Pages Navigation Menu

Parapsychology articles and news

Psi Experiments website announces the launch of its second psi experiment

After successful launch of the psi experiments website I’m proud to announce the launch of the second experiment.


Over 1500 people have participated in the first experiment and the results will be published during the coming two weeks, after the analysis will be complete. Yet, following constructive criticism regarding some design flaws of the first experiment, this second experiment is better designed to overcome these flaws and provide more reliable results.

Everyone is invited to participate in this second psi experiment and spread the word among your friends. The more participants, the more reliable will be the results.

I’m also glad to add a new sponsor on board. This new sponsor is Immrama Institute and they will provide their Insight CDs as prizes. Insight CD
is a powerful tool to help you explore expanded states of meditation and consciousness, achieve deeper levels of self awareness, reduce stress, enhance creativity, and improve your overall mental and emotional well-being.

To those participating, if you’d like to learn some ESP I can advice you to read my articles:

Developing ESP – First steps
Exercises to develop ESP ability – part 2

For those interested to learn more about psychic abilities I can refer to the course Developing psychic powers, which, for a reasonable price, includes three books with over 500 pages of material to develop your psychic abilities.

   

 

 

14 Comments

  1. I think the Insight Cd is a very good thing to have beacause it can
    help you explore the expanded states of meditation and consciousness
    and it can help you achieve deeper levels of self awareness,reduce
    stress,and more.

  2. Thanks for replying.

    Target per trial randomisation is a must I’m afraid. Even a miniscule bias can appear to be significant psi if enough trials are done, and most effects in parapsychology are small.
    I get your point about making this test ecologically valid by concentrating on one target, and your efforts are laudable, but the first step should be proof by principle and that means randomisation per target.

    I would run this experiment to begin with, and then add in the randomisation later on, or even better, have a mixture of randomised / non-randomised trials under a blinded protocol, and look to see if biases actually influence any “psi” effect in the randomised condition.

    This could be a clever experiment investigating both psi and any interaction with peoples natural biases.

    Good luck!

    Michael.

  3. This is extremely poor science.

    The crtiticisms of the previous experiment have not been acted upon. Random selection of the target per trial is essential, there is no getting around it.
    Why is this important? As mentioned earlier, peoples choices are not random. If the target selected happens to be a target that people have a natural bias towards choosing, then that is going to look like psi, when it is not.

    It is sad that the previous comments have not been acted upon. For that reason I am not going to bother with this site anymore.

    Thank you.
    Michael.

    • It’s quite easy to do random selection of targets, technically, but then this information would only be an entry in the site’s database. This would be like a really “light” signal for people to lock on. I try to make more of a “real” world situations.

      If this is a total must, then I will run an experiment like this, maybe even with the same products. What do you say, that in 2 weeks, I’ll restart the experiment, where this time also the target will be selected at random, no envelope, nothing, except some bytes in the computer. Then it will be possible to compare how biased people are.

      I understand that there’s a problem of possible personal preferences to a specific office product but hopefully, this time it’s much better, since they are all kind of “neutral”, so I expect the results of this experiment to be really closer to chance than the previous one, where there were too strong cues to bias a preference upon, like colors and position.

      If the approach I use here is problematic and you say that people in the end only select things based upon the image or personal preference, then there’s a problem with psi in the real world as well.

      Real psychics, if they exist, should be able to distinguish real “signal” from ther subjective personal preference. For example, if you come to a psychic asking him for a location of a lost object and give him several possibilities, I’d expect for him to be able to overcome his subconscious preferences and only lock on to “real” information.

      That being said, the participants of this site is general population and is not controlled, so clear results can not be expected at all. Yet, many people find the experiments interesting.

      I can also discuss a future experiment design before publishing it, if you would like so, to make it better before its release.

      • Ah, Jacob, you have touched on the biggest problem “real” psychics have in producing useful real-world information. We have to be able to filter out our own biases in order to obtain accurate results, but it is extremely difficult – in fact, it is the single most difficult part of the process (in my experience and that of others I know). Every piece of information we all receive through all of our senses has to be filtered for relevancy. For instance, just separating out relevant verbal information in a normal environment full of auditory input is a complex process, and it can be impossible or nearly so for those with any of a variety of physical and/or psychological conditions from poor hearing to ADD to PTSD to Autism.

        In attempting to find that relevant “needle” of psi data in the massive “haystack” resulting from constant bombardment of all the senses is not just an “oh well, anybody who does this *should* be able to,” it is *the* task. It is the very essence of the work. The understanding I have come to after decades of studying the process in myself and others is that the very nature of a “psi” experience is to be able to find (capture? notice?) that particular little piece of data, isolate it from the virtual storm of data that we all are constantly receiving.

