation.
g) The chemical reaction expressed as [insert combustion reaction] explains the boiling water.
This is not about water boiling in general, nor about flames in general, but about flames caused by one particular reaction.
Predictive because you can suppress the reaction, and that will stop the water boiling.
h) Tom's wife asking him to boil some water explains the boiling water.
This is intended to explain a very narrow set of water boiling conditions.
Predictive. Correlate Tom's wife making the request and Tom fulfilling it. Secretly ask Tom's wife to make the request at selected times, and see if Tom complies. Control for Tom boiling water for his own purposes.
i) God asking Tom to boil some water explains the boiling water.
This is also intended to explain a very narrow set of water boiling conditions.
Predictive. Correlate God making the request and Tom fulfilling it. Secretly ask God to make the request at selected times, and see if Tom complies. Control for Tom boiling water for his own purposes.
If you cannot see God, cannot see him making requests, and cannot make requests of God, then there's no prediction that I can see.
Alternatively, if you tack on the addendum that you can never control for what would happen in the absence of the explanation, you're implicitly saying that your predictions non-existent.
In summary, each of these explanations explains something different. It's not the case that they all explain the same thing just because they all relate to "boiling water". Their explanatory power relates to their predictive power.
DL:
In the explanation case we have:
My data points fall on the straight line y=ax+b, AND I predict that future measurements will lie on this same line (to within some level of precision).
The difference is critical.
Q1: Given your comment, is this a fair statement now? If not, please tweak it as necessary.
"I predict that what I see can be explained in terms of what I've seen before, to some level of precision."
Q2: How do you quantify the level of precision (tolerance) if not from the observed data? Once again, it seems your entire prediction is based on what you've already seen.
DL:
This question has to do with the data used to form the predictive model.
A group of 50 scientists observe an athlete running the 100 meter dash. The athlete is observed running the race dozens and dozens of times and so they have formed a predictive model to determine his running time. It is 10.0sec +/- 1.5sec
One weekend, ten of the scientists come to observe the athlete again. During their observation the athlete runs with a time of 4.2sec and the scientists are amazed. They attempt to repeat the time, but the athlete never comes close. All future runs fall within the original predictive model.
Q: Do we include the 4.2sec run in the predictive model? Why or why not?
Secondary to that question (which I think will help us a lot), what kinds or categories of things would your model consider as explanations of this hypothetical event, doctor(logic)?
Also, doctor(logic), should we gather from your 1:46 am comment that "predictive" always implies some sort of correlation? Or can "predictive" have other senses besides?
SteveK,
Q1: Given your comment, is this a fair statement now? If not, please tweak it as necessary.I'm not sure what you mean by this question. Can you rephrase it?
"I predict that what I see can be explained in terms of what I've seen before, to some level of precision."
Let's use the graph analogy. I have some data points on a graph. An explanation consists of a curve passing through those data points, and such a curve always interpolates and extrapolates.
However, I cannot infer which curves explain the data points without using inference from the data points.
So, yes, it's "based on" what I've already seen, but not equivalent to what I have already seen. Indeed, the curve could be something that has never been seen before. So it would be misleading to say that the explanation is a restatement or "only" what has been seen before. It contains predictions so it cannot be only what has been seen before.
A good example is General Relativity (GR). GR was invented based on special relativity, which was in turn based upon explanations and observations of electromagnetism and Newtonian mechanics. Several interesting GR solutions were predicted, including black holes. Black holes had certainly not been observed before GR. So, yes, GR was based on what we had seen before, but it predicted things never seen before.
Q2: How do you quantify the level of precision (tolerance) if not from the observed data? Once again, it seems your entire prediction is based on what you've already seen.Uncertainty and precision may appear because our senses or instruments are fallible, or because the effect we're looking at is not acting alone. For example, in the case of the ball dropping from the tower, we might anticipate that winds, humidity or other factors affect the rate of descent. Since we're not measuring these factors in our experiment, we're not controlling for them, so we would have to assume that these other factors are relatively small. That is, our explanation says "assuming air friction and moisture are negligible, the time to fall is 3.3 seconds +/- 0.1 seconds, where the uncertainty relates to all the factors we're not measuring, and to the limitations of our instruments."
