Observing someone modeling pain tolerant behavior can influence one's own perception of pain

Social factors have been shown in dozens if not hundreds of studies to have widespread and profound effects on numerous health outcomes. For example, people who are isolated tend to live shorter and less happy lives, and knowing others are available for social support can buffer us against the ravages of everyday stresses in life and decrease the risk of cardiovascular disease, diabetes and other conditions. Given this immense influence of social relationships on so many aspects of health, it has been surprising to me that there has been relatively little research exploring the impact of social factors on pain!

Indeed, I think this topic is so important that I chose to make it the topic of my Ph.D. dissertation.

In the process of becoming familiar with the scientific literature, it occurred to me that one way in which one's social environment might influence pain is through social modeling. That is, the degree to which people would find some experience painful would depend on the responses of others in their social environment. In other words, after watching someone stoically absorb painful shocks you would feel less pain from those shocks than if you had just observed someone responding more weakly to the pain stimuli.

In the 1970s, psychologist Albert Bandura, developed what has become known as social learning theory. The essence of social learning theory is that people can learn new responses through observing others in lieu of direct experience. Bandura, in his 1977 book said, "Learning would be exceedingly laborious, not to mention hazardous, if people had to rely solely on the effects of their own actions to inform them what to do. Fortunately, most human behavior is learned observationally through modeling"

I thought this was pretty straightforward stuff so when I hit the literature I expected to find lots of studies looking at social modeling and pain. I was surprised to learn that I would have to go all the way back to 1976 and a paper written by Kenneth Craig and presented at an American Psychological Association meeting.

Here's how it went...




Participants sat down beside another individual whom they believed was another participant but who was in fact working for the experimenters (a confederate). Participants underwent a series of electric shocks of increasing intensity. The confederate appeared to likewise receive the shocks, but in fact did not. After each shock both individuals rated the pain on a wooden scale marked with numbers 0 to 10 with a sliding metal pointer to indicate pain intensity. A wooden barrier separate the participant and model. Participants were randomly assigned to one of 5 conditions:

1. Interactive tolerant: Both participant and model verbally declared their pain ratings after each shock. The model consistently indicated shock ratings at 75% of the participants' ratings.
2. Subject verbal only: Participants declared their ratings verbally but the model remained silent but indicated pain using the sliding scale. This condition was intended to control for the presentation of oneself as pain tolerant. 
3. Model verbal only: Similar instructions to subject verbal only except that in this case it is the model who declares her pain ratings and the participant who remains silent. In order for the model to continue to declare shock ratings at 75% the level of participants, they used a concealed mirror to spy on the participant's ratings on the slide indicator.
4. Both silent: This condition controlled for the fact that the mere presence of a companion can reduce the aversiveness of a situation. 
5. Subject active only: only the participant received and rated shocks.

Here's a graph that I created, which was adapted from Craig's:

As you can see, participants exposed to a tolerant model reported considerably lower levels of pain and persisted longer than individuals not exposed to a tolerant model. And this difference became larger as shock intensities increased. For example, in the both silent condition (in which case if there was an influence of the partner it would only be the easing of stress due to the mere presence of another person going through the same aversive experience) a stimulus intensity of 3 mA induced an average rating of 5 (on the 0 to 10 scale), whereas for those exposed to the tolerant model, the same stimulus intensity induced a much lower average pain rating of about 2! Expressed another way, the same pain rating of "5" was reported by participants receiving very different shock intensities: It took only 3 mA to prompt this rating in people without the tolerant model but participants exposed to the tolerant model would require an 8 mA shock intensity before rating it a 5!

One thing I like about this study is that they did a reasonably good job of controlling for alternative explanations. For example, if the presence of another person compelled participants to "show off" or somehow present themselves as stoic, then the subject verbal only group should have produced pain ratings not much different from the two groups who were exposed to the tolerant model, which, as you can see in the graph, is not the case.

Of course there still remains many fascinating questions. For example, in the subject verbal only condition, participants may still be trying to show off pain tolerance by giving lower ratings that they otherwise would. The reason why participants in the tolerant model conditions rate pain even lower than the subject verbal only group is perhaps simply because they have a specific figure to which to anchor their own responses. Participants might think, "hmmm, my initial impression was that it was more painful than a 4 but now that I think about it... yeah I guess it wasn't so bad... I'll also give it a 4"

So future studies will have to be a bit more clever about this. I have some ideas but I'm not telling :-)

And after finding this paper, I discovered that Kenneth Craig published some similar research around the same time.

Craig, K., & Weiss, S. (1971). Vicarious influences on pain-threshold determinations. J Pers Soc Psyc, 19(1), 53-59.

Craig, K., & Neidermayer, H (1974). Autonomic correlates of pain thresholds influenced by social modeling. J Pers Soc Psych, 29(2), 246-252.

Craig, K., Best, H., & Ward, L (1975). Social modeling influences on psychophysical judgments of electrical stimulation, J Abn Psych, 84(4), 366-373.