Do Dreams Predict the Future? A test! [Archive]

rsaavedra

7th October 2007, 03:04 PM

If I hit a deer in the next 24 hours, it will be strong evidence that dreams predict the future.

If I don't hit a deer in the next 24 hours, it will be evidence that dreams do not predict the future.
Very arbitrary to choose 24 hours. Who says an effectively predicting dream would have a specific time frame for the prediction?

A possibly hidden assumption of yours seems to be that *all* dreams necessarily predict the future. Maybe not all dreams fall under that category. Tricky thing would be to properly classify a dream as predicting vs. non-predicting, before checking whether the predicting outcome checks out.

One other thing: hitting a deer in the next 24 hours, or in the next 30 days, or whatever timeframe you choose, wouldn´t really be "strong" evidence that dreams predict the future. It will be just an increment of one in the counter A = "Successful predicting dreams". You should keep track of another counter B = "Failed predicting dreams." (But after what timeframe would you consider a dream failed if the outcome doesn´t check out? What if it checks out the day after your limiting time frame expires?)

After keeping track of many many people, and many many of their (predicting) dreams, then if the ratio of A / B is somehow significantly large (how large?), then *that* might represent strong evidence that dreams classified as "predicting ones" do predict the future.

JoeTheJuggler

7th October 2007, 03:48 PM

I just woke up from a vivid dream that I hit a deer with my car at night. In 24 hours I will report in this thread if this dream came true.

If I hit a deer in the next 24 hours, it will be strong evidence that dreams predict the future.

If I don't hit a deer in the next 24 hours, it will be evidence that dreams do not predict the future.

Near misses or odd occurrences, like hitting a person, will be counted as inconclusive.

The point of this is to demonstrate how I think metaphysical predictions should be tested.

Discuss...
First, I think you're on the right track in one way: reporting or writing down possibly precognitive dreams at least eliminates faulty memory, which I think is behind many instances of people claiming to have psychic dreams.

On the other hand, I don't think what you've described is a good way to test claims of metaphysical predictions. Whether or not you hit a deer or a person in the next 24 hours is not evidence of anything. An event like that could happen purely by chance, and right now there's no way to distinguish whether the dream was simply a coincidence or it was a glimpse at the future.

I think you'd first need to clarify what is being claimed. (Is it claimed that the dream will come true in 24 hours? If not, why that time limit?--as rsaavedra pointed out.)

Counting odd occurrences like hitting a person as a "hit" (i.e. prediction fulfilled) doesn't make any sense whatsoever--unless the claim is that the dream somehow means "an odd occurrence" will happen in the next 24 hours.

Once you've got the claim worked out, then you can begin to set up a test that should be able to provide evidence for or against the claim.

One of the biggest problems is sorting out what is merely a coincidence from what can't properly be explained that way. To do that, you'd ideally need to know what the probabilities of these predicted events are. (And a sample size of 1 isn't going to allow you to rule out coincidence with any level of confidence.) Just a little bit of shoehorning can make a seeming low probability event into a very high probability one.

For example, you might be able to calculate the probability of hitting a deer. If you include hitting a person, or hitting a dog, or hitting a squirrel or any other animal, or running into a billboard that has a picture of a deer on it, or hitting something that is "dear" to someone, or hitting a John Deere tractor, or nearly hitting a deer, or nearly hitting a person, etc. you will end up with a much higher probability. And "any odd occurrence" isn't as far-fetched as it might sound. I'm sure there are believers who could retrospectively interpret a dream of hitting a deer to fit whatever happens the next day: you accidentally caused the failure of a project at work, you got stuck in traffic because of someone else's accident; etc.

If the claimed ability isn't something that's repeatable (just a once-in-a-lifetime event), I'd discount it. By definition an event that isn't repeatable isn't testable.

As an aside, I hope you're not motivated to make this prediction come true since you're in a position to make it happen!

Madalch

7th October 2007, 04:05 PM

The idea that dreams (as opposed to daydreams) has always struck me as ridiculous.

I remember when I was about three or four, I woke up crying from a dream I had in which a skeleton was strangling Bruno Gerussi (from the Canadian show The Beachcombers). It never came true, and Bruno himself lived another twenty years after that.

When I was twelve, I dreamed that my sister and I both turned into some weird human-dimerotron(? the pre-dinosaur reptile with the huge sail thing on their backs) hybrid. That never came true.

In high school, I dreamed I had mucles like Conan (the Barbarian, not O'Brian), but instead of a sword, I had a magic tree branch which burned with a golden flame, but was never consumed. Never came true.

In college, I dreamed that my grandfather had come back (two months after his funeral), and that I was the only one who found this strange. He's still dead.

In university, I dreamed that I went to Germany to find my "real" grandfather's grave, accompanied by my then-girlfriend,who had somehow changed from Czech to Chinese. Helena never changed her ethnicity, and I never learned that my grandfather was anyone other than who I thought he was, and the closest he had ever been to Germany was riding in a bomber.

The other night, I dreamed I had an argument with my wife, threw a pot of tea at her, and walked alone on the streets of Amsterdam. I haven't fought with my wife lately, and Amsterdam's a long walk from here.

LostAngeles

7th October 2007, 06:36 PM

So, I'm the only person who voted yes on the second option. Guess I should explain why.

I dream in two categories: Freaking Surreal and Entirely Plausible. While the Entirely Plausible are the more annoying (since I have to take a moment to realize that none of that did happen), the Freaking Surreal ones that I remember often have an element affected by some concern I've had. This is either annoying or beneficial. Occasionally it helps me in sorting some stuff out, be it a homework problem or some cock up in the rest of my life. Sometimes just being able to pick out something that's been bugging me because it occurred in a dream is helpful.

And that's how my dreams are, "predictive.":p

karmicserenade

7th October 2007, 06:48 PM

all righty, I have a few dreams that have come true, one was just something that had occurred, I dreamt about the event, not knowing it had happened to a friend, it was exactly the way it had occurred too! Another was the rain, it was just raining in my dream and the next day it rained haha...that might not prove anything to the rest of you, but to me, it seems interesting to say the least.

Gord_in_Toronto

7th October 2007, 08:12 PM

Graham Ross

7th October 2007, 10:03 PM

If my dreams came true my waking life would be a lot more interesting.

Unalienable

8th October 2007, 03:18 AM

I have precognative dreams. Wait, let me restate that: I've had dreams before, which in retrospect, seemed to be strikingly precognative.

When I was much younger I started to pay careful attention to them. I realize that might be superstition, or complete folly, but it's a harmless superstition in the worst case, so why not? One might think of it as an ongoing paranormal experiment.

Over the years I've taken notes about these dreams and have worked out the mechanics of what seems to be happening--the "rules of the game" if you will. Here are my observations:

The Primary Effect - I have a "special dream" which is extremely vivid. When I wake up, I know for sure that it's one of "those" dreams. Not all vivid dreams are precognative, but when I have one of these special dreams I just "know it". I make a mental note of it.

Frequency - A long time ago (when I was about 20) to have about a dozen precognative dreams a year. Lately I haven't been getting them very much at all--maybe only 1 or 2 in the last year. I did in fact have one about two weeks ago (in it, I saw my finger cut severely, and it was dangling by a single bloody red tendon or vein.) Mind you, I have hundreds of dreams each year, as I assume everybody does, but only a few of these special dreams that I watch out for.

Time Between Dream & Manifestation - Usually, between a day to a few months after the dream, it comes true. In one case it took about a year, but that's exceptional. I know at all times which dreams I have 'pending' and I scratch them off my mental list as they manifest themselves.

The Big Twist - There almost always seems to be a critical difference between reality and the dream. It's as if an important detail becomes twisted on purpose. For instance, regarding my finger-cut dream, it might manifest itself with a horrible accident to a toe instead. Or it might be deep gash but not nearly as bad as in the dream. Or it might even happen to somebody I'm with, and not me at all. Usually what really happens is something that I never even thought of when trying to predict what's really going to happen. But once it manifests itself, I think "oh, of course, like in my dream!"

The Unavoidable Destiny - Attempts to escape the fate of the dream never work. E.g., I certainly don't want to lose a finger--I make my living with my hands. So if I take very good care of my hands, wear gloves, avoid machinery that could possibly harm me, then maybe some crazy dog will come up and bite off my toe. Only then will I learn the twist: that the 'finger' really meant 'toe', but otherwise the dream was dead on. And so I've concluded, there's just no escaping the fate of a precognative dream. It never happens exactly like the dream, but it always happens.

How Often Do They Come True? - So far, every single one. However I must stress two points: First, some of them (most of them even) are so mundane that it would be surprising if they didn't come true; second, when it comes true it always requires some mental stretch to make the facts fit the prediction, due to "the twist". Yeah I know, that's a loophole you can drive a truck through.

Metaphors - Sometimes the dreams speak metaphorically of the future. For example, I had a dream where I was fighting a person at the top of a tall building, and we were in a death struggle trying to push each other over the ledge. I ended in up great conflict with the antagonist in my dream, but there was never any fight on a high ledge.

Mundane Dreams - You would think that this mystical power would only be used to convey matters of great importance to me. Apparently it doesn't work that way. Most of my precognitive dreams are very mundane, of no interest to anybody (sometimes not even me), and most all of them were "easy guesses" that can be dismissed as coincidences. They are not guesses though, at least not conscious guesses--I have no control over what predictions my dreamscape comes up with.

If all of my precognative dreams were limited to the mundane variety, I would not even bother mentioning this. It's those few exceptional dreams that are so specific, so unlikely; when they finally happen they positively floor me and make me continue to play this mental game with myself.

My Current Prediction - My current prediction based on my recent precognative dream is this: The image of a severed finger will play a role in my life in the fairly near future.

I am giving plenty of allowances for the "twist"--it might not really be a finger, or it might not be my finger, or it might be a giant metaphor for something that will only make sense to me when it finally happens.