We are assuming that the uncontrolled environmental factors and user and instrument errors are pretty much random, so they fall into a Gaussian distribution around the mean. The +/- is a standard deviation, which is a measure of how wide the distribution is.
The more experiments we do, the better we understand the distribution, and the smaller we can make our uncertainty.
So, yes, the uncertainty for an explanation is something we infer from past experience.
SteveK,
One weekend, ten of the scientists come to observe the athlete again. During their observation the athlete runs with a time of 4.2sec and the scientists are amazed. They attempt to repeat the time, but the athlete never comes close. All future runs fall within the original predictive model.If you have a predictive model of this event, sure.
Q: Do we include the 4.2sec run in the predictive model? Why or why not?
However, it's difficult to cook up an explanation for a one-time event if you weren't collecting any other data.
Here's an explanatory theory. Someone tampered with all of the stopwatches that day. It's incredibly unlikely. However, from what we know about human runners, a 4.2 second run is even less likely. This theory predicts that examination of the stopwatches will locate the defect or at least a way to tamper with them. It predicts that you will find a person or persons who had means, motive and opportunity.
Alternative theory: Perhaps a cocktail of drugs and steroids can cause a man to run more than twice as fast as the fastest human. Predicts that the runner will have drugs in his system. Predicts that drugs may be found (or may have already been found) that can do this.
Suppose we rule out all of the theories we can come up with. In that case, the 4.2 second run remains unexplained.
You can include the statistic in the runner's statistics if you like, and his average speed may be stellar for a short while, but eventually, the average will be diluted by the statistics of his mediocre (by Olympic standards!) running.
Tom,
Secondary to that question (which I think will help us a lot), what kinds or categories of things would your model consider as explanations of this hypothetical event, doctor(logic)?Anything properly predictive, i.e., predictive, and not artificially shielded from verification.
Suppose I sign a contract with you in which I promise to clean your windows. If a clause in the contract says that I cannot clean your windows on weekdays, and not on Saturdays or Sundays either, then I'm not really signing a contract to clean your windows, even if there's some pretense to doing so. It doesn't matter that the contract says your windows will be smudge-free after contract execution, the whole thing is a sham because there's no intent or mechanism to ever get to the clean-window state.
Likewise, there's no point in us claiming a predictive explanation if the predictions can never be verified by the rules of our explanation.
Tom,
Also, doctor(logic), should we gather from your 1:46 am comment that "predictive" always implies some sort of correlation? Or can "predictive" have other senses besides?Hmmm. This question is rather vague. Everything, even the meanings of words, implies some degree of correlation (e.g., I correlate the word Elephant with sensations and thoughts of elephants).
Suppose I observe a cup collects 3" of water every June, and I predict that it will do so next June. I have to correlate June, the cup, the water, and the year to confirm the prediction.
Suppose I observe odd numbers sum to even ones. I predict this will always be the case. I prove a theorem that shows this to be the case in general. Doesn't the proof correlate odd numbers and even sums in all cases?
What is the alternate sense of prediction you're thinking of?
Suppose we rule out all of the theories we can come up with. In that case, the 4.2 second run remains unexplained.
Suppose 10 of the 50 scientists formed their own predictive model such that the 4.2sec run met the conditions of predictability. One groups says the 4.2sec run can be explained by the data, the other says it remains unexplained. How does this get settled?
SteveK, isn't your last question a little disingenuous? Are you sure you can't imagine a valid process by which such conflicts get resolved?
More broadly, what is the point of all these questions asked of DL, who answers them? Let's get to the point, eh?
SteveK,
Suppose 10 of the 50 scientists formed their own predictive model such that the 4.2sec run met the conditions of predictability. One groups says the 4.2sec run can be explained by the data, the other says it remains unexplained. How does this get settled?It gets settled by verifying the predictions of their theory.
If you see a car crash, it isn't explained merely by the fact that you know that brake failure, intoxication, speeding, steering failure, etc are all explanatory. Knowing the explanation means deciding which explanation is most probable.
You generally need all sorts of evidence about that particular event to do this with high confidence. For example, you will be able to tell the driver's blood alcohol level after the crash, and thereby verify that intoxication was a cause (or part of the cause).
However, if you lack these details, you might still know something in light of past verifications. If you know that, historically, brake failure accounts for 1% of accidents and alcohol intoxication for 40%, you might justify a 40% confidence level that the accident was caused by alcohol intoxication. Of course, relies on the availability of statistics.
