Randomness and the mind: looking for psychic effects in games of chance. Helmut Schmidt.
Randomness and the Mind
Looking for Psychic Effects in Games of Chance
Does the outcome of a game of chance depend on the player's mental effort? Certainly effort can help where skill and strategy count. But what about pure chance games like roulette, or those computer games in which we can do nothing but wish for success?
With these questions we strike the core of an ongoing controversy about the existence of some "psychic' effects that may not quite fit into our present worldview.
The first laboratory reports that gamblers and other people might mentally affect the outcome of dice falls were published in 1942 by L. and J.B. Rhine who labelled this the "psychokinetic effect' (PK).
These early claims were, rightly or wrongly, much criticized. The effects were rather small. Even the best performers scored only slightly better than chance would dictate. Statistical evaluation methods were required to detect the effect--like a weak signal in much noise. The critics wondered in particular whether the manual recording of thousands of dice throws might have left room for systematic recording errors.
The introduction of electronic test equipment has revitalized this kind of research. We no longer have to worry about recording errors or the monotony of dice throwing. Any home computer can be programmed to act as electronic dice tumbler or coinflipper to tally the scores and to present the participant with stimulating feedback.
Using such test equipment, several research groups have reported weak, but, nevertheless, significant PK effects. And this type of work is being continued in physics and engineering departments of universities, industrial research laboratories, and private research foundations.
But with the reported effects rather weak and at odds with our conventional thinking patterns--with what we call common sense--many critics still see room for doubt in the reality of psychic effects. Could it be, perhaps, that the researchers overlooked something, that they made some blunder, or that they even have an urge to deceive themselves or the public?
This psychic controversy is carried out in the home territory of computer enthusiasts. And we certainly want to know whether or not mental effort matters in games of chance.
After some experimenting, I am no longer neutral in the controversy. But rather than discuss my own results, I want to show you how you can try to catch PK in action. If a large number of experimenters were able to observe the effect, the question about its existence would be settled once and for all.
Experimenting in this field can be great fun. We can design and play challenging games, and at the same time do fundamental research. This may be one of the few areas of science in which the newcomer can still contribute to progress.
From Dice to Radioactive Decay
The early researchers saw psychokinesis as a mental "force,' comparable to the known electric or magnetic forces. Surprisingly, however, the participants in their experiments seemed able only to affect rolling dice; they could not exert a measurable force on a die placed on a sensitive scale or suspended as a pendulum on a string.
What makes a rolling die different is the element of randomness that enters into the outcome. And today it appears that PK might be able to act wherever truly random processes are in progress.
The idea that part of nature may be governed by pure chance, was one of the more sensational outcomes of modern physics. The textbook example of a pure chance process is radioactive decay. If I have a sample with many atoms of radioactive Strontium 90, then I can register the individual atomic breakups as clicks on my Geiger counter, and I can calculate that a Strontium atom has an average lifetime of 30 years. But according to quantum therory it is impossible to predict when a particular atom will decay. There seems no mechanism inside the atom that determines the decay in a computerlike (deterministic) manner. The decay is rather the result of pure chance.
The first successful attempt to mentally affect radioactive decay was reported by R. Chauvin in Paris. In this experiment children tried to speed up or slow down the counting rate of a Geiger tube during one minute time intervals. This appeared to work more ofted than not. Two 13-year-old boys were particularly successful as long as the experimenter could keep them sufficiently motivated and excited about the experiment.
Chauvin's neat experimental setup has some minor practical disadvantages: the base counting rate must be re-calibrated frequently, the power supply must be very stable, and the experimenter must make certain that a fraudulent participant does not introduce his own radioactive source (such as a radioactive watch dial) to affect the counting rate.
Therefore, in my own experiments, I used the randomness of radioactive decay slightly differently, as basis for a digital random number generator (RNG). I then had the participants try to affect the output of the RNG.
The basic idea behind the RNG experiment is very simple: you have a very fast and regularly advancing binary counter which you stop at the random time at which the Geiger tube registers the next signal. Then the lower digits of the counter stop in truly random positions, provided that the average waiting time for the next signal is long enough that these digits have been turned over many times.
A True Random Number Generator
To implemint the RNG with a computer, you can use the external circuit of Figure 1a which connects to an input port Q of the computer. The flipflop in Figure 1a changes state with every decay signal, i.e., at random intervals. A machine language program loop can provide the needed fast counter. Figure 2 shows a corresponding flow chart.
But remember that your computer may steal computer time for the direct memory access of the screen display. In this case you must disable the interrupts to guarantee the smooth progress of the counter while you are waiting for the next random number. If you don't want the screen to flicker for each random number request, you must generate and store all needed random numbers at the start of the game. You could also avoid screen blanking by including a fast hardware counter in the external circuitry.