Just of the sake of example, I would consider any of these highly positive manifestations of the dream:

- I lose a toe, or a finger, or have a toe/finger reattached surgically
- I find myself at the emergency room due to some finger-related injury
- My wife or a close friend loses a finger or toe, or comes close
- I catch my finger in a car door and hurt it very badly

Just to show I'm not a total loon who is ready to grab at straws to satisfy the prophecy, here's a list of example of things I would not consider to be manifestations:

- I bang my finger on a table by accident and it stings for a minute
- My cat dies (the cat is a metaphor for a finger, you see)
- I cut my hand with a knife by accident (blood in the dream, blood in real life)
- Somebody gives me a gift: a red ring (sort of makes it look like my finger is bleeding)

I know, some of you may say "if you allow yourself so many variations, then it must come true, because you'll find something, or appeal to this vague concept of 'metaphor' if nothing else arrives." A legitimate criticism, but I argue that it's not entirely true. Nothing that I can recall happening in the last 10 years of my life would qualify as a manifestation of this dream. So if something in the next few weeks does, that's pretty interesting. Doesn't prove a damn thing, but it's interesting.

Anyhow, I'll keep you posted.

Disclaimer

I am well aware of rational explanations for everything I've described. I know you all can post list of rational arguments of why this is entirely bunk, and I could too, but I'd just be preaching to the choir.

Regardless of what happens it's a fun game that I am playing on my own anyhow, so I figure I might as well document it. Let's hope I type my follow-up soon, with all ten fingers operational.

Mr. Scott

8th October 2007, 03:53 AM

I have precognative dreams. Wait, let me restate that: I've had dreams before, which in retrospect, seemed to be strikingly precognative.

One problem with that is, if dreams only seem precognitive after the fact, then the aren't really precognitive. I think this is called a "postdiction" instead of a "prediction," whence prediction is not noticed or reported or interpreted until after the event. It makes the prediction rather useless, which is one of my working definitions of not-real.

What's interesting about the current dream is:

1) It was isolated. The entire contents of the dream consisted of driving my car at night and striking a deer. It lasted just two seconds.

2) Immediately awakening after the dream and having an ominous feeling like this was about to happen.

Instead of publicizing the prediction after the event it predicted occurred, I announcing it before the predicted event. This will eliminate the post-diction fallacy.

Why the 24 hour limit? Well, there has to be some limit after which we need to say the prophesy failed to come true. Otherwise, we stack the deck in favor of some random future event will give the illusion of fulfilling the prophesy. I pick 24 hours because I will certainly fall asleep and dream again, and I don't see any reason a prophetic dream should stay valid through the next sleep cycle. I drive through deer-infested areas almost every day. At least once a week I see a deer beside the road just aching to run in front of my vehicle. If there's no time limit, then just by a flip of a coin a deer hit may come. This fear of hitting one is on my mind a lot, so why wouldn't it appear in a normal non-paranormal dream?

Well, there's about 10 hours to go. No deer encounter yet.

Unalienable

8th October 2007, 07:55 AM

One problem with that is, if dreams only seem precognitive after the fact, then the aren't really precognitive.

I think you misunderstood me. Postdiction, if I understand correctly, is taking credit for precognition after an event has taken place, with no proof that the prediction was ever previously made. Like if you're watching a baseball game and you say "I just KNEW he was going to hit a home run!" but you didn't tell anybody until the ball was out of the park.

I admit, I have not always told other people about my precognative dreams, mostly because they are very boring material and nothing that would make a good story. But on other occasions I have, and those are always the more interesting occasions anyhow, when my dream is something that seems unlikely.

But this time I most certainly announced it in advance, and thanks to the technology of the internet, I've done so in a way that's basically incontrovertible.

What's interesting about the current dream is:

1) It was isolated. The entire contents of the dream consisted of driving my car at night and striking a deer. It lasted just two seconds.

Time passes strangely in dreams, but I'd say "two seconds" is a good estimate for my dream length as well. Just a single, sharp, clear vision of a dangling finger, combined with an emotion of horror.

2) Immediately awakening after the dream and having an ominous feeling like this was about to happen.

Oh, absolutely. I always know when a dream is one of the 'special' ones, it's unmistakable. When I awoke, the first thing I did was checked my hands (how can you not?) then realizing it was just a dream, I took note that I just had another precognative dream. I'm not in the habit of writing down my dreams, but I told my wife.

Instead of publicizing the prediction after the event it predicted occurred, I announcing it before the predicted event. This will eliminate the post-diction fallacy.

Isn't that what I've done too? I'm sorry if I have sort of hijacked your thread, but I figure if we're going to do one precognative dream experiment surely we have room to squeeze in a second one.

So as far as I can tell I'm playing the exact same game you are, with two important distinctions: #1) I am not limiting myself to a 24-hour time window, and #2) I am allowing leeway for possible ways that the dream can manifest in a way not literally exactly like the vision I saw.

You say that hitting something other than a deer (say, an elk) would be a failure, or at least inconclusive. I would not label that as "inconclusive", I'd call that a "hit". In my experience, that's about as good as it gets.

I don't see any reason a prophetic dream should stay valid through the next sleep cycle.

Frankly I don't see any reason why prophetic dreams should exist at all. But IF there really is such a thing as a precognative dream (and of course that's one mighty big "if") then who are we to dictate the laws they must obey?

If we want to be scientific about this, the best idea is to collect all possible data, and then we'll see if there is a pattern that can be demonstrated. Obviously if we make our testing windows too large we will dilute the confidence that we are witnessing a real phenomenon--but that doesn't mean we should just arbitrarily pick some exact number of hours and minutes to represent our cutoff time. We'll end up with statistics like these:

* % of precognative dreams that are fulfilled
* average/median of time between dream and fulfillment of dream
* "closeness of match", i.e. how faitfully the dream describes the fulfillment
* probability that the dream would be fulfilled anyhow

Depending on what the final numbers look like, we might have reason to believe there is a real phenomenon, or we might not.

Anyhow, good luck to both of us. Let's hope neither dream comes true.

ImaginalDisc

8th October 2007, 07:59 AM

I had a dream that hindsight and confirmation baises cause us to think dreams predict the future, but it hasn't come true.

:(

tkingdoll

8th October 2007, 08:13 AM

Starrman

8th October 2007, 08:25 AM

The good news for believers is they can have it both ways.

If you don't hit a deer, the prophetic dream made you more cautious in your driving, which saved both you and the deer from a great deal of grief.

Bri

8th October 2007, 08:30 AM

I think you misunderstood me. Postdiction, if I understand correctly, is taking credit for precognition after an event has taken place, with no proof that the prediction was ever previously made. Like if you're watching a baseball game and you say "I just KNEW he was going to hit a home run!" but you didn't tell anybody until the ball was out of the park.

I admit, I have not always told other people about my precognative dreams, mostly because they are very boring material and nothing that would make a good story. But on other occasions I have, and those are always the more interesting occasions anyhow, when my dream is something that seems unlikely.

But this time I most certainly announced it in advance, and thanks to the technology of the internet, I've done so in a way that's basically incontrovertible.

You're right, what you described would not technically be postdiction. It would be retrofitting.

-Bri

strathmeyer

8th October 2007, 09:00 AM

sophia8

8th October 2007, 09:16 AM

I had a vivid dream this morning that I beat somebody to death. (This is true, I really did have this dream - it was some unknown guy who was bugging me). So if I don't beat some man to death within 24 hours will this prove that dreams do not foretell the future?

JoeTheJuggler

8th October 2007, 09:49 AM

I think you misunderstood me. Postdiction, if I understand correctly, is taking credit for precognition after an event has taken place, with no proof that the prediction was ever previously made.
Yes, and this is precisely what you describe in the first few paragraphs of your first post. You admit that it's only in retrospect that the dream is a prediction. Ahead of time you only made a mental note of the dream because it was vivid. You could very easily be misremembering the dream after the fact.

Try writing them down. Then you can get an idea of the numbers, and the details won't be as likely to change.

Check out this essay on coincidence (http://skepdic.com/lawofnumbers.html). Coincidentally, one of the examples he uses is precognitive dreams.

You say that hitting something other than a deer (say, an elk) would be a failure, or at least inconclusive. I would not label that as "inconclusive", I'd call that a "hit". In my experience, that's about as good as it gets.

If you allow hitting an elk to count, what about hitting a dog? What about almost hitting a deer? Almost hitting a dog? What about a friend or relative hitting a deer? How about a friend of a relative hitting a deer?

This is called shoehorning. If you allow this to go on, then, as I said above, even low probability events become high probability. The net you cast can get really wide in a hurry. Before long, you've got a so-called prediction that is almost impossible NOT to score a hit.

<derail>Check your spelling of precognitive.</derail>

Spektator

8th October 2007, 11:59 AM

One of my friends says he had a vivid dream once in which he saw himself as a wealthy, successful novelist. He practices "visualization" and "affirmation" in which he deliberately conjures up visions of himself as a wealthy, successful novelist.

He is fifty. He has been doing this, he tells me, for thirty years now.

He has never written a novel. He is neither wealthy nor successful.

Unalienable

8th October 2007, 01:03 PM

This is called shoehorning. If you allow this to go on, then, as I said above, even low probability events become high probability.
I understand what you're saying but I believe that a great deal of leeway can be allowed, as long as you're willing to do the math behind it.

For example, suppose I have a dream that an Boeing 747 falls out of the sky and hits my house. I write my dream down in as much detail as possible, I post it right here timestamped on the JREF forum, etc.

Then, shortly thereafter, not a Boeing 747 but a meteor hits my house instead.

This is a startling coincidence. True, there are easily thousands of scenarios where I would claim a "hit": a military plane might have hit my house; a helicopter hitting my house; an airplane hitting my neighbor's house, etc. Maybe it happens on the first day after the dream, or the second day, or a week later.

But all of these events can fit neatly under a single umbrella of "large object from the sky falling on a house in my neighborhood within a month." Even though that umbrella contains millions of possible events, each event is so mind-bogglingly unlikely, that their totality is still highly unprobable.

I'm sure you'd agree that the likelihood of any large object from the heavens falling anywhere within a 10 mile radius of my house in any 30-day period are very small. (If anybody doesn't agree, contact me and we can arrange for a private wager.) Large objects falling from the sky happen rarely, and there are a lot of 10-mile radii in the world, so just do the math. Agreed, probabilities like these are hard to estimate, but they can be estimated. Actuaries do this kind of work all the time.

