Have you ever had that feeling of being watched? In the late 90s, biologist Rupert Sheldrake searched for a psi experiment that was simple enough to be carried out at high school level. His thinking was that such an experiment could introduce young adults to the concept of psi research and teach them a little science along the way. It would be an easy way to show that psi was real. So the ‘being watched’ experiment was evolved.
The Staring Studies Are Simple
What he came up with was a staring experiment. It’s not only a simple concept, but the execution of the experiment is also straightforward. You have one person designated as the starer and another person designated as the subject who cannot see the starer. The starer then randomly stares or doesn’t stare at the subject and the subject records whether they thought they were being stared at. The subjects are given from 10 to 20 seconds to make up their minds.
If you do this enough, you find that people correctly guess that they are being watched or stared at, at much better than chance levels. In some experiments, it was as much as 10% better, and in an Internet experiment, it was only about 3%. (The advantage of an Internet experiment is that you can do a lot more trials.) It’s not something that people get right all the time, but it is definitely a testable effect.
It’s like that in most psi experiments and why statistics are necessary most of the time to discover an effect. If psychic ability was so easily replicable that we didn’t need statistics, we wouldn’t have the present controversy. Nevertheless, it was a successful experiment by any sane standard.
The Devil is in the Details
Even with a simple experiment like this one, the devil is always in the details. You begin staring studies research by surveying general populations to determine whether it’s even a thing. Sheldrake referred to three different surveys: (Braud, Shafer & Andrews, 1990; Sheldrake, 1994; Cottrell, Winer & Smith, 1996) to show that a large number of people, 70-97% of the population, claim to have experienced psychically knowing that they were being watched or stared at. This alone made it worth checking out.
There are several ways to conduct this experiment, depending on budget. You can do it via CCTV, where the starer does their work via a live feed from the participant. Therefore, you can measure galvanic skin response and do away with participant reporting, or do it on the cheap and have the starer and the participants in the same room and just make sure that the participants keep their backs turned and self report. Try blindfolding the participants, or you can do a combination of the above, depending on how strict you want the experimental controls to be.
It’s useful to do staring studies several different ways, because this tells you if the extra controls are necessary. If, for example, you do one experiment with CCTV and another experiment where people just have their backs turned and the results from the two experiments are similar, then it’s highly likely that the additional step of using CCTV was unnecessary. (That’s what happened, by the way. The CCTV experiments did not have results much different from the less well controlled experiments.)
Controls and Potential Problems
Experimenting with controls is also easy. A group of people have their backs turned and are never stared at. (The controls guess at rates that are strictly at chance levels.) Still, even in something as simple as staring studies there are a number of things to watch out for. Someone may try to peek. Or they might hear something from the starer that cues them. If so, this problem would show up in the control experiments as well, which it didn’t.
Peeking requires a turning of the head. People cannot see directly behind them, even with peripheral vision, so this would be an obvious move that would alert the experimenter to the problem. If you think hearing is the problem, you can also use a video feed and stare from another room.
A more difficult possibility is when feedback is given, participants might pick up on subtle sensory cues and get better at guessing when they’re stared at. Then there is the possibility of outright cheating. (This can be prevented with better controls.) If sensory clues were the problem though, the CCTV experiments would have had different results. This didn’t happen.
There is also the possibility of faulty recording which a problem encountered in all research that relies on humans recording their experiences.
And the last is the experimenter effect. This particular problem is the most interesting thing to come out of the staring studies. You see, because this experiment was so inexpensive, more than one skeptic tried their hand at replication. And a strange phenomenon popped up. They were unsuccessful even when another person reported positive results with the same group. More on this later.
Crossing the “t’s” and Dotting the “i’s”
All of these issues were brought up and dealt with in various papers written on the study. This is typical of parapsychology experiments. Scientists must be extremely thorough because they can and will be criticized for everything, no matter how trivial. Skeptics will breathe fire on them if they butter the wrong side of the toast.
Overall, the experiments were successful. Like most psi experiments, the effect wasn’t particularly large, but thousands of trials demonstrated a very real effect. Sheldrake received more staring studies than the ones he included, but some didn’t get published because they didn’t follow the rules. A rather obscure problem can result from publishing the results of every study submitted. It’s known as the “file drawer effect” (the fancy pants name for this is “selective reporting”).
“File drawer effect” refers to studies that were completed, but never published. If only positive studies are published and studies with no effect are buried, it can skew the results of a meta analysis to a false positive. This was certainly a possibility in a situation like this. Schools which performed the experiment and didn’t get positive results might have failed to submit those studies. If all Sheldrake got back were positive studies, but there were also a bunch of failed studies that never made it to him, then the combined results would not show what really happened.
Sheldrake solved this problem with pre-registration He had an experiment registry and the experiment had to be recorded in it before it started to be included in the overall assessment.
Skeptic Problems Of Being Watched or Stared At
Skeptics did their own experiments, but as usual, even when they got the same result, tried to explain them away: (Wiseman & Smith, 1994, Colwell, Schröder & Sladen, 2000). Wiseman and Smith claimed that the results were due to "the detection and response to structure" present in the randomization. In other words, the subjects were correctly guessing what was coming next because they saw a pattern in the randomization.
What makes this particularly idiotic, in my opinion, is that the particular randomization process used by everyone was one that Wiseman and Smith had themselves suggested. (Counterbalanced randomization: there are the same number of hits on both sides.)
In any case, this complaint was a nonstarter because if it were true, then the subjects would have learned equally well in staring and non – staring trials. This didn’t happen.
There were other cases where skeptics hypothesized alternative explanations, but failed to follow up on them, apparently not caring whether they were right or not, as long as they could find a way to explain away positive results. I don’t know why anyone bothers with these people.
Being Watched – The Experimenter Effect
The experimenter effect was already a known problem in scientific experiments, but this particular staring study highlighted it in a very clear way. When skeptics themselves were the starers, they tended to get chance results. In one case, this occurred even though the skeptic was side by side with a parapsychology researcher, Wiseman & Schlitz (1997). In that case, Wiseman, someone who has a public persona as a psi skeptic and enjoys the spotlight, and Schlitz, an enormously competent psi researcher who has stayed mostly out of the spotlight, could not have been more different in their motivations and intent.
Getting positive results and admitting the reality of psi would have marginalized Wiseman and hurt his public persona. (In 1997, this happened to everyone who took psi research seriously.) Schlitz? Not so much. By staying under the radar and avoiding being in the middle of controversies, she was free to scientifically explore and learn. Hers was a much better psychological position to be in for psi experiments.
If you were aware of this backstory, then the rather dramatic results of the experiment, where Wiseman could not beat chance results at all while Schlitz made it look easy under the same conditions and using the same participants, would be the expected outcome. It also explains why this experiment struggled with replications. It would be difficult to find people quite as different as these two.
The Staring Studies Were Consistent With Other Experiments
There isn’t a lot more to say about the staring studies. It was a simple experiment, the skepticism was weird and unhelpful as usual, and just as importantly, the results of these experiments were consistent with other research.
At the end of the day, the staring experiments are just another replicated experiment confirming the existence of psychic ability. If you need evidence, well then, there it is, along with all the rest. So the question is now answered; can you sense if you are being watched? The answer is yes!