        That is why naturally occurring instances most often come when one is in an extraordinary situation of intense experience and/or deep emotion and therefore in a state of extreme focus. Those of us to whom these experiences occur with more regularity seem to have a more highly developed ability to focus, to filter for that particular little piece of data. In order for me to more consistently find the needle in question, along with enough context to make it meaningful and useful in, for instance, a missing persons case, I was further trained – similar to athletic training – to focus, and to learn how to tell what is coming from my mind and what is coming from elsewhere. Have you ever seen a belly dancer roll a line of quarters across her tummy? I’ve tried to do it, and I haven’t the foggiest idea how to isolate the particular fine muscle movements that feat requires. Moreover, once she finds those muscles, she has to practice until she can find and control them consistently. I think what I do is a lot like rolling those quarters.

        I therefore concur with the others who are stressing the importance of using the most neutral targets possible in testing, so the subjects don’t just choose the targets that most easily catch their attention: the most likely result in an experiment with large numbers of subjects of varying levels of experience and ability. There is a catch-22 in this, of course. With less-neutral targets you are more likely to get “good” results, but the results will not be the result of what you are testing for. Since I believe good results come from experience and practice in learning to tell the real target from the flotsom, I don’t know how a large mixed population will perform – I would predict the results would be rather bad. Let me stress I am not saying that is because it is a gift given to the chosen, but because of the inexperience of the majority in filtering out the irrelevant and focusing on the relevant. I don’t want to sound negative, in fact I would love to see more experiments like these, and to see them succeed. Good luck!

  4. I thought I and others were clear and constructive in our explanations of the main defect in the previous experiment. You seemed to acknowledge the problem, yet here’s the same mistake again.

    I don’t understand this:

    I decided not to select the target photograph each time by computer since I’ve also done the following:
    The photograph which was randomly selected out of the four, was put in a sealed envelope.

    The reason to select a target for each trial is to make the trials independent, so statistical methods can determine the significance of the result. I, and probably others, can help you with experimental design and computer implementation.

    Oh, there’s also a cut-and-paste oversight:

    The string in the next paragraph is the result of one-way hashing of an answer string, which describes in which box is the ring.

    That’s obviously from the previous experiment.

    • Hello, Bryan.

      The reason why there’s only one photograph selected as correct answer is because I wanted to have some real-world “evidence” to the answer. I have a hypothesis that psi might work better with real objects than with selection registered only as some byte in computer memory, which is much less materialistic. Also remote viewing requires a target, a real one.

      My understanding (and correct me if I’m wrong) is that since I shuffle the locations of the displayed images in the form using a true random process, this will eliminate the location phychological bias or preferences, like for experiement 1, where people would pick specific locations more since they are numbered. And I had no way to shuffly the location of the ring or shuffle the photo in experiment 1.

      In this experiment, since I record the location data, if there’s a bias towards specific location, I will know it, and I’ll be able to use that data in final analyses. For example, if people would choose location 1 (left most) 25% of the time, I will know that. But, since the real answer was randomly ditributed between all 4 locations, this would not so much make a difference to me. Again, correct me if I made a mistake here. I’ll be glad to use more help with statistical analysis.

      Regarding answer hashing. Yes, the explanation text is copied from experiment 1 but the hashed value is different since the anwer string is different. Take a good look at it.

      • We have no quantitative measure of how biased or correlated peoples’ choices are. No matter what he outcome, I do not see how you could combine the trials to the trials into a statistically significant result.

        Should you restart the experiment? I think that once you’ve started it and said that it will run a certain way for a certain time, you’d be open to more criticism if didn’t follow through. A better experiment would be nice, but I see no great rush.

        And yes, I noticed that the SHA-1 hash is different this time. Note that it only prevents one form of cheating: changing the target after the fact. There are other ways to cheat, such as leaking the answer or miscounting the votes. Fortunately there’s no reason to believe you are anything but sincere.

        The criticism so far has only covered this one major defect. There are a number of issues to consider, and several lessons to be learned from other on-line PSI experiments. You might want to discuss your ideas here or on sci.skeptic, as a few of us have spent some time studying such things.

        • I’d be glad to discuss it here.
          Could you recommend me a way to :
          1. Have random targets
          2. Have some more real world, longer lasting equivalent to the answer than some bits in computer memory.

          I wanted for the 3rd experiment to use the following setup. Tell me your thoughts on it. The card referred can be the famous Zener cards:

          1. Ask the subscribed people to go to the website, and select one of the five cards. Their selection will be recorded in the database with their identification (database ID). I will also ask each participant to either print or draw their selection and leave it in their home with the number of their ID written on it. This will create a real-world target.

          2. A week or two later, I’ll sent each subscribed participant a list of 5 ID numbers and will ask them to guess which card these participants have selected. The list will be created randomly from the entered selection in step 1. The participants will enter their 5 guesses.

          In this experiment, although there can be a strong psychological bias towards selecting specific cards, since I record that information, I can use the statistics of step 1 to normalize the statistics of step 2. For example, if in step 1 people select specific card at 30%, than guessing it at 30% would be considered exactly chance, even though there were 5 cards.

          Anyway, this can be a very interesting psychological study since often these cards are used in testing, and as such, the biases of people should be known a-priori to the guessing.

          What do you think, Bryan and Michael?

        • Some theories of PSI phenomena do not depend on a physical target. One theory is that a fundamental mechanism if PSI is information channel from the future to the present; predicting an outcome to be shown later should work whether or not there is a physical target when the guess is made.