My point is, splitting hairs over what kind of large object it is that hits my house, or even which house, is going way overboard. If precognition was actually a real phenomenon, but subject to the kinds of variances that I've suggested, your methodology would easily conclude that it doesn't exist. That's just as irresponsible as creative-retrofitting.

Believe me, I'm not trying to distort the facts so that no matter what happens we all jump up and down and declare that a miracle has taken place. I'm approaching it from the totally open minded point of view that "I don't know if this phenomenon is real or not" and then trying to figure out the truth by examining the evidence. The best way to do that, which I have figured out, is this:

- For each dream, carefully define the umbrella of events that would be considered a "hit"
- Estimate the probability (p) of any event within that umbrella occuring on any 1 day
- Wait to see if the event happens, within a certain timeframe.
----> If it does occur, record the number of days that it took (d)
----> If it doesn't occur within the timeframe, let d be undefined (n/a)

Then you can make a table which looks sort of like this:

DREAM# p d
-------- ----------- ------
1 0.002 15
2 0.08 n/a
3 0.000017 47
4 0.15 12
5 0.000237 n/a
etc.

The key here is that the umbrella is carefully defined before the dream manifests itself. Otherwise, the experimenter would be in a position where no matter what happens, he just expands the umbrella to contain that event.

With enough data, we can determine whether what we are witnessing represents a statistically significant phenomenon. If it doesn't show any statistical significance, then maybe we're wasting our time. But if it does, the work still isn't done--an outrageous claim like this will require much more evidence to establish it. We get other people to do similar studies, and so forth, until finally it becomes so reproducible and confirmed that it's accepted as scientific fact.

Sadly, my method comes at a price: for starters, the "umbrella" is not always easy to figure out. Next, computing estimates for the probability of these events will often require a lot of effort--and if we set them too low, we will end up with false positive results. Finally, what do we make of situations where only one person (e.g. the dreamer) is witness to the dream's ultimate fulfillment? We want cold hard facts, not bigfoot sightings.

The superiority of this method over yours is that using my method we can detect even small effects with enough work. Your method would quickly detect the most grossly obvious effects, but would be utterly blind to small effects, regardless of how much work you spent on the study. I prefer to make no assumptions as to the size of the effect, the timeframe that it operates on, its accuracy rate, etc. I just want to know if the phenomenon exists whatsoever.

Alice Shortcake

8th October 2007, 01:17 PM

Madalch

8th October 2007, 02:29 PM

Bloody hell! Why can't my dreams be as interesting as yours?!
I'm sure if you paid enough attention to them to remember them, you'd find some gems.

These are just the more memorable examples (out of how many tens of thousands of dreams?)- I usually forget my dreams in the first few minutes of being awake. I'll be standing in the shower, thinking, "That was such a weird dream I had last night", and then realize that I have no memory of what it was.

They're also the ones that could be posted without violating rule 10. In high school, we were supposed to keep a "dream log" for psychology class, but I simply told the class that I had no dreams that week, since all the dreams I had were either profane, perverse, or would have gotten me beaten up.

And none of those dreams came true, either.

Mr. Scott

8th October 2007, 02:33 PM

Mr. Scott

8th October 2007, 02:35 PM

TX50

8th October 2007, 02:47 PM

I have never seen any evidence (personally - thus anecdotal) that dreams
have any precognitive function at all.

The only thing I do find intriguing about dreams is trying to work out where
some of the dream imagery comes from, and how my brain takes the merest
hints of ideas from the "waking world" and conflates them into a
dream "narrative". Some dream things I can spot the provenance of (eg.
seeing a friend in a dream wearing a dark blue coat with gold buttons. That
may come from reading a picture book about the 17th century Royal Navy)
and some I just can't place (eg. me dodging mountains made out of
marshmallow falling from the sky; waking up on being hit and crushed by
one. Where that idea comes from I have no idea).

Mr. Scott

8th October 2007, 02:56 PM

I would want to know the probability of you hitting a deer before I comment. Do you live or regularly drive in an area populated by deer? Have you ever hit a deer before? Are deer accidents commonplace, rare, or unheard of in your area? Are you about to drive somewhere where you normally don't go but deer do?

Etc. See, this affects not only why you would dream such a thing in the first place, but its likelihood of coming true.

I've hit deer at least twice (once causing significant damage to my car) but not recently. Neither time was it preceded by any metaphysical foretelling.

I regularly drive in an area and at a time in which I see deer eager to get hit. This, I believe, is what instigated the dream -- my daily fear of an accident.

The 24 hours is up. I haven't hit a deer, haven't had a close call, don't know of anyone who hit one or had a close call with a deer or anything else, and didn't even see a dead deer by the road. Nothing to shoehorn, nothing to retrofit.

Tonight, I will be driving through the highest risk deer accident zone at the worst time since the dream, and I don't mind extending the deadline until after that, but this is the kind of wiggle room that the Randi Challenge would never permit.

In the meantime I offer this YouTube entertainment: a near miss that resembles those I see often:

A6VS6qsLJeY

JoeTheJuggler

8th October 2007, 04:54 PM

I understand what you're saying but I believe that a great deal of leeway can be allowed, as long as you're willing to do the math behind it.

Why?

Also, how do you limit how much leeway? Seriously, there are countless scenarios that are almost the same as what was predicted.

My point is, splitting hairs over what kind of large object it is that hits my house, or even which house, is going way overboard. If precognition was actually a real phenomenon, but subject to the kinds of variances that I've suggested, your methodology would easily conclude that it doesn't exist. That's just as irresponsible as creative-retrofitting.

That'd be perfectly fine if the prediction was "a large object will hit my house", but that wasn't it, was it?

Believe me, I'm not trying to distort the facts so that no matter what happens we all jump up and down and declare that a miracle has taken place. I'm approaching it from the totally open minded point of view that "I don't know if this phenomenon is real or not" and then trying to figure out the truth by examining the evidence.
I'm a strong advocate of the maxim, "Extraordinary claims require extraordinary evidence." For precognition of any kind to be real would be an amazing thing--and it would require some serious rethinking (if the future is predetermined, then there is no free will, for one thing). It's really an extraordinary claim. It's way outside of known science, so there's no reason to make the rules laxer.

Also, the history of paranormal investigation is full of examples of exactly what you say you're not advocating.

For example, would you read "I don't walk alone" as the same thing as "I was confined to a wheelchair for the last few years of my life" for example? That's just the way researcher Gary Schwartz interpreted an Allison Dubois pronouncement as related here (http://skepdic.com/essays/gsandsv.html).

Again, allowing even a little shoehorning (http://skepdic.com/shoehorning.html) really distorts the whole thing. What you're left with is similar to the generalities you see in daily horoscopes. "Expect a change in your home life." "An important person is about to enter your life."

JoeTheJuggler

8th October 2007, 04:56 PM

I'm saying it will be evidence that dreams don't fortell the future, not proof.

It's still not evidence either way. With a sample size of one, you can't say anything about it at all. (It's actually just an anecdote, even though you at least ruled out issues of faulty memory by stating the prediction ahead of time.)

You could say it proves that not ALL vivid dreams will come true within 24 hours, but then you'd be making a straw-man fallacy since I don't think anyone holds that position.

Also, the 24 hour time limit was completely arbitrary, wasn't it? (Or was that in the dream somehow?)

Fnord

8th October 2007, 10:29 PM

I just woke up from a vivid dream that I hit a deer with my car at night.

Dreams are not precognitive. Many are expressions of supressed or unrequitted desires. Many are also expressions of unacknowledge desires.

Here are my guesses, for entertainment purposes only:

1) You fear that you will accidentally hurt someone very "dear" to you, while doing something completely normal.

2) You want to hurt someone who may be very "dear" to someone else, while are safe and protected in your powered armor.

3) You have a general animosity towards women.

Disclaimer: I am not a medical professional, and I have no training as a psychologist or psychiatrist. Therefore, I haven't a clue.

Unalienable

9th October 2007, 01:44 AM

Ia great deal of leeway can be allowed, as long as you're willing to do the math behind it.
Why?
Because the math will end up measuring exactly what we are trying to measure, regardless of the amount of leeway that we allow.

Maybe that is a little opaque, let me try to explain.

First, ask yourself this: what exactly are you trying to measure? It seems to me, the measurement in this case is "How lucky are the test subjects at predicting the future based on precognitive dreams?" If it's true that the subjects have a "normal level of luck" then we've basically come up empty for evidence for precognitive dreams, at least with the test subjects we've selected.

If we do happen to find ourselves with a lucky subject, then we have to ask ourselves "Is it really just luck, or is something going on here?" That's why more testing is required. And if more testing bears out the original finding, we test some more. Still not satisfied? Then we test some more. Maybe we get it to the point where the only way that you can explain the data with pure luck is 1 in a billion, but that's still not good enough for you. No sweat, we collect even more data, until it's 1 in a trillion. And on and on it goes, until even the most hardened skeptic in the world has to confess "something weird is going on here."

Also, how do you limit how much leeway? Seriously, there are countless scenarios that are almost the same as what was predicted.

By "leeway" I mean two different things: #1, how wide we stretch the umbrella (e.g. if you dream you hit a deer, does hitting an elk count?), #2, what time period we allow for the fulfillment of the dream (e.g., if you dreamt it today, would an event fulfilling the dream 200 days from now be considered as data?).

Keep in mind that both of these decisions must be made after the dream is recorded and before the dream's fulfillment. As just about everybody in this thread has said at least once, it is folly to wait to see what happens and then, only as an afterthought, make these decisions.

By my methodology, you can stretch either #1 or #2 as much as you want. If you make the error of stretching #1 too far, you can obliterate the possibility of getting a positive result. You will not create false positives by stretching the umbrella -- if anything, you'll create false negatives.

Likewise, #2 (the number of days you allow) can be as long as you have patience/funding for. It really doesn't matter. If you can afford to do a 20 year study, more power to you.

That'd be perfectly fine if the prediction was "a large object will hit my house", but that wasn't it, was it?

I can't help but to think you still don't "get it" ... it just doesn't matter. You might dream something as commonplace as "It will rain", and allow yourself 365 to see the dream come to a conclusion. And even with parameters like that you can still try to measure the phenomenon using the scientific method.