With the arrangement shown in Figures 1a and 2, you get truly random numbers. Even if you produce millions of them, no computer should be able to detect any systematic pattern in the sequence.
I had great fun and no difficulties in setting up the Geiger tube circuit. But if you don't need too many random numbers at once, there is an easier way. You can use the randomness in the timing of a manual buttonpush. Instead of the Geiger tube setup in Figure 1a, you connect the electronically clean switch of Figure 1b to the computer input. For an even simpler setup, you can let a computer key take over the role of the external switch in Figure 1b as well.
From the time that your brain decides to press the key, as the nerves conduct the signal and the muscles execute the command, so many truly random processes in your body have entered in that the timing is (within certain limits) truly random. With a fast machine program loop, you should ger perfect randomness.
For preliminary experiments you might be somewhat sloppy and create the counter with a Basic loop, particularly if you need only a binary decision (even or odd counter reading). You might also find a memory location that is automatically incremented with each screen refresh, and use that as counter.
It is possible to use electronic noise as a source of randomness. The basic idea is again very simple. Feed white noise into a pulse counter so that the counter advances rapidly at random intervals. Whenever you need a random number, read the counter. In this manner you can generate random numbers at very high rates. But the setup requires more external circuitry and much care in design. In particular, you must filter the noise so that no marginally weak signals can reach the counter.
The Size of PK Effects
The PK effects reported in rigorous experiments have been rather small. That may be the nature of the phenomenon, or it may be that test conditions tend to inhibit an effect that depends much on enthusiasm, confidence, and emotional involvement.
Take as the first example an experiment that you could easily set up for yourself. When a key is pressed, the computer makes a binary random decision for a "head' or "tail,' and displays the result as a red or green signal respectively. You choose your favorite color, say, red. And then you wish for red to appear whenever you press the key. You might do this in a very relaxed atmosphere, taking much time to visualize the target color. You could also use a more aggressive approach, fighting for each trial. In either case you should start a test session only when you feel in top shape, and keep the sessions short (perhaps 20 trials per session). Stop whenever you want. The computer tallies the hits and misses (red and green signals) and at the end of each session you record the scores, manually, or automatically on disk. If you like the setup and the results look promising, then you might go on and try to accumulate "statistically significant' results (see Table 1).
When I tried this experiment, I used two colored lamps as a display. The lamps were in sockets close to the participant; the RNG, computer, and recorder were in another room. It was easy to verify that the RNG alone, running by itself in the absence of onlookers, showed no bias. But as an additional safeguard, I used in the PK test each of the two sockets (connected to "heads' and "tails') equally often as the target side. The participant simply inserted his favorite color lamp into the target socket. In this way, even a constant generator bias could not simulate a PK effect.
I did not seem able to produce PK effects myself, but I found several promising volunteers who enjoyed the experiment. When we felt sufficiently confident, I began a "real experiment' with a total of 3000 trials. The 35 participants contributed different amounts, depending on their availability and interest.
We obtained an average success rate of 54.5% (where 50% is the chance expectancy). With the large number of 3000 trials the result is statistically significant (Table 1), i.e., the outcome cannot be reasonably explained by pure chance.
This was a very slow experiment, with an average trial rate of perhaps one every 15 seconds. With higher trial rates one might hope to gather data more efficiently.
To study an extreme case, I used random noise to obtain binary decisions at a rate of 300 per second. This time, the heads and tails were clicks in a right or left headphone. That way, you cannot quite distinguish the individual clicks. But you still hear statistical intensity fluctuations. And for obtaining more heads than tails you can aim at more action in the right headphone. Since this takes much concentration, I used only short runs of 1000 trials (about three seconds).
In a total of N=200,000 trials (200 runs), we got a success rate of only 50.39%. But the result was still significant (Table 1).
The decline of the scoring rate at very high speeds is certainly disappointing, but not surprising. There was really not enough time to focus on each individual event.
Right and Wrong Mental Attitude
If PK is real, one wonders what practical role it may play in our lives. With the effect rather weak, nature may have decided to use it only sparingly. To succeed in life, even I would rather trust my powers of sound reasoning than my PK powers.
But you don't have to be a professional gambler or a salesman to realize that there are some situations in which we can do nothing but wish for a good outcome. And if PK could help us to increase our chance of success by only a few percent, it might be worth the effort.
We can simulate real life situations with games and study how chance may be affected by the right or wrong mental attitude.
The results of the following simple game may surprise you. A player advances in single steps by pressing a key. With each step, the computer obtains a true random number in the range from 1 to 4. For a 1, 2 or 3 you hear a pleasant low tone, you have made a "safe' step. If, however, the random decision is for a 4, you hear an unpleasant loud beep or explosive sound. The player's goal is to take as many safe steps as possible.
