Hack 68. Search for ESP

Though most scientists agree that there isn't much evidence that ESP actually exists, they might be wrong. You or your friend or your monkey might have ESP, and there's no time like the present to find out!

The term extra-sensory perception (ESP) was coined to describe perceptions that are independent of the traditional five senses: sight, hearing, touch, taste, and smell. The first to use the term was a psychologist at Duke University in the 1920s and 1930s named J.B. Rhine. There was much excitement at the time, as Rhine and his colleagues were able to identify individuals who seemed to exhibit ESP abilities. In the popular press and some of the scientific writing of that period through the 1970s, it was even taken for granted that there was such a thing as ESP and that we all had the trait to a certain degree.

Today, though, you don't really hear much about ESP, and most scientists have concluded that such a thing probably does not exist. More specifically, it hasn't met the criteria for scientific acceptance that any other hypothesized phenomenon is expected to meet, such as experimental evidence, replicated studies, and so on. You can add to the data, though, by conducting your own studies and identifying whether you or your friend might be psychic.

Identifying Psychic Abilities

Though there is a wide range of supposed psychic abilities, ranging from reading minds to moving objects with one's mind, the traditional way to study ESP has been to use a deck of cards called Zener cards. Zener decks have 25 cards with matching backs. The face of each card displays one of five symbols: a circle, cross, square, star, or wavy lines, as shown in Figure 6-7.

Figure 6-7. Zener cards

If you don't have a deck of these cards handy, you can make them pretty easily with a pack of blank index cards and a black magic marker. Just make sure that no one can see right through them (unless they are psychic, in which case they can see right through you, too). Make 5 cards of each symbol for a total of 25 cards.

There are a few different ways you can use a shuffled deck of Zener cards to conduct an ESP test:

• One person tries to guess the order of the cards by announcing each one before it is turned over.

• One person looks at the face of each card and attempts to "send" it to another person telepathically who is sitting nearby.

• A person in another room or in a distant location looks at the face of each card and attempts to send it telepathically to another person over a great distance. Sometimes, the receiver imagines that they are in the room with the sender and can see the card.

With whatever method you choose, the procedure is to go through all 25 cards and keep track of the hits and misses. How many cards out of the 25 did the subject correctly identify? In some studies, the receiver is told how he is doing while they are going through the whole deck; sometimes, he is not told until the end of the experiment. The outcome variable is the number or percentage of cards that were correctly identified.

In ESP research, the person who is trying to read someone's thoughts is the receiver and the person who wants her mind read is the sender.

Analyzing the Results

If the results are what would be expected by chance alone, treat the outcome as evidence that the subject is not psychic. If the subject gets many more correct than would be expected by chance, that outcome suggests that the subject might have ESP.

So, what would be expected by chance? If you are guessing for 25 cards and there are five cards of each type, chance alone would get about 5 correct. Imagine, for example, that you guessed star every single time across all 25 times. You would be guaranteed 5 hits and 20 misses because you know star will come up exactly five times overall. If you guessed randomly each time among the 5 possibilities, your average success would also be 5 out of 25, or 20 percent.

What if you had a higher success rate than 20 percent, though? What if you were correct 6 out of 25 times, for a success rate of 24 percent? Should we treat that as evidence that something other than chance is playing a role here? What we need is a statistical analysis of the different possible outcomes, to identify what percentage should be considered so unusual that it must be evidence for the presence of something so unusual.

A statistical test reveals only whether chance is the best explanation for an outcome. For our experiment, a statistically significant outcome doesn't prove that ESP is at work, only that chance is not the best explanation. After all, the best explanation for a high hit rate might be that the receiver sees the cards reflected in the sender's glasses, or some other less interesting cause.

We know that over the short run (or in a small sample, to use the stats jargon), results that differ from the population are common. We also know, though, that a large difference from that population value is uncommon, especially over the long run (or with a large sample). In fact, the probability of finding a difference of a given size between a sample value and the population value is directly related to the size of the sample.

For ESP experiments, the sample size is the number of guesses or trials, and the population is the known distribution of the different symbols across all trials. The population value for any number of guesses is 20 percent correct; that is what would be expected by chance. If there is a large difference between the sample value and this population value, then something other than chance is likely operating.

The statistical analysis appropriate here is something called the Z-test for comparison of an observed proportion to an expected proportion. It is similar to other common statistical tests, such as t tests [Hack #17], which calculate a difference and determine how frequently such a difference would be found if a given sample really was randomly drawn from a population with certain characteristics.

The probability of any difference is based on the size of the sample. For example, if after 25 attempts, a person guessed 24 percent correctly instead of the expected 20 percent, the information needed for this analysis would be:

• A sample size of 25

• An observed proportion of .24

• An expected proportion of .20

Without showing the formula and calculations for this particular analysis, I'll show you the result. By chance alone, with 25 guesses, a subject will guess at least 24 percent of the cards correctly 31 percent of the time. Another way of saying that is that out of 100 subjects going through your study, 31 of them will get this result or better. So, a hit rate of 24 percent is better than average, but not so unusual that I would call The National Enquirer just yet.

What about other hit rates or if you test with more than 25 trials? Table 6-17 shows the chance of guessing given percentages of cards (or higher) correctly. This table assumes an expected hit rate of 20 percent.

Table Likelihood of selected ESP hit rates

Number of guesses	Percent correct (hit rate)	Probability of hit rate or better
25	20 percent	50 percent
25	30 percent	11 percent
25	40 percent	1 percent
25	50 percent	.01 percent
100	20 percent	50 percent
100	30 percent	1 percent
100	40 percent	.00001 percent
100	50 percent	.000000000001 percent

Notice the dramatic drop in likelihood for extreme outcomes as the sample size increases. For example, with just 25 guesses, the chances of getting 40 percent correct is about 1 percent; if you went through a pack of 25 cards 100 times, you are likely to do that well or better just one time. If you took 100 guesses, though, maybe going through the deck four times, you would get 40 percent or better correct just 1 out of 100,000,000,000,000 times!

How Much Is Enough?

If you want to conduct ESP experiments, you should establish a standard for how unlikely a performance must be for you to consider it evidence that something other than chance is the operating factor. Typically, in statistical research, if a result is likely to occur by chance 5 percent of the time or less, the result is considered statistically significant. For ESP experiments with 25 Zener cards and 25 guesses, you will guess 8 or more cards correctly about 7 percent of the time. You will guess 9 or more correctly just 2 percent of the time. So, some standard between 8 or 9 hits is scientifically reasonable.

The skeptic in me feels compelled to leave you with a warning. If you perform this experiment and get a significant result on yourself or someone else, that's pretty cool. If you can repeat the finding, though, replicating the experiment with the same person and getting similar results, that's when it will start to get exciting! If that happens, send me a telegram immediately. I'll sell my house, buy a train, and we'll hit the road to fame and fortune!