Of course I'm not saying "therefore if it rains in the next year, this guy can see the future with his dreams." But you can make a table like the sample one I provided above, with all the p's and all the d's, and figure out the odds that the test subject could come up with these predictions through pure chance.

If he's dreaming commonplace things like "It will rain" or "I will meet somebody named John or Johnson or mabe Jack", then the statistical signficance of these events could not only be small, but actually downright negative.

In my sample table above, I was of course just making up numbers for the sake of example, but if you look at my "Dream #4" there is an intentional example of exactly that situation: the probability of the event happening in one day (p) was estimated at 0.15--that's very high, akin to "It will rain". And d=12 there, so we know it took 12 days to happen. Well that's not really a "hit" ... that's a failure! Even though the person's dream did indeed come true, it provides evidence that the dreamer is worse than to be expected at predicting the future, at least in that case. Just because we allowed him 12 days instead of 1 doesn't invalidate the study. Allowing a large number of days for the dream to be fulfilled would only invalidate the study if we were so braindead as to think that it represents evidence for precognitive dreaming. If you are wise enough to recognize that this is evidence against precognitive dreaming, then go ahead and let d=1000 if you like.

I'm a strong advocate of the maxim, "Extraordinary claims require extraordinary evidence."

I think most people at this forum are as well. And yes, dreams (or anything else) that can predict the future are quite extraordinary.

It's way outside of known science, so there's no reason to make the rules laxer.

Nobody is suggesting making the rules of science laxer. Allowing 365 days instead of 1 day, or expanding the envelope to include elk as well as deer, is not making anything laxer. If we are seeking a 90% confidence level, or a 99% confidence level, or higher, we can do it.

To summarize this longwinded post in sentence: If the phenomenon does in fact exist (which I doubt) my methodology would prove its existence, within any required confidence level, regardless of how much leeway you allow.

Maybe that sounds contradictory. If we relax the constraints of what constitutes a successful dream fulfillment then surely we are slanting the test in the direction of finding a positive result, right? WRONG! Not if we also do the accompanying mathematics to figure our a proper value of p for the new definition.

Also, the history of paranormal investigation is full of examples of exactly what you say you're not advocating.

I know that, and thank you for saying.

Unalienable

9th October 2007, 02:14 AM

By the way, this is only a little off-topic, but I have a pet theory for why procognitive dreams seem to exists so strongly, for some.

My theory is that it's a protection mechanism developed over countless years of evolution. Our minds, while we sleep, try to create likely scenarios which would either cause us harm, or be advantageous. This is information which needs to be sent to the conscious brain in order to help the individual. Therefore the dream comes along with some message that punctuates its importance, as if to say "Watch out for this! It will be relevant in your life very soon!"

That's why most precognitive dreamers say exactly what Mr. Scott and myself said: that there was something weirdly important about a certain dream that normally does not apply to the typical dreams. We wake up thinking "Gosh--that felt so real, like it was a warning!" It's not really a crystal ball into the future, but just a guess based on logic that you worked out in your sleep.

For instance, Mr. Scott dreamed that he would hit a deer in his car, and he lives in deer country. This is his brains way of telling him "This is a danger, avoid it." Now that he has this image of hitting a deer firmly in his mind, the likelihood of him actually hitting a deer should only go down (not up!) because he is watching out for it. Whether he does or not hit a deer is up to chance as usual, but now it's slanted in his favor (and the deer's favor as well!) because he'll be on his guard for it.

Likewise, if I have a dream that I find some money on the sidewalk, that's my brain's way of telling me "This is a plausible way to improve your condition, be on the lookout." And without even realizing it, my eyes scan the ground for lost money with more frequency than otherwise. In this case, the odds of me actually finding lost money go up, which is what my brain interprets as a net benefit to my well-being.

The above is purely speculative but I think it's highly plausible. Precognitive dreams can actually help the individual in real ways, sometimes by "coming true" but other times by failing to come true. So why shouldn't evolution have forged a mechanism of that kind, to help with our survival and reproduction?

Mr. Scott

9th October 2007, 04:12 AM

Tonight, I will be driving through the highest risk deer accident zone at the worst time since the dream, and I don't mind extending the deadline until after that, but this is the kind of wiggle room that the Randi Challenge would never permit.

Didn't see a deer, didn't have a close call with anything.

To those who claim that dreams are precognitive, may I ask that you propose a mechanism which would make this work?

For example:

- Have future events already happened in the future, and the information of their occurence has traveled backwards in time to us?

- Has some intelligence of the universe planned to make the event happen and the dream is a warning we can use to circumvent the event?

- If precognitive dreams never let us prepare for them, why then do they happen? Why such a complex, useless mechanism?

- Why the variable time delay of hours or months? Where is the information stored in this time? How is it determined the length of time it's stored?

- What is the intelligence that converts the actual event into symbolic or imprecise information (like a toe becoming a finger, or a deer becoming someone dear to us)?

I personally don't believe dreams are metaphysically precognitive, but do believe they represent recent experiences or hopes and dreams -- a kind of noise in the brain during "system update." There's some good science behind this but I haven't time to look it up right now.

sophia8

9th October 2007, 04:36 AM

EHLO

9th October 2007, 07:17 PM

Years ago, I read one of those "interpret my dream for me" magazine columns. One of the readers sent in an account of a recurring dream in which he was trapped in a blazing house. At the end, he added that he was a fireman. The columnist - a remarkably sensible woman - replied that had he been in any other profession, she would have been worried by his dream. But since being trapped in an inferno was a likely hazard for a fireman, his recurring dream was a way for his subconscious to get him thoroughly accustomed to the scenario; then, if he really was trapped in a fire, he wouldn't be frozen by panic.

Ha, a sensible dream interpreter indeed, the only people of that ilk that I have come across would almost certainly have concluded that the dream was a precognition or the re-living of a past life experience...

I had a semi lucid dream this morning that I was Matthew Broderick flying over the north Australian outback, which for some reason was full of wind farms. I'd like to get an analysis of that!

JoeTheJuggler

9th October 2007, 07:38 PM

Because the math will end up measuring exactly what we are trying to measure, regardless of the amount of leeway that we allow.

Maybe that is a little opaque, let me try to explain.

You're wrong.

It's not the least bit opaque, it just isn't a proper way to run an experiment.

Have you heard of the Texas Sharpshooter's Fallacy?

If I shoot a slug into a wall from some distance, then after the fact walk up to where it hit and draw a tiny little circle around the slug in the wall, what are the odds of hitting that exact spot?

By allowing "leeway" what you're opening the door to is not stating a hypothesis until after you look at the results. That's what's going on with the sharpshooter's fallacy too. It doesn't even matter how many times you do this (how big your sample size is) if you allow this. I can always hit the wall and then draw a tiny circle around the slug and claim extraordinary shooting abilities.

It really doesn't matter at all that you can calculate the odds of hitting that little target if the target didn't exist before running the test.

I understand that you said you can somehow allow lots of leeway but define what counts ahead of time, but you really can't.

If you allow hitting an elk to count, would you allow hitting a dog? Would you count almost hitting a deer? Would you count a close friend actually hitting a deer to count? There are an infinity of these, and you cannot anticipate them all.

Further, there's no reason whatsoever to allow any near misses to count as a hit. (Why would you count hitting an elk but not count hitting a dog? Why would you count someone else's hitting a deer? Or why not, considering that you would count hitting an elk?)

Again, it is an extraordinary claim. As such, methodology should be even tighter than circumstances where the hypothesis isn't so unparsimonious--not more lax.

JoeTheJuggler

9th October 2007, 07:49 PM

To continue more to the point you've been making that the probability of each scenario would make up for the fact that leeway is being offered.

That's no so. At best you're making a whole plethora of hypotheses to test against one trial, and you'll toss out any that don't fit. In the current example, you'd like to allow hitting an elk to count (even though the precognitive dream was of a deer). Doesn't that mean you have two hypotheses? Or is it now just one broader one (I will hit a deer or an elk-- or even broader, hitting any animal)? I suspect if the dreamer did hit the deer, you'd want to use the full very long odds against hitting a deer and not the much lower one of hitting any animal.

Let's take a simpler example (where all the possibilities are known, unlike these dreams): predicting the roll of a die. Let's say you pick the number 4, but state ahead of time that you'll allow leeway, so 3 and 5 will also count. What if the 4 comes up? Do you say the prediction was of a 1:6 probability event? In fact, it was a 1:2 probability event given that definition.

With precognitive dreams, the odds of the event as dreamt are probably very long, but when you start allowing shoehorning, they drop dramatically. Especially if such leeway is poorly defined, as I contend it must be since you can't possibly anticipate all possible near miss events.

Also, again, the history of people doing this sort of research is to do precisely that. Since this is known, the only way to tighten the methodology is. . . . to tighten it.

Bri

9th October 2007, 09:16 PM

Additionally, Unalienable doesn't specify a specific time frame other than "if something in the next few weeks does [happen], that's pretty interesting." Of course, if nothing happens within "a few" weeks, he'll just extend the time limit.

Just of the sake of example, I would consider any of these highly positive manifestations of the dream:

- I lose a toe, or a finger, or have a toe/finger reattached surgically
- I find myself at the emergency room due to some finger-related injury
- My wife or a close friend loses a finger or toe, or comes close
- I catch my finger in a car door and hurt it very badly

So if he or anyone he knows gashes a thumb or a finger at any time from this point forward, he'll claim a hit. Such an event wouldn't be rare at all -- there's a very good chance that someone you know will at some point cut their finger or thumb or catch a finger or thumb in a door, and possibly even have to go to the emergency room. I've done it a couple of times myself. It would probably be rarer if nobody he knows has something happen that would qualify as a hit based on his shoehorning examples above.

-Bri

Beth

9th October 2007, 09:27 PM

If I shoot a slug into a wall from some distance, then after the fact walk up to where it hit and draw a tiny little circle around the slug in the wall, what are the odds of hitting that exact spot?

I think, Unalienable can correct me if I'm wrong, that he is trying to describe a process more like drawing a circle that includes the bullet that encompasses 1/4 of the wall and then concluding that there was a .25 probability of hitting in that quarter.