Paul:
More broadly, what is the point of all these questions asked of DL, who answers them? Let's get to the point, eh?
Tom has specifically set aside this as a Q&A so we can learn what exactly DL thinks. Didn't you read this?
Clarifying questions and comments are in order; criticisms and statements of agreement are not. We need to get to the point where we all agree on what we're talking about, before we actually start the conversation.
Tom,
Given Paul's "get to the point" comment, are we going to do this in another blog post or are we going to do that here?
It seems to me that the definition of "explanation" as used by DL that we are trying to reduce beyond ambiguity is tied to the scientific method of inquiry. DL's focused responses appear to be related to that method of observation, prediction, and explanation. If that is the case, it is not going to get us very far in a debate about whether God "explains" anything because the definition, at its core, excludes mechanisms that are usefull in discussing God.
Intangibles, such as love, hate, anger, etc. clearly exist and have real effect on humans. Yet, to try to use the scientific method to prove such intangibles, or as a basis of explanation for observations does not work very well.
Oops, you're right, SteveK, I momentarily forgot the different purpose of this thread.
DL:
It gets settled by verifying the predictions of their theory.
It was verified to within the expected limits of the model. In this hypothetical situation, one group of scientists interpreted the data prior to the 4.2sec run in such a way that the probability of a 4.2sec run was 1 in a billion. A 4.2sec run may occur again during observation, or it may never occur again - that's what "1 in a billion" means.
I'm trying to understand how this gets resolved. Tipping my hat to Paul, I can imagine any number of ways it gets resolved, but all of them require deviating from the original prediction model that was supposed give us the answer.
Getting to the point will be on the next post... which is going up now! Provided the next post is agreed to be a good statement of dl's position, of course.
Michael,
I responded to your comment on Tom's new thread.
SteveK,
You seem to interpret an uncertainty (+/-) as a prediction that something deviant will happen at a certain rate. That's not always the case.
In your original example, one will predict that a runner takes 10 seconds +/- 1.5 seconds. However, when we collect enough statistics, we'll see that the distribution is not Gaussian. If the distribution was Gaussian, we might expect to see runs of 9 seconds every dozen attempts or so, and maybe an 8 second run every hundred attempts (sorry, I'm not doing the detailed math here). We don't see those super fast races. So the prediction is that the runner will never run a 4.2 second race, and we have high confidence in that. It's not that we predict the runner will run a 4.2 second race once in a billion times.
If all the wasps you see are less that 3" long, will you predict that 1 in a trillion wasps is as big as a Cadillac? I don't think you would say that. You would be confident to 1 part in a trillion that you won't see a wasp that big, and that's not the same thing as predicting one.
Let's suppose that, somehow, the 10 running judges are predicting the 1 in a billion super-fast run without any reason for doing so. I suppose that, in principle, this rule could be explanatory (though perhaps not an inference). If we get more folks to take up running, we should see more super-fast times in the future.
If all the wasps you see are less that 3" long, will you predict that 1 in a trillion wasps is as big as a Cadillac? I don't think you would say that. You would be confident to 1 part in a trillion that you won't see a wasp that big, and that's not the same thing as predicting one.
You are using extremes and it's not necessary. All that's required is to step outside the prediction, no matter how small that step is.
Suppose I see a wasp that was 3.1" long? The prediction model failed and the wasp can't be explained - yet here it is sitting in my hand.
Question related to my last comment.
Q: Does you model explain the thing in your hand, or does the thing in your hand explain your model?
Actually the size of the step out of the predictive model is important. I have to agree with doctor(logic) on that. It depends on the shape of the probability distribution.
If every wasp observed in history was 3" long plus or minus 0.01 inches, then 3.1" would be a failure of a prediction model--or else a sign that something new must be accounted for in that model. But I doubt wasp sizes are so tightly distributed as that!
doctor(logic)'s predictive model has considerable strength for things like these. The question is not going to be whether it's wrong--for clearly it works for many purposes--but whether it's overly restrictive; or conversely, whether it's fully explanatory for everything that both stands in need of explanation and can be explained. That's where I'll be headed with it.
Well, Tom, I obviously haven't got this down yet. I'm working on it though and trying my best to understand.
