Apr 13 2020
Demand Characteristics in Psychological Research
I have written quite a bit about the body part ownership illusion (sometimes called the “rubber hand illusion” because of the original study design). The idea is that the brain constructs all our sensations, perceptions, and experiences, including the sense that we own, occupy, and control our various body parts and our whole body. All it takes, apparently, is synchronization between seeing the “rubber hand” being touched and feeling your own hand being touched. Visual-tactile synchrony triggers the sensation of ownership. There is a robust replicated body of research supporting this conclusion. But particularly when people are the subject of research, no conclusion is beyond reconsideration.
That is the core strength of science – it constantly questions and reexamines its own assumptions and conclusions. Psychological research especially needs to do so, because human behavior is horrifically complicated and we cannot directly see (well, not yet) what is happening in someone’s mind, so we have to infer what is happening from things like behavior. That inference is usually based on some construct – some idea about how people operate and how this will translate into their behavior in a psychological experiment.
For example, there is the now famous marshmallow test. In this robust series of experiments, children were offered a marshmallow (or some treat) and told they could eat it now, but if they wait a few minutes the researcher would be back with a second marshmallow and they could have two. The construct for these experiments is that children with more ability to defer gratification through executive function will be able to hold out for the second marshmallow. So in the end this is meant as an experimental measure of executive function. This conclusion was accepted for decades – until it was questioned. There is another possible interpretation of the results – at least some of the children who don’t hold out and go right for the initial treat may not trust that the researcher will be back with more. They will take the bird-in-the-hand. This is a rational response – especially if you have lived your life with adults who are less trustworthy and where basic resources may be limited. And in fact children from lower socioeconomic backgrounds, with less trust and stability, generally do worse on the marshmallow test.
We may now be facing a similar reinterpretation of the body owernership illusion experiments, although at this point I don’t think it is going to turn out that way. A new paper, however, does point out a very important consideration that this and similar research must take into consideration – the role of demand characteristics. The idea itself is nothing new. In psychological experiments the study design must take into consideration the fact that subjects subconsciously try to figure out what the experimenter wants and then gives it to them. Any subtle cue that one response is more desired than another can affect the outcome. So the influence of demand characteristics must be carefully controlled for.
What the authors of the new paper ( Dr Peter Lush, Research Fellow at the Sackler Centre for Consciousness Science at the University of Sussex, and colleagues) are arguing is that the typical controls in studies of the body ownership illusion have not adequately ruled out demand characteristics. In particular, subjects who are suggestible may simply be responding to the suggestion that they own the rubber hand, and don’t actually feel that they do because of some circuit in their brain. To support this, in a preprint paper (not yet published in the peer-reviewed literature) they report that they found:
“…substantial correlations between response to the Rubber Hand Illusion and response to imaginative suggestion , or phenomenological control, in a large sample of 353 participants.”
Lush is making a valid point, and other researchers would do well to accept his recommendations for tightening up future research into the body part illusion. It also seems likely that the positive response to the body part illusion is likely contaminated with some subjects who are just highly suggestible (as is, probably, most of psychological research). The question remains, however, is this responsible for the research outcomes, or just a small contamination? I suspect it’s the latter, but we will need some robust evidence to settle it.
The reason I suspect that the body part illusion is real and not just all an artifact of research protocol is because there are several independent lines of evidence supporting it. We know, for example, that people with specific localized strokes can have disorders of body ownership, they can have phantom limbs, or even supernumerary limbs, or even alien hand syndrome in which they don’t feel as if they control a body part. These phenomena are often spontaneously reported by patients suffering brain damage – not subjects in an experiment who may be experiencing suggestion.
I would also point out the Lush’s own conclusions are subject to multiple interpretations. For example, it could be possible that they same underlying neurological phenomena would make someone susceptible to the body part illusion, and also susceptible to suggestion – even the the two are not directly causally related. In other words, he found a correlation between suggestibility and positive results on the body part illusion, but that correlation could be due to both phenomena being related to a third neurological phenomenon. We can come up with several constructs to explain this correlation. Perhaps the common link is a relative decrease in reality testing. Do those who do not experience the body part illusion simply reject the sensation synchrony because at a higher level they “know” the rubber hand is not part of their body? Those same people are less prone to demand characteristics because they also reject suggestions. Are they just more stubborn, critical, or oppositional? Perhaps all of these things are happening at once in some proportion.
But I love the fact that researchers are questioning even well-established research constructs and suggesting how they can be reinterpreted. But also, they are tying their criticism to specific future research that can settle the difference of interpretations. Their criticisms are testable. Because psychological research is subject to so many different models of interpretation, we need to explore as many different interpretations as possible. This will only make the conclusions more robust. This is also why we need to tie psychological research to neuroanatomical correlates and brain function as much as possible, because that provides some of the independent lines of evidence we need to know if our constructs are actually based in reality.