Done properly, it doesn't seem an unreasonable approach to me. Hard to do properly though.

JoeTheJuggler

9th October 2007, 10:17 PM

Yes, I recognize that he's trying to account for this leeway in interpreting hits ahead of time, so it's not really the Texas Sharpshooter's Fallacy in theory.

It still makes no sense. Why would you allow events that weren't predicted to be considered hits in a test of precognition?

Why would you allow some unpredicted events to count but not others?

If you say ahead of time that you'll count hitting an elk, and then the person hits a cow (not something anticipated), how can you defend not including that as a hit? What about watching a movie (or reading a book or magazine article) that involves someone hitting a deer?

There's no way you can anticipate all of the possible near misses, and there's no way you could defend only the ones you anticipated when a great many more make just as much sense.

Also, doing it the way Unalienable suggests would punish an actual, legitimate, dead-on hit because you'd have to include the probability of all the other things you "leewayed" in to the prediction. In other words, it would devalue a real hit.

Bri

10th October 2007, 06:44 AM

I think, Unalienable can correct me if I'm wrong, that he is trying to describe a process more like drawing a circle that includes the bullet that encompasses 1/4 of the wall and then concluding that there was a .25 probability of hitting in that quarter.

Done properly, it doesn't seem an unreasonable approach to me. Hard to do properly though.

Except he's not doing that. He's not listing all of the possible hits ahead of time. He's retrofitting an event into a hit after the fact. Unless I misread, Unalienable said that there was no way he could anticipate all events that would constitute a hit ahead of time, which means that there is no way to calculate the probability of a hit ahead of time.

So to be fair, he's drawing several circles on the wall beforehand, then firing, and upon missing any of the circles he drew, he draws a circle around the bullet hole. He then calculates the probability of hitting all of the circles ignoring the fact that there are a million other events that he would have considered a hit. If his bullet had hit a different spot that didn't contain a circle, he still would have called it a hit with exactly the same probability.

-Bri

Dan O.

10th October 2007, 08:12 AM

If you wanted to pursue this direction of measuring predictions, you should turn it into a wager. One party believes they have inside information that some event will occur, the other party believes that the events will occur at their normal frequency. If the two parties can agree on specific events and time frames that will constitute a hit and the payment that will be made should there be a hit or no hit then the wager can be made.

After a sufficient number of such wagers we can evaluate wether the dreams have been financially beneficial.

Bri

10th October 2007, 08:39 AM

That's a good suggestion. I would never take such a bet unless Unalienable were to list the specific events that would constitute a hit beforehand, unless the probability of one of the events occurring by chance was below 50%, and unless a reasonable time frame was specified.

However, I'd be perfectly willing to wager on my own prediction (even though I would be making it up without any precognition at all) based on the same rules that Unalienable sets for himself (no time frame, no predetermined list of what constitutes a hit, allowing myself to determine whether an event constitutes a "hit" after the fact, etc.). Care to wager, Unalienable?

-Bri

Beth

10th October 2007, 08:48 AM

JoeTheJuggler

10th October 2007, 10:03 AM

Unless I misread, Unalienable said that there was no way he could anticipate all events that would constitute a hit ahead of time, which means that there is no way to calculate the probability of a hit ahead of time.

I think he's saying that he can anticipate all the possible near-misses-that-would-count-as-hits ahead of time and calculate the probability of treating all of them as a single prediction. (Great big circle covering most of the wall.)

I'm arguing 1) that's not possible and 2) there's no reason whatsoever to treat misses as hits.

JoeTheJuggler

10th October 2007, 10:12 AM

To continue with analogy of circles and bullet holes, think of it as a typical target with a series of nested concentric circles. The bullseye is the actual prediction. The closer to the bullseye, the lower the chance that it is simply a random coincidence. The farther away, the greater the probability of a coincidence. I think Unalienable is simply trying to determine how large the circle has to be if it is made large enough to encompass the actual event. Then he computes the probability of a hit based on the size of the circle.

I think that's a pretty good analogy, but I think it's really bad experimental design. Basically, you'd score a bullseye at its full value (probability of 1:gazillion) even though you've effectively made the definition of any hit at all a much higher probability (1:10 perhaps?). If you hit the bullseye, you can't just scrap all the other "hypotheses", so you would have to treat it, in Unalienable's terms, as a single umbrella hypothesis and score it a hit with a 1:10 probability.

Either way, it's not right.

Also, hitting an elk (or a dog, or an infinity of other equally "near" misses) isn't almost hitting a deer the same way that hitting a spot near a target is almost hitting a target. In other words, with the bullseye analogy, we can measure proximity, but the only way to do that with the deer dream would be to measure how many inches you came from hitting a deer in a near miss.

Other events that can be shoehorned to fit the dream are all equally distant--since they were not hitting a deer. (Their probability is irrelevant here; we're just talking about how near they are to being a hit.)

Bri

10th October 2007, 12:10 PM

There are at least two metrics that I can think of that might be tempting to use to determine how good of a match an event might be considered in relation to a prior prediction. The first is how "close" the event is to the prediction, and the second is the probability of the event occurring by chance. How rare an event is might actually be possible to calculate, at least in theory. If it were possible to calculate how "close" an event were to the prediction, we could normalize the two values and multiply to get a "score" from 0.0 to 1.0 for an event. A score of 0.5 would represent something like a guess rather than an actual prediction -- it would not be particularly rare nor particularly close to the prediction. We might consider an event that got a higher score (i.e. that was "closer" to the prediction and rarer) to be considered a better match for a particular prediction.

This scheme would automatically regulate for "unimpressive" predictions (predictions of common events or vague predictions that could be easily matched to many events) since the maximum score would already be limited by the score of the predicted event.

In Unalienable's case, for any given prediction we could perhaps find a few events that occurred within a given timeframe (perhaps Unalienable's lifetime) that best fits his prediction, and we could calculate a score to see how well they really did. We would then choose the one that has the highest score and assume it to be the predicted event. If the score wasn't significantly above 0.5, it would likely have happened by chance or wasn't very close to the prediction. We could even determine a threshold for a "hit" (say 0.8 or 0.9).

But to see if Unalienable had any real powers, you'd have to do this on many predictions.

I think Unalienable would be surprised at how unspectacular his predictive powers are given such a measure. Most of the events (such as you or anyone you know slamming their hand in a door) wouldn't even come close because they are more likely to happen or not (their score would be well below a 0.5). But, if it happened, losing a thumb instead of a finger might be both rare enough and sufficiently close to the original prediction to receive a fairly high score if it happened and if Unalienable could show that a significant number of his predictions received similarly high scores.

BTW, according to this article (http://findarticles.com/p/articles/mi_g2601/is_0013/ai_2601001392):

Traumatic amputation is the accidental severing of some or all of a body part. A complete amputation totally detaches a limb or appendage from the rest of the body. In a partial amputation, some soft tissue remains attached to the site.

Description

Trauma is the second leading cause of amputation in the United States. About 30,000 traumatic amputations occur in this country every year. Four of every five traumatic amputation victims are male, and most of them are between the ages of 15-30.

Traumatic amputation most often affects limbs and appendages like the arms, ears, feet, fingers, hands, legs, and nose.

Causes & symptoms

Farm and factory workers have greater-than-average risks of suffering injuries that result in traumatic amputation. Automobile and motorcycle accidents and the use of lawnmowers, saws, and power tools are also common causes of traumatic amputation.

Presumably Unalienable would count traumatic amputation of any type to be pretty close to his prediction. Assuming Unalienable is male and American, and that there are around 150 million males in the U.S., he has a 1 in approximately 6250 chance each year of having a traumatic amputation. If he is between ages 15 and 30 the chances are higher. If he belongs to one of the risky occupations or regularly engages in one of the risky behaviors his chances increase further. While slim, Unalienable's chances of a traumatic amputation aren't as slim as one might think.

Still, if you consider that well over 50% of the population slams their hand in a door at some point in their lives, the event that Unalienable slams his hand in a door would not be close to the event predicted and would score well below 0.5. And given that even closer to 100% of the population knows someone who has slammed their hand in a door, that event would score very close to a 0.0.

-Bri

JoeTheJuggler

10th October 2007, 12:37 PM

JoeTheJuggler

10th October 2007, 12:42 PM

And this umbrella method really is retrofitting. If you don't know what the "prediction" was until after the event, it is not a prediction.

Bri

10th October 2007, 01:13 PM

I contend that it's impossible to calculate the second metric, how close a near miss event is to the prediction, AND that it's impossible to anticipate all the possible "near" misses.

It's not going to be an exact science for sure, but I think it's possible to make a general guess as to how close a particular event is to a prediction. You don't have to anticipate all near misses -- this is an after-the-fact calculation. If you think a particular event that occurs is close to the predicted event, calculate the score. If you later experience another event (within the time limit), you could calculate the score for that one and see if it's a better match for the prediction. The goal is to find the event that best fits the prediction. There will always be a best fit -- the question is how good of a fit is the best fit event?

I also say they're all equally close in that they're NOT the event. Would a close call with a deer be somehow more of a hit than my brother actually hitting a deer? How could you compare the relative proximity to the prediction of hitting an elk and seeing a movie where hitting a deer is a prominent part of the plot?

They might all be about the same distance from the original prediction. Certainly closer than being anally probed by an alien, which although rare would not be even remotely close to the predicted event.

(Surely just seeing a deer on or near a roadway wouldn't count. . . . or would it?)

Any event would "count" but wouldn't yield a very high score I wouldn't imagine.

And I repeat my query--is there any valid reason to count a miss as a hit at all (i.e. to allow shoehorning or "leeway")?

Yes, because that's what the claim is. If it were accurate, a premonition such as "be careful -- something unusually bad is going to happen to you today" might be useful. Even if you were unable to change the outcome of the event (as I think Unalienable indicated) such a prediction might still be interesting.

-Bri

JoeTheJuggler

10th October 2007, 02:09 PM

Any event would "count" but wouldn't yield a very high score I wouldn't imagine.

So if any event counts, then there's no chance of the "prediction" not coming true. It's a bad design.