Let me try this....
1) I have my first experience ever. No model therefore no explanation.
2) I experience it again in the same way and under the same circumstances and so I form a predictive model.
It seems to me I have a problem though. I'm attempting to use an unexplained event to explain another unexplained event. All future events are explained in terms of the first event, which to this day remains unexplained.
I'm obviously missing something, so help is appreciated.
Good point. There's a bootstrapping problem there, isn't there?
SteveK,
It seems to me I have a problem though. I'm attempting to use an unexplained event to explain another unexplained event. All future events are explained in terms of the first event, which to this day remains unexplained.Explanation is the recognition of a pattern. If you only have one data point, you can't see the pattern. The more points you see, the more the pattern emerges.
And emergence is exactly the right term. A single data point doesn't "contain" any explanation in the same way that a single geometric point doesn't "contain" any essence of the square or dodecahedron of which it is a part.
There's no bootstrapping problem because explanations are relationships between data points, and relationships between explanations.
DL:
A single data point doesn't "contain" any explanation in the same way that a single geometric point doesn't "contain" any essence of the square or dodecahedron of which it is a part.
Interesting theory, but one that doesn't seem correct to me. If each data point explains nothing then how does adding more data points help you? It's as if you are adding zero's together and somehow getting to the number one.
A more intuitive theory would be that each data point has a limited amount of explanatory power.
Switching gears a little, assuming your theory is correct, what does the prediction model tell us about the outcome? I'm talking about the significance, or meaning, of the outcome. If the model says we have an valid explanation, what meaning should we derive from that?
If each data point explains nothing then how does adding more data points help you? It's as if you are adding zero's together and somehow getting to the number one.A single brick doesn't make a wall, but many bricks do. A single point doesn't make a line, but many do. A single data point doesn't make a pattern, but many do.
It's the pattern that emerges, yet no single item in the pattern contains the whole pattern.
Paul:
It's the pattern that emerges, yet no single item in the pattern contains the whole pattern.
I agree with everything you wrote, Paul. I just think the conclusion may be wrong. Your conclusion is: Therefore each point, each brick, contributes nothing to the whole pattern - nothing.
Is there a prediction model we can use to help us?
Where did I say that each brick contributes nothing? You're the one who said that adding up zeros can't get you to one, not me.
How about this: A pattern is a *description* of data points that uses less information than the sum of the data points. So a brick contributes nothing to a description, but it does contribute one small part to the data points that are being described.
OK, Paul.
One thing that hampers our analysis is the English language. It's not optimized for this sort of thing. We tend to be sloppy about the following terms:
Justified explanation versus potentially explicable.
Inexplicable versus potentially explicable.
Verified explanation versus potential explanation.
A thing is explained (and explicable) when we have justified verification of one of its candidate explanations.
A thing is potentially explicable and unexplained when we can imagine one or more candidate explanations, but don't have significant verification of the candidates.
A single data point is potentially explicable and unexplained. However, it cannot be explained at all because you will need another data point to verify and gain confidence in even the simplest explanation of the original data point.
This one will help me with the perceived bootstrapping problem.
Q: Can a single unexplained event/experience have meaning? Why or why not?
SteveK,
What do you mean by "meaning" in this context?
A single unexplained data point is emotionally and factually significant. It also suggests that there's something we don't know (or else we would have expected the data point).
DL:
What do you mean by "meaning" in this context?
Does it tell us anything beyond "this is unexplained", or in the opposite situation "this is explained"?
SteveK,
I'll paraphrase your question to one that I find less ambiguous. Let me know if I'm missing your intended meaning.
Q: Can a single event for which I (presently) have no explanation tell me anything beyond "this is unexplained."
No.
Suppose that I could infer some other knowledge. In that case, I would be able to predict something else in light of the new knowledge I just inferred. But if I could predict something in light of my inference, I would have an explanation for the event. That contradicts the premise that I have no explanation for the event.
DL:
Suppose that I could infer some other knowledge. In that case, I would be able to predict something else in light of the new knowledge I just inferred. But if I could predict something in light of my inference, I would have an explanation for the event. That contradicts the premise that I have no explanation for the event.
Huh? Aren't you predicting something in light of the new knowledge gained (inferred) from all previous data points?
Commenting by HaloScan
