Personal testimonies. They can often be a very powerful tool of persuasion, so there is no wonder why religious folks are so happy to share them in an attempt to win over converts. But despite their emotional potency, can we really trust personal anecdotes as objective descriptions of real events?
Consider the following experiment :
Suppose you are shown the following film footage of an automobile accident. [show collision]. If I were to ask you, “About how fast were the cars going when they hit each other?” what would your guess be? 25 miles per hour, maybe? Or perhaps 35?
Well, as it turns out, people are actually not very good at visually estimating the speed of impact for automobile accidents such as this. So even though the true speed was only 25 mph, most control groups tend to estimate values on the order of 34 mph.
Now let’s shake things up a bit. Although it is interesting that people tend to over-estimate the collision speed, we are here for another reason. Suppose I replace the word “hit” with the word “smashed.” In all respects, this is exactly the same question as before, but now I’ve inserted a verb with just a slightly more potent connotation. Surely, the mere phrasing of a question would not influence the answer a subject gives, would it?
As it turns out, simply replacing the word “hit” with the word “smashed” can bias the subjects’ responses to over 40 mph. And to make matters even more interesting, a whole continuum of values can be obtained from a whole slew of verbs. This is the basis for what is called a “leading question,” wherein the question itself actually biases the response given.
Now after performing an experiment such as this, there are basically two ways to interpret the results. One possibility is that subjects are simply uncertain in their response, and the nature of the question biases the answer. For example, suppose you’re not sure if the speed was closer to 30 mph or 40 mph. When you hear the word “smashed,” you might subconsciously take it as a subtle hint that the speeds were relatively high, and so you respond accordingly.
An alternative explanation is that the question itself actually “tricks” the subject’s memory into seeing the accident as being more violent than it really was. That is to say, when you retroactively review the video footage in your mind, the word “smashed” conjures up a memory of an event that was more intense than reality. If this view is correct, then the very nature of the question may actually create new, peripheral memories around the event to coincide with the added intensity.
To test these explanations, the experimenters gathered three groups of 50 and again showed them a film depiction of an automobile accident. The first group was asked how fast the cars were going when they “smashed” into each other; the second group how fast when they “hit” each other, and a third group was not asked about speeds at all.
One full week later, the subjects returned for another series of follow-up questions. Buried within the survey was the question “did you see any broken glass?” Now in reality, there was no broken glass in the film, so it is reasonable to expect that very few of the control subjects would respond with a “yes.” And this is exactly what happened. Out of 50 control subjects, only 6 of them reported seeing any broken glass. Subjects conditioned by the “hit” question also responded reasonably well, with only 7 out of 50 answering “yes” to the broken glass. Yet the subjects who were pre-conditioned with the word “smashed” answered with over twice as many “yeses” as the other two groups.
Now isn’t that interesting?
So it seems that not only does the question itself bias the responses given, but it can also physically generate commensurate memories of events that never occurred. This phenomenon is called the “misinformation effect,” wherein post-event misinformation can actually integrate itself into a person’s memory of a real event. And that’s just what happens from simple, leading questions. Imagine what could happen if we are also placed under social pressure to mis-remember events. For example, when asked to recall any broken glass, suppose that you must first sit and listen as another subject confidently asserts, “yes I absolutely saw broken glass.” These situations have also been well-studied [show papers], and all of them show a significant tendency for social pressure to increase the level of memory distortion.
So how do such experiments apply to the case of these folks here? Can we really trust them at their word when they speak about miracles from God?
At best, these people simply experienced intense emotional arousals during times of heavy psychological duress. If we then sprinkle in a little social pressure from other believers to ascribe such events to the hand of God, we are statistically guaranteed to insert, omit, and distort important details from our memories.
This is why eyewitness testimony is universally regarded as the weakest form of evidence. It is not because these folks are deliberately lying to us in some organized conspiracy, but rather because we have a well-documented tool by which social pressure and mis-information can combine to create specific memory details with no physical correspondence in reality. Even the mere act of retelling the modified story can combine with insufficient justification effects to further solidify any distortions [insert model with memory]. Unless these people can devise a quantifiable model for their experiences and then reproduce their effects on demand, we have no choice but to be skeptical of all personal testimonies.
So the next time you hear stories from people about the time they felt the power of the Holy Spirit, witnessed the hand of God, or took part in miracle cures, ask yourself: Can I really trust the validity of such a story? Or is this person just another poor sucker who fails to appreciate his own power of self-deception?
- E. F. Loftus and J. C. Palmer, "Reconstruction of automobile destruction: an example of the interaction between language and memory," Journal of Verbal Learning and Verbal Behavior, Vol 13, No 5, pp 585 - 589 (1979)