Neuroimaging Placebo Effects

Here's a link to an interview with Tor Wager (Psychology, Columbia University) on NPR's Talk of the Nation.  It lasts about 10 minutes.  The interview is about a month old and concerns recent neuroimaging results from his lab that support the belief that placebos can have genuine analgesic effects (a subject that I also discuss here).

Comments

[Neil]

This strikes me as a classic instance of the overvaluing of neuroscience (as compared to other sciences). Of course placebos have an analgesic effect. How do we know that? Subjects report that they have an analgesic effect, and since pain is a phenomenal property we take their word for it. Since analgesics have an analgesic effect - and phenomenal properties are psychological properties, and psychological properties cannot float free of brains - placebos must produce changes in the brain.

[Adam Kolber]

My post wasn't intended to suggest that the only purpose of the study was to confirm a placebo effect. So, I'm sure that Wager et al. are also interested in the neural mechanisms of placebo effects (e.g., to what extent do they relate to expectations of pain relief, conditioning effects of medical treatments, selective attention to pain symptoms, etc.) Neuroimaging may help address such issues.

On the other hand, Wager does comment in the interview to the effect that self-reported pain analgesia may reflect efforts to please investigators, rather than genuine pain relief. The comment suggests that neuroimaging may uniquely reassure us that placebo effects are not just artifacts of this desire to please investigators. Could double blind studies address this issue without neuroimaging? I would think so, though I'm really not certain and welcome further thoughts.

[Neil]

Neuroimaging might be able to address the questions about causal mechanisms. But I'm not convinced that it's the best way. All these questions can be addressed more directly by traditional psychological manipulations: separate belief from desire to please for instance; make pains more or less salient, and so on. Remember that neuroimaging is downstream of behavior; there is an extra inferential step in using neuroimaging to infer causation. That is, we know what various parts of the brain do only by inference from their role in behavior, directly or indirectly. So even though the neural level is a more basic causal mechanism, paradoxically we do better at studying causation at a less basic level.

By the way, standard dispositional explanations of the placebo effect refer to an *unconscious* desire to please. If the unconscious desire leads to a genuine reduction in pain, my criticism is compatible with the dispositional explanation (for that matter, even if the desire is conscious the criticism may be justified). If there is an actual reduction in pain - that is, unless the patient is lying to please the doctor - then there must be a neural correlation of this reduction.