Also, how do you say one event that wasn't predicted is somehow closer than any other event that wasn't predicted? You said you could make a "general guess" but what would it be based on?

Is there any argument that any of the near misses I've suggested (hitting a dog, seeing a movie or reading an account about hitting a deer, or a relative hitting a deer, etc.) are not as valid as any other near miss? Is there any reason to say that one is closer to the prediction than any other?

Yes, because that's what the claim is. If it were accurate, a premonition such as "be careful -- something unusually bad is going to happen to you today" might be useful. Even if you were unable to change the outcome of the event (as I think Unalienable indicated) such a prediction might still be interesting.

I thought the claim was that some vivid dreams were precognitive. If you define that to include shoehorning, then it's not a testable claim (not even a falsifiable hypothesis).

I disagree about the utility of such dreams (which has nothing to do with the truth value of the claim, BTW). If any precognitive dream gives you no information at all until after the event, it is useless.

If a dream of hitting a deer can mean anything from actually hitting a deer to some of the near misses that I've suggested (seeing a movie or reading an account about hitting a deer or a relative hitting a deer), then you can't even generalize it to a warning to "be careful of everything".

A former girlfriend and believer in all things woo told me one morning that she'd had a vivid dream about me being in a car accident, and that I should be extra careful. Now I'm fully aware that driving a car is about the most dangerous thing any of us do, and an admonition to be careful is always a good reminder, but she thought of it as a premonition or psychic warning.

I had a lot of driving to do that day, but I wasn't in an accident or any close calls. I did get stuck in a big traffic jam that was caused by an accident (which is nearly a daily occurrence). When I told her about it, she shoehorned in the accident that caused me some relatively minor inconvenience as a "hit".

"See, my dream was right! There was an accident!"

AmyWilson

10th October 2007, 02:21 PM

Yes, dreams predict the future. :)

My father has told me of dreams of his that have come true.

It's true.

Bri

10th October 2007, 02:36 PM

So if any event counts, then there's no chance of the "prediction" not coming true. It's a bad design.

No it's not bad design. The question isn't whether a particular prediction predicts an event, but how well it predicts an event. A score can indeed be assigned to any event and any prediction, but a score of 0.5 would indicate that the prediction was no better than a random guess. A "successful" prediction would have to be higher than some threshold (say, 9.0 or greater).

Also, how do you say one event that wasn't predicted is somehow closer than any other event that wasn't predicted? You said you could make a "general guess" but what would it be based on?

Any event that was completely unrelated to the prediction would have a score of 0.0. No need to guess in that case, unless you feel that there is a very loose connection in which case you can give it a 0.1 or 0.2. Who cares? If it's less than 0.5 it's no better than chance anyway, which means a definite "miss."

Is there any argument that any of the near misses I've suggested (hitting a dog, seeing a movie or reading an account about hitting a deer, or a relative hitting a deer, etc.) are not as valid as any other near miss? Is there any reason to say that one is closer to the prediction than any other?

There might be, but I doubt it would really matter since none of those would likely be close enough and rare enough to meet the threshold.

I thought the claim was that some vivid dreams were precognitive. If you define that to include shoehorning, then it's not a testable claim (not even a falsifiable hypothesis).

It's testable, but obviously somewhat subjective. It is a pretty good tool for being more accurate and less subjective as far as how "close" an event is to the prediction. That's my point.

I disagree about the utility of such dreams (which has nothing to do with the truth value of the claim, BTW). If any precognitive dream gives you no information at all until after the event, it is useless.

I can agree with that. However, my example does give you enough information before the event to possibly avoid the event. If it were possible to reproduce such a prediction with any accuracy, it might be something worth looking at further.

If a dream of hitting a deer can mean anything from actually hitting a deer to some of the near misses that I've suggested (seeing a movie or reading an account about hitting a deer or a relative hitting a deer), then you can't even generalize it to a warning to "be careful of everything".

Yes, but again if you generalize it that way the score will be pretty close to 0.5 or a bit above (nowhere near the threshold of 9.0) since it would be neither particularly rare nor very close to the predicted event. And indeed if you predict hitting a deer and any of those events occurred, it probably would be slightly peculiar compared to chance, though nowhere near statistically interesting unless similar results could be reproduced many times, in which case it might be worth looking at further.

A former girlfriend and believer in all things woo told me one morning that she'd had a vivid dream about me being in a car accident, and that I should be extra careful. Now I'm fully aware that driving a car is about the most dangerous thing any of us do, and an admonition to be careful is always a good reminder, but she thought of it as a premonition or psychic warning.

I had a lot of driving to do that day, but I wasn't in an accident or any close calls. I did get stuck in a big traffic jam that was caused by an accident (which is nearly a daily occurrence). When I told her about it, she shoehorned in the accident that caused me some relatively minor inconvenience as a "hit".

"See, my dream was right! There was an accident!"

Using this scheme, the fact that car accidents by other people are nearly a daily occurrence places the probability very high, which would make the resulting score very low. It would be nowhere near the 9.0 threshold that would qualify as a "hit."

-Bri

JoeTheJuggler

10th October 2007, 06:26 PM

I understand what you're saying, but I only see it as practical method if you arbitrarily limit the number of near miss events to just a few (to keep the probability low enough to rule out chance hits with some confidence).

I say arbitrary because I still don't see why hitting an elk is "closer" to the prediction of hitting a deer than any of the things I mentioned (a relative hitting a deer, hitting a dog, seeing a movie or item on TV about someone hitting a deer, reading an account about hitting a deer, etc.)

So this method would be OK, if you limit the number of possible near misses to just a few, but it still doesn't make sense to me as a way to test the claim of precognition.

Say you set up just a few near miss scenarios, and you end up with an umbrella probability of 1:1,000,000. (You still can't assign separate probabilities to each event and only count the best fit after the fact. That's essentially making a hypothesis after the test is run. In the sharpshooter's analogy, it'd be like drawing a several circles of various sizes on the wall, then shooting. If you hit a circle, you can't just ignore the total area of the wall that was in circles and calculate the probability of hitting just that circle that you did hit. Unalienable was right about that--you have to include the entire "umbrella".)

Let's say none of those events occurs, but one of these others that I mentioned (arbitrarily excluded to prevent the probability increasing so much as to make the test pointless) happens. Believers will surely question why you included what you did but excluded what happened. (And rightly so, IMHO.)

Unalienable

10th October 2007, 06:41 PM

Except he's not doing that. He's not listing all of the possible hits ahead of time. He's retrofitting an event into a hit after the fact. Unless I misread, Unalienable said that there was no way he could anticipate all events that would constitute a hit ahead of time, which means that there is no way to calculate the probability of a hit ahead of time.

No, I said no such thing. You absolutely must define what constitutes a "hit" in advance!!! How many times do I have to say that before people stop saying that I claim otherwise?

I think what's happening here is that two of my posts have become blurred into one.

I started this thread describing purely anecdotal observations that I made which were outside of any kind of controlled experiment. I never presented this as some sort of evidence for precognition. I just remarked at how interesting this effect appeared to be real, fully acknolwedging that if a legitimate study was done this "mysterious phenomenon" would likely vanish completely.

Then I came back and outlined a method which I believe would constitute a legitimate study, and I still stand behind my method. I think some people here read my first post, immediately recognized the possibility of retrofitting and other flaws, and then assumed that my subsequent methodology would be based on the same fallacies. But if you read carefully you'll see that I did not.

The "Texas Sharpshooter Fallacy" has nothing to do with my proposed methodology. I advocate drawing the circles first, then seeing if something hits them. The amount of time that you wait before something hits the circles doesn't really matter, but it should go without saying that if enough time elapses, you've lost any chance of demonstrating that the dream was precognitive and in fact start to enter into the realm where you have evidence against it, even if the dream comes true.

That might sound like a contradiction, but it's not. If I dream that it rains, and we then have a giant drought, my dream will still come true... 45 days later. So call it a "hit" if you like, but the bottom line is that the dream prediction was worse at predicting the future than what one would expect by chance. And that's what we are really trying to measure: whether or not dreams are better predictors of the future than what we would expect by chance.

Also, doing it the way Unalienable suggests would punish an actual, legitimate, dead-on hit because you'd have to include the probability of all the other things you "leewayed" in to the prediction. In other words, it would devalue a real hit.

That's absolutely right. There is always the possibility that a sound experiment will fail to demonstrate the existence of a real phenomenon because, as you don't fully understand the phenomenon, you might fail to make an experiment which will prove it. At that point, you go back to the drawing board and make another experiment: perhaps next time you make the event-sets smaller, or bigger, or you simply give up.

Unfortunately this entire study is mired in difficulties. I can name the three which I think are the biggest problems.

(1) The job of coming up with a suitable "umbrella of events" to constitute a hit is very tricky business.

(2) The math is hard. If some fool gave me a million dollars to actually pursue this research, the first thing that I would do is hire a small team of insurance actuaries. These guys are amazing, they can figure out the odds of anything. Have you ever heard of "meteor insurance"? Or "animal attack" insurance? If I wanted to, I could buy "insurance against an alligator biting off my left index finger", and these guys can actually estimate a good "p" value for that happening.

(3) Sadly, actuarial statistics is not an exact science, which makes our job even harder. We would have to witness a very large effect to be able to have faith in what we are witnessing. Critics of our experiment would always be able to say that we rigged our estimates of "p" so small that expected results seem unusual. The ideal dream would be something like "lottery numbers" where the odds are very simple to compute.

In short, I continue to maintain that my methodology is perfectly sound, but in practice would require a lot of time and money to actually do right. And no, I have no interest in actually doing it, but I find it interesting to discuss.

Unalienable

10th October 2007, 07:03 PM

Consider the problem from the point of view of someone who has dreams that seem, subjectively, to come eeriely close to later events, but it's not exactly the same. It can be difficult for the person who has experienced the premonition to simply dismiss it as coincidence. But how can you compute the probability of getting as close as you actually got after the event occurred? I think that is what he is trying to accomplish.