          If you do want physical targets, you could automatically print some, perhaps bounding the number a thousand or so to stay within budget. You could also test whether printing the target effects success rates.

          If I understand your proposed Zener-card experiment, the sender will know his or her card and ID, and receiver will know the ID. That’s vulnerable to cheating. Also, given the normalization method, have you figured how you would combine the results with people hitting and missing the different probabilities?

          A lesson from previous on-line work is that people are likely to drop out before completing the trial. Randin notes (and over-states) the problem his preliminary analysis of his on-line experiments. You just need to ready for it, and determine in advance how to score partial submissions.

        • I hope I’m not being too discouraging. Good psi experiments are tough to design.

          This design is fairly well thought out, but I’m afraid its not quite there.

          Biases are context dependent — you cannot depend on people selecting “targets” in the first part having the same biases as people selecting “responses” in the second part. Generally though, shared biases with the Zener symbols are pretty weak (though some individuals showed strong personal biases).

          I wouldn’t use Zener symbols though — they really make lousy targets from a psi-elicitation viewpoint (generally attributed to their extreme boring quality). Say you get a response close to the highest sustained zener scores in history [Synchronicity Digression -- I was just trying to remember the average score of the individual with what is probably the highest sustained Zener based score, known as "CJ" from the Colorado study (I did some analysis of his score-sheets at one time). So I had a question about CJ in mind -- my eye glances at the captcha and it is: CJQ]. I can’t quite remember but it was something like 6/25 instead of 5/25 — a shade less than that, I think, but use the simple number. Then you will need about 400 responses to expect to be just significant at the .05 level.

          Your best bet is to get a fairly large set of fairly interesting digital photos. Each “target holder” gets one randomly chosen photo from the total set which they should print out and put in an envelope. Here is an idea to help provide a psychological focus: they should also print out a unique symbol — say a common dictionary words (selected from a set that is not to similar) inside a distinctive border and enclose that in the same envelope.

          Before seeing the target pool your “percipients” are given the image associated with their target. They should then try to get any impressions that they can. When they are ready, present their target and three other randomly selected members of the complete pool, in a random order. They pick the one that most closely matches their impressions.

          I know that’s a bunch of work, but as long as you are careful about your random numbers (hint — do not use a prepackaged pseudo-random number generator unless you REALLY know what you are doing) it should be rigorous and includes some of the elements that seem to produce larger effect sizes.

        • Thanks, Topher.

          I’ll take the notice of Zener cards.

          But the problem with digital photos, in my opinion, is that not everyone can print from their computer. I’m afraid that’s just too much work for most participants.

          I didn’t understand what the words from the dictionary for, though.

        • No one has unambiguously, convincingly demonstrated any strong effect of physical factors such as target size, distance or physical (as opposed to virtual) existence. Psychological factors, are another thing, however — what people “feel” like they can accomplish. That’s why I don’t dismiss your “physical existence” design criteria. If you feel its needed it cannot be dismissed.

          With multiple physical targets out there, there may be a feeling from percipients that they don’t know what target they are trying to get. You can’t identify the person holding the target — both because of a loss of confidentiallity and because it opens up the possibility for collusion. That leaves telling people to identify the target that’s “In the envelope with this symbol” where “this symbol” is unique.

          I don’t think that it is physically necessary — psi (by which I mean whatever it is that causes small but consistent anomalies in parapyschologists experiments) seems to work as a whole, identifying which target as well as its contents is no harder than just identifying the contents. Psychologically, though, is another matter, especially with mostly inexperieced participants.

          You are right, random people [rinting out pictures (at least color ones) could be a problem. That would suggest that a single person (ideally not you — someone you have limited contact with) preparing all the targets.

      • I should give my qualifications — I’m a computer scientist who has been involved with parapsychology for somewhere around 35 years. My specific area of expertise is statistical analysis of psi experiments and statistical computing. I am, for example, a member of the Parapsychological Association and I’m one of the authors of the old Psi FAQ found in many places on the web.

        You have eliminated some biases by using objects you judge to be similar and neutral, by using black and white photos and by rearranging the order. Its not good enough I’m afraid:

        1) If people can perceive a difference then they will have preferences, whether they are aware of it or not. Three of the four target objects could cause harm if used almost but not quite properly, the fourth could be seen as providing control. Is that going to cause a significant bias? I don’t know, and neither do you. We can play the “One of these things…” games in different ways and get different answers and some of them will unconsciously cause some effect.

        2) Maybe you have eliminated all the systematic bias, but for the experiment to count as evidence you have to be able to prove that you have done so. The only way you can eliminate biases that you don’t even know about is to randomize all perceptable aspects of the experiment.

        Imagine that I show a 100 people cards with the numbers 1-10 arranged in various orders and ask each to pick one. There will be some positional biases which will cancel out, but you’ll probably still get as many as 30% of a group of people with a Western European (including “American”) background (I’m unsure about other groups) choosing “7” — and what can be more neutral than the purely abstract entities numbers?