Try first a positive approach. Spend some time putting yourself in a calm, confident mood. (Try to recall and savor a situation in which you were particularly successful.) Then start the game. After a false step, stop for a moment to regain your confidence, and stop the game for the day whenever you feel apprehension or fear of a false step welling up.
Next explore the effect of a negative mental attitude. Imagine vividly how scared you are of a false step (Think about stepping on a bomb, getting an electric shock, or feeling the dentist's drill hitting a nerve). How are you scoring now? Can the negative attitude work against you, so that more than 25% of your steps are bad?
Fortunately, being scared can be fun (that's why we watch Hitchcock movies), so experimenting with your friends or with groups of children can be quite enjoyable.
Quasi Random Generators
In slow games, like the last one, the timing of key presses can provide all the randomness we need. For fast action with many random decisions, an external random source with radioactive decay or electronic noise comes in handy. But can we perhaps use the Basic RND function instead?
This function is based on a quasi-random algorithm. Starting from a seed number, the algorithm derives a sequence of numbers that appear to be random. Nevertheless, the whole sequence is pre-determined by the seed number.
Listing 1 contains its own quasi-random algorithm. After a seed number is entered, the program prints HEAD or TAIL in quasi random sequence, and after 100 trials the score is tallied. For a simple PK test you might display the heads and tails as high and low tones. Note that using the same seed number twice, leads to the same sequence and the same score. And no PK effort can change that.
We might give the PK mechanism a chance to act by selecting the seed number randomly, by the timing of an initial key push. Then the outcome would be unpredictable and you would almost always get different sequences of heads and tails. But with the seed number and the outcome of the game fixed at the start, you would feel rather foolish making a mental effort during the game. You wouldn't want to assume that such effort could retroactively aid in the selection of a favorable seed number. But try it any-way, particularly with participants who do not know the details.
There is another way to bring the psychic element into play more directly. Use the quasi-random generator to supply a sequence of heads and tails, but provide an "option' button such that pressing the button inverts the decision of the quasi-random generator.
Then, starting from the same seed number, you can get different game histories, depending on when you press the option button. Working always with the same seed number, you might soon learn at which times in the game to press the button to succeed. This game would test your memory and your skill.
To make it a clean psychic game, you must start each time with a truly random seed number. Since you don't know the seed number and the subsequent quasi-random sequence, there is no systematic strategy for success. If you show in the long run a significant winning tendency, it can be due only to some psychic mechanism.
You could call the mechanism in this case precognition. You might foresee when the quasi-random generator is going to work against you, and then hold the option button down to invert the decision.
In the laboratory, precognition and PK appear closely related, and it is often impossible to distinguish one from the other. They are probably based on some common "psi principle,' in which psi includes PK, precognition, and the other forms of ESP.
Design of Psychic Games
Should we modify available games or design new ones for the study of psi effects?
Some games may need little change. When the player has skillfully guided his missile toward the enemy ship, you can let a truly random cointoss determine whether the ship blows up or the missile is repelled.
In the end, you display next to the total score (resulting from skill, strategy, and chance), the psi score that reflects the player's "luck' in the game, i.e., his success with the chance decisions.
Challenging the player to display not only his skill and cleverness, but also his psychic powers may add a new dimension to the game. Even though the effects are small on the average, the correlation between your momentary mental state and your psi score may be striking.
Most of the previous work has been done with very simple games in which skill and strategy play no part and in which we have only one thing upon which to concentrate. (A collection of ten such games for the Atari is available through the Mind Science Foundation, 102 W. Rector #215, San Antonio, TX 78216. Text and Basic listings, $15; same with listings on disk, $23. If you can feel challenged by simple games, they might give you the best opportunity to explore your own PK. You can liven up the games with mental pictures and discover what mental state seems best for success.
Extremely fast games that strain your eyes and your hand muscles might not work quite as well on the psychic level, because you may be too distracted--but we don't really know. With some people the complete involvement in some skill task may carry a subconscious PK effect.
The real challenge for game designers might lie in setting the stage for a basically simple chance event. When the plane drops its bomb, let the bomb gather speed gradually and emit the eerie sound of the bomb, and then let a random event decide whether the bomb really explodes on impact.
Table: Table 1. Statistical Significance.
Table: Listing 1. Quasi Random Algorithm.
Photo: Figure 1. (a) Geiger Counter Circuit. The flipflop is toggledat random time intervals. Position the source for an average counting rate of about 40/seconds.
Geiger Tube: Amperex 18504
Source: Sr90, 0.1 C (from Nucleus, P.O. Box R, Oak Ridge, TN 37830).
(b) The electronically clean manual switch can also provide truly random timing.
Photo: Figure 2. A random number, RN, is generated by stopping a fast counter at a random time, when the input Q goes from 0 to 1. The lower bits (or digits) of RN are truly random, provided that the average waiting time for a Q change is long enough, so that these bits (or digits) have been turned over many times.