To continue with analogy of circles and bullet holes, think of it as a typical target with a series of nested concentric circles. The bullseye is the actual prediction. The closer to the bullseye, the lower the chance that it is simply a random coincidence. The farther away, the greater the probability of a coincidence. I think Unalienable is simply trying to determine how large the circle has to be if it is made large enough to encompass the actual event. Then he computes the probability of a hit based on the size of the circle.

I understand what you're saying Beth, and it sounds very tempting, but that's exactly the kind of nonsense that would lead us to false conclusions. After the dream is noted, my life will be filled with dozens of notable events every day, all of which can be used to compare against the original dream. Even the most dismal failure of a prediction could be placed somewhere, on some bullseye chart, if you invent the chart after the fact.

Therefore I suggest instead defining the circle in advance. True, you don't know how big to make this circle to prove the phenomenon is real, but then again, we don't really even know IF the phenomenon is real. All you can do is try.

One way to prove that something real is happening, is to go through these steps:

(1) Have the dream and record it.

(2) Create the event-umbrella based on the dream.

(3) Compute the probability of an event occuring within the umbrella during any single timeslice (e.g. 1 day).

(4) Wait for an event to fall within the event-umbrella and record the number of timeslices (days) it took to come to fruition.

Repeat those four steps many times over and you can start to get a picture of whether or not these event-umbrellas are better predictors of the future than we should expect by chance.

Consider for the sake of analogy, suppose somebody approached JREF with the claim, "If you look at a playing card from the deck, I can get a psychic sense of what card it is. I will often get the suit right, or the denomination (value) right, or sometimes I will be only one pip off (e.g. I guess a 7 when it's really an 8), and I can do this with incredible accuracy ... but in my experience, I will very rarely get the exact card right." In other words, they are an amazingly good "near guesser" but they are horrible at making direct hits. This is a testable claim and fully worthy of the million dollars. It would be trivial to design an experiment to test this claim to a 99.9% confidence, provided that you can nail them down to define exactly what it is that they claim to be able to do. The fact that the applicant does not claim to be able to predict the cards dead-on does not invalidate the paranomal claim.

Bri

10th October 2007, 08:34 PM

I understand what you're saying, but I only see it as practical method if you arbitrarily limit the number of near miss events to just a few (to keep the probability low enough to rule out chance hits with some confidence).

I think you misunderstand. There will only be one event that actually occurs that is the "closest fit" to the dream. This is the event that Unalienable would say the dream predicted. We can then calculate a score that essentially tells us how close the event was to the dream that supposedly predicted it. My guess is that the vast majority would be way beneath the threshold of 9.0 (or whatever you use as the threshold) which means that the prediction would not be particularly impressive.

So this method would be OK, if you limit the number of possible near misses to just a few, but it still doesn't make sense to me as a way to test the claim of precognition.

The number is limited to only one in most cases, unless more than one "near miss" occurred (which would be highly unlikely). In that case, calculate the scores for both and use the highest.

-Bri

Bri

10th October 2007, 08:57 PM

No, I said no such thing. You absolutely must define what constitutes a "hit" in advance!!! How many times do I have to say that before people stop saying that I claim otherwise?

For the dream you posted about severing a finger, you posted four possible "hit" scenarios, but you also said that any similar scenario would also be a hit. Where did you define a precise list of scenarios that constitute a hit in advance, and say that no other scenarios would count as a hit?

I think what's happening here is that two of my posts have become blurred into one.

That's possible. Perhaps I missed the list.

The "Texas Sharpshooter Fallacy" has nothing to do with my proposed methodology. I advocate drawing the circles first, then seeing if something hits them.

OK, so where did you define the "circles" in any precise way that wouldn't allow for retrofitting?

The amount of time that you wait before something hits the circles doesn't really matter, but it should go without saying that if enough time elapses, you've lost any chance of demonstrating that the dream was precognitive and in fact start to enter into the realm where you have evidence against it, even if the dream comes true.

The time element can also be included as a third metric by calculating the likelihood of the event occurring at a particular time.

In short, I continue to maintain that my methodology is perfectly sound, but in practice would require a lot of time and money to actually do right. And no, I have no interest in actually doing it, but I find it interesting to discuss.

I think it is a useful exercise to discuss the values you would place on the metrics. I think they would be rather enlightening. We can pretty well estimate the probability of an event occurring by chance (even taking into account the time factor), but the more difficult metric is determining how "close" the event is to the event in your dream.

For example, you said that you would consider slamming your hand in a door to be a "hit." We can estimate that it's not unlikely that you will slam your hand in a door at some point during the remainder of your lifetime (the longer the time limit, the more likely you are to slam your hand in a door during that time). But estimating exactly how "close" that is to your original dream is difficult.

Still, given the fact that slamming your hand in a door during your lifetime isn't a rare event, even if we deemed it to be extremely "close" to the event described in your dream (which it isn't) the event would still score a relatively small amount -- over 0.5 probably, but definitely well below any reasonable threshold that would point to the premonition being real. Let's say that 1 out of every 4 people your age will slam their hand in a door during the remainder of their life (using your expected lifespan). I think that's being generous since it would likely be more than that, but let's just say it's 25%. That would mean that the "probability" metric would be 0.75. That number would then be multiplied by the "closeness" metric, so 0.75 would be the highest score you could get, even if you were to deem the event to be "dead on" with the event in the dream (which it clearly isn't). So even if you did happen to slam your hand in a door, I'd say that your score would end up around 0.55 or so, which isn't very impressive compared to a threshold of 0.9. In other words, your dream didn't really predict that event much better than a random guess would have.

-Bri

Mr. Scott

10th October 2007, 09:35 PM

The job of coming up with a suitable "umbrella of events" to constitute a hit is very tricky business.

Unalienable, I'm trying to divine what your position on this issue is and I'd like to know why you voted in this survey that you believed such dreams literally came true, but now are talking about a "suitable umbrella of events." It reads like backpedaling to me.

I personally do not believe dreams metaphysically predict the future in any way shape or form, because:

A) I know how the mind plays tricks on us with habits like shoehorning and retrofitting. (Not to mention people who lie for attention.)

B) There is absolutely no hypothesis one can come up with for a physical mechanism which would account for it.

Bri

10th October 2007, 09:36 PM

One way to prove that something real is happening, is to go through these steps:

(1) Have the dream and record it.

(2) Create the event-umbrella based on the dream.

(3) Compute the probability of an event occuring within the umbrella during any single timeslice (e.g. 1 day).

(4) Wait for an event to fall within the event-umbrella and record the number of timeslices (days) it took to come to fruition.

I think this is very close to what I'm talking about. The threshold score I mentioned is simply the size of the circle, and the metrics are simply a way of determining whether a particular event falls within the circle or not. However, I completely disagree as to the size of your circles and/or your criteria to determine whether a particular event falls within the circles if you claim that a dream about severing a finger could reasonably be considered a prediction of slamming your hand in a door. I think the threshold score (the "size" of the circle) needs to be up around 0.9 rather than around 0.55 (which is where I estimate slamming your hand in a door to be in relation to your dream). Remember, that 0.5 would be the average score given to some random event if the "dream" were simply made up and were not a premonition at all.

The reason I think the metrics are useful is because they make you look at the size of your circle more realistically compared to chance. At first you might think that slamming your hand in a door is reasonably "close" to the event in the dream, but considering how often people slam their hands in doors, that event occurring at some point in your life would not be at all unusual regardless of the dream.

To even qualify for a threshold of 0.9, an event would have to have a 1 in 10 or less probability of occurring within the given time frame. If only 1 in 10, the dream would have to be spot on to score above the threshold, if a bit less than 1 in 10 it could vary slightly from the event and score above the threshold, and if much less than 1 in 10 it could vary more from the event and score above the threshold.

-Bri

JoeTheJuggler

10th October 2007, 10:24 PM

I think you misunderstand. There will only be one event that actually occurs that is the "closest fit" to the dream. This is the event that Unalienable would say the dream predicted. We can then calculate a score that essentially tells us how close the event was to the dream that supposedly predicted it. My guess is that the vast majority would be way beneath the threshold of 9.0 (or whatever you use as the threshold) which means that the prediction would not be particularly impressive.

That doesn't sound quite like what Unalienable is proposing, but the problem with this one is that it's still a fishing trip. If you don't state a clear hypothesis before you start, it's not kosher.

You're saying whatever event happens that is "closest" to the prediction (and I still don't get how you make that assessment) becomes the hypothesis, and all that's left is to calculate the odds of it happening times the "closeness to prediction" quotient. It's still an after-the-fact calculation, and you're still ignoring the probabilities of all the other possible events that didn't happen that could have also counted--in fact, there's an infinity of them since there's no such thing as a miss in this set up.

I suppose I could be persuaded to this if you could tell me how to assess closeness to the prediction (of events not predicted) and how you could calculate those ahead of time (which is what Unalienable is suggesting).

JoeTheJuggler

10th October 2007, 10:30 PM

I understand what you're saying Beth, and it sounds very tempting, but that's exactly the kind of nonsense that would lead us to false conclusions. After the dream is noted, my life will be filled with dozens of notable events every day, all of which can be used to compare against the original dream. Even the most dismal failure of a prediction could be placed somewhere, on some bullseye chart, if you invent the chart after the fact.

Therefore I suggest instead defining the circle in advance. True, you don't know how big to make this circle to prove the phenomenon is real, but then again, we don't really even know IF the phenomenon is real. All you can do is try.

One way to prove that something real is happening, is to go through these steps:

(1) Have the dream and record it.

(2) Create the event-umbrella based on the dream.

(3) Compute the probability of an event occuring within the umbrella during any single timeslice (e.g. 1 day).

(4) Wait for an event to fall within the event-umbrella and record the number of timeslices (days) it took to come to fruition.

Repeat those four steps many times over and you can start to get a picture of whether or not these event-umbrellas are better predictors of the future than we should expect by chance.

Consider for the sake of analogy, suppose somebody approached JREF with the claim, "If you look at a playing card from the deck, I can get a psychic sense of what card it is. I will often get the suit right, or the denomination (value) right, or sometimes I will be only one pip off (e.g. I guess a 7 when it's really an 8), and I can do this with incredible accuracy ... but in my experience, I will very rarely get the exact card right." In other words, they are an amazingly good "near guesser" but they are horrible at making direct hits. This is a testable claim and fully worthy of the million dollars. It would be trivial to design an experiment to test this claim to a 99.9% confidence, provided that you can nail them down to define exactly what it is that they claim to be able to do. The fact that the applicant does not claim to be able to predict the cards dead-on does not invalidate the paranomal claim.

