cooper@pbsvax.DEC (Topher Cooper HLO2-3/M08 DTN225-5819) (10/30/85)
decwrl!dec-rhea!dec-kirk!williams asks: >About these psi tests: How is the input stimuli randomized? >I hope you are *NOT* going to tell me that the experimenter >*CHOOSES* something at " random ". > > John. Good, an intelligent question. Your hopes are fulfilled: I will not tell you that the experimenter chooses the target at "random". Such a procedure is obviously completely unacceptable. Many techniques have been used, I'll describe briefly the major ones (roughly in order of their widespread adoption): Card Shuffling: A deck of cards is shuffled using multiple riffle shuffles or overhand shuffles. Most frequently the deck is a Zener deck (sometimes called the Rhine deck or the ESP deck) which consists of 25 cards, five each of five standard symbols (star, square, circle, wavy lines, and triangle). This is called a "closed" deck. An "open" deck is created by shuffling five closed decks together then dealing off 25 cards. In an open deck the probability of any particular card being any particular symbol is still 1/5 but that symbol could occur anywhere from zero to 25 times in the deck. Open decks are used when the subject receives immediate feedback after each call. Standard playing decks, and various specialized decks (e.g., "clock" cards, and animal picture cards) have also been used. The shuffling might be done by hand or by machine. Random Number Tables: tables such as the Rand Table of random digits have been used quite a bit. Position in the table may either be sequential usage (i.e., one experiment starts where the last left off) or by using dice and a complex mathematical transformation (the transformation was supposed to prevent PK from being used unconsciously by the subject or experimenter to select a position which would correspond to the subjects arbitrary calls; the idea was that it was supposed to be an ESP not a PK experiment). The resulting numbers would be used to arrange the targets, whether a Zener deck or some other target type. Various Mechanical RNGs: There have been a good number of these, some good some poor. I won't go into details, no single mechanical RNG has gained widespread use for any amount of time. Pseudorandom RNGs: PRNG's are considered adequate for much work, if they have been properly checked for randomness and if circumstances allow adequate security. They have been used both in the same way as tables, and directly in interactive computer tests. They cannot be used in precognition tests. Subject Action Random Bit Generators (RBG's): A flip-flop is oscillated between two states (or a tight program loop simulates the same) until the subject takes an action (hits a button). The flip-flop is stopped and its state represents the random bit. The button hit may represent the subjects guess or the bit might be used in other ways. Since the period of the oscillator is much shorter than the subject's reaction time, and even shorter than the variance of the nervous system (i.e., even "precisely" the same action would cause different results because of variation of response time of neurons) this can be considered random. A variation uses a full counter instead of the flip-flop and uses the result to randomly seed a PRNG. In case you don't recognize the description, one example of this is using a time-of-day clock to seed a PRNG. Radioactive Decay RBG (Schmidt RBG): Similar to the previous except a beta source is placed near a Geiger counter and its the firing of the Geiger counter which stops the oscillator. This has been used quite a bit in recent years both for ESP and PK (yes folks, people seem to be able to bias radioactive decay, or at least the behavior of instruments which measure it). The oscillator frequency is usually in the MHz range, while the average period of the beta-source/Geiger counter is about .01 seconds. Electronic Noise RBG: The same again but noise from a back-biased diode is amplified and the oscillator is stopped when it crosses some threshold. Proper use of both this and the Schmidt RBG require frequent control runs to make sure that no bias has crept in. Further questions? Topher Cooper USENET: ...{allegra,decvax,ihnp4,ucbvax}!decwrl!dec-rhea!dec-pbsvax!cooper ARPA/CSNET: cooper%pbsvax.DEC@decwrl Disclaimer: This contains my own opinions, and I am solely responsible for them.