I think the only quibble I have with your set up now is step number 2. (Aside from the issue that shoehorning is to be allowed at all when the claim is for a precognitive dream. If you don't know what the dream means ahead of time, it's simply not precognitive.)

You can do this with picking a playing card, but I don't think anyone does anything like that with precognitive dreams. With dreams, my problem is the same as before: how can you identify all the possible near misses? and, what constitutes nearness (i.e. how do you measure "nearness" of events that weren't actually predicted)?

JoeTheJuggler

10th October 2007, 10:47 PM

(1) The job of coming up with a suitable "umbrella of events" to constitute a hit is very tricky business.

(2) The math is hard. If some fool gave me a million dollars to actually pursue this research, the first thing that I would do is hire a small team of insurance actuaries. These guys are amazing, they can figure out the odds of anything. Have you ever heard of "meteor insurance"? Or "animal attack" insurance? If I wanted to, I could buy "insurance against an alligator biting off my left index finger", and these guys can actually estimate a good "p" value for that happening.

(3) Sadly, actuarial statistics is not an exact science, which makes our job even harder. We would have to witness a very large effect to be able to have faith in what we are witnessing. Critics of our experiment would always be able to say that we rigged our estimates of "p" so small that expected results seem unusual. The ideal dream would be something like "lottery numbers" where the odds are very simple to compute.

In short, I continue to maintain that my methodology is perfectly sound, but in practice would require a lot of time and money to actually do right. And no, I have no interest in actually doing it, but I find it interesting to discuss.

I'd be willing to cede that 2 and 3 are just really difficult, but I still say difficulty number 1 is impossible.

I don't think anyone can anticipate all the possible near misses that would go into the umbrella. If the umbrella fails to include all possible near misses (which I think are equally legitimate--or rather equally not), then it's just an arbitrary selection.

The classic example of shoehorning, for me, is when Gary Schwartz took Allison Dubois' message from a departed friend, "I don't walk alone" to mean that his friend was confined to a wheelchair in her later years. You could probably ask 100 people to write down what they thought "I don't walk alone" would mean in this context, and I doubt that a single person would write down that it means "I used a wheelchair before I died".

We brainstormed a little in this thread on what other interpretations you could give to a dream about hitting a deer while driving your car, but I'm sure there are MANY more possibilities.

And I still say a precognitive dream ought to give you foreknowledge of future events. If it only seems to match an event after the fact, it's not really foreknowledge.

But I do get what you're saying, Unalienable, that blurry impressions or approximations might be what the claim is.

In that case, I simply don't give it any merit. If the claimed paranormal ability boils down to saying "something important might happen soon but I won't know what it will be until after it happens", it's pretty much a waste of time.

Bri

11th October 2007, 06:53 AM

That doesn't sound quite like what Unalienable is proposing, but the problem with this one is that it's still a fishing trip. If you don't state a clear hypothesis before you start, it's not kosher.

The hypothesis is that using the metrics, a "predictive" dream will score above a 0.9 compared to the "best fit" event that occurs within the given time frame. Pretty simple, really. The metrics are subjective, of course, similar to the Drake Equation (http://en.wikipedia.org/wiki/Drake_equation), and are useful only in helping to determine how impressive the prediction is compared to the predicted event when the two aren't identical. I think that I'd actually add a third metric for specificity to prevent a very vague prediction from achieving a high score:

S, specificity of the prediction (0=vague, 1=specific)
P, probability of the actual event occurring by chance within the time limit (0=high, 1=low)
C, closeness of the actual event to the predicted event (0=distant, 1=close)

score = S * P * C

You're saying whatever event happens that is "closest" to the prediction (and I still don't get how you make that assessment) becomes the hypothesis, and all that's left is to calculate the odds of it happening times the "closeness to prediction" quotient. It's still an after-the-fact calculation, and you're still ignoring the probabilities of all the other possible events that didn't happen that could have also counted--in fact, there's an infinity of them since there's no such thing as a miss in this set up.

Note that the first two metrics can be calculated ahead of time, and will limit the highest score possible. So most of the time we don't even have to wait for the event to occur before we can show that the person was "hedging" the prediction (not necessarily consciously) by either making it vague or making it fit events that are likely to occur anyway. So, it's a useful exercise just for that reason.

And after the event has occurred, we don't care about what could have happened -- we only care about what did happen. At that point, we want to know why the person thinks the prediction was successful, whether it really was successful, and if so how successful it was.

I don't know of anyone who has had difficulty identifying the event that was predicted by their premonition. In fact, I think Unalienable said that he "just knows" when the event occurs that was predicted by the dream. But even if he didn't, if the dream was even in the least bit predictive, it should be relatively easy to pinpoint a single event that it predicted (otherwise the score will be too low from the first two metrics to matter anyway). If there happens to be more than one event predicted by the premonition (I've never heard of a psychic saying that two different events were equally predicted by the same premonition, but I suppose it could happen) then you would calculate both and use the highest score.

The idea here is that when someone says "See, my dream came true!" you can break it down and see just how predictive the dream was compared to the actual event that came true. The closer the score to 0.5, the more likely that the dream couldn't have predicted the event at all -- 0.5 means that the event could just as easily have been "predicted" by a random guess. A score of 0.9 would be pretty impressive -- it would be an event that was not only very close to the dream, but that also wasn't likely to have occurred by chance.

I suppose I could be persuaded to this if you could tell me how to assess closeness to the prediction (of events not predicted) and how you could calculate those ahead of time (which is what Unalienable is suggesting).

I'm not sure that you can calculate "closeness" with any accuracy since it's entirely subjective, but the exercise is still useful. In reality, it's usually the other factors (vagueness and probability of the event occurring) that are being used to consciously or subconsciously "hedge" a prediction most of the time anyway, since those are the two that can be assessed before the event occurs. I'm not sure that Unalienable was even aware that he was using those two factors to essentially guarantee that his prediction would come true when saying that slamming his hand in a door would count as a hit. Just by calculating the odds of Unalienable (or someone close to him) slamming his hand in the door and comparing it to even a vague notion of "closeness" to the original dream, one can see that the dream could not very well predict that event much better than any random guess might no matter how "close" one decided the event is to the dream.

-Bri

JoeTheJuggler

11th October 2007, 09:49 AM

The metrics are subjective, of course, similar to the Drake Equation (http://en.wikipedia.org/wiki/Drake_equation), and are useful only in helping to determine how impressive the prediction is compared to the predicted event when the two aren't identical.

I don't think ANY of the factors of the Drake Equation are subjective--just unknown. There's a big difference. How you estimate an unknown quantity is subjective, but the things they're measuring is not. We just don't have the information.

Note that the first two metrics can be calculated ahead of time, and will limit the highest score possible.
So you think every possible near-miss can be known ahead of time? I don't think it's possible. And if you don't have all of them, then the definition and probabilities are arbitrary.

And after the event has occurred, we don't care about what could have happened -- we only care about what did happen. At that point, we want to know why the person thinks the prediction was successful, whether it really was successful, and if so how successful it was.

Again, if you don't count the probabilities of all the other events that you might have counted as hits (if they'd happened), you're basically creating a whole bunch of hypotheses and then choosing only the best one to test after the fact.

That's what I call a "fishing trip". It's also called data dredging. If you don't state the hypothesis you will test before the test, you can sift through data to find "evidence" of almost anything. I understand you're not purely doing that (by tying all the various near misses to a single prediction), but it's the same in practice. You still don't know which probability you're testing until after the fact. So this method depends entirely on the validity of the "closeness" measure.

I don't know of anyone who has had difficulty identifying the event that was predicted by their premonition. In fact, I think Unalienable said that he "just knows" when the event occurs that was predicted by the dream.

Again, if he "just knows" AFTER the event occurs, then the dream was not a prediction.

I'm not sure that you can calculate "closeness" with any accuracy since it's entirely subjective, but the exercise is still useful.

I disagree. I don't think it's just a matter of subjectivity. The problem I see is that we're trying to compare misses. In only a few circumstances does comparing misses makes sense. (Submit several guesses of my height, and if they're both wrong, we can compare which is closer.) Comparing the miss, "I hit an elk," and the miss, "My brother hit a deer," doesn't make any sense. Neither one is "I hit a deer", and one is not "closer" than the other. I say neither one is the fulfillment of the prediction "I will hit a deer". Knowing the prediction "I will hit a deer" doesn't favor one or the other the way it favors what it really predicts, "I hit a deer." Ahead of time, the prediction doesn't give you any idea which of these two would be more likely (because of the prediction, not because of real probabilities--I'm still talking about your "C" factor--I'm willing to accept the "P" factor as something difficult but knowable to some range of confidence).

Bri

11th October 2007, 03:00 PM

So you think every possible near-miss can be known ahead of time? I don't think it's possible. And if you don't have all of them, then the definition and probabilities are arbitrary.

You don't have to know every possible near-miss. You only have to know the single event that the person who had the premonition claims is the event being predicted. You can then calculate how "impressive" it would be that the premonition predicted that particular event by looking at how rare the event is and also how close the event is to the prediction.

Again, if you don't count the probabilities of all the other events that you might have counted as hits (if they'd happened), you're basically creating a whole bunch of hypotheses and then choosing only the best one to test after the fact.

I'm not sure what the probabilities of other events that didn't happen have anything to do with it. For any prediction, there are going to be lots of possible events that could have happened, but didn't. Some will be rare and some will be likely to occur (even though they didn't). Some will be closer to the actual prediction and others will not be. I don't think events that don't happen tell you anything at all about the premonition. The only event that tells you if the premonition is interesting is the event that actually occurred that is claimed to have been predicted by the premonition.

That's what I call a "fishing trip". It's also called data dredging. If you don't state the hypothesis you will test before the test, you can sift through data to find "evidence" of almost anything.

The hypothesis is that using the metrics, a "predictive" dream will score above a 0.9 compared to the "best fit" event th