Is it possible that placebo effects are merely the result of biasing attention towards signs of improvement?

Well known doc decked out in white lab coat, in famous pain clinic hands you a packet of pills and says with confidence that the medicine he is giving you is a potent painkiller and that you should feel less pain shortly after taking it. You take the pill when you get home. Sure enough, your pain eases considerably. You really do feel less pain. Thank goodness, you think, for the existence of such wonderful analgesics.

Yes, analgesics are indeed wonderful, but you didn't get one. The pills the doctor gave you contained only starch. You got a placebo. "But I really did feel less pain," you protest. Absolutely, but it wasn't because of anything in the pill itself. So how on earth did this placebo work to quell your pain? Well, placebo researchers would say that the confident doctor and clinical context promoted a strong expectation of analgesia and it is this expectation that prompted the brain to release endogenous (within the body) painkilling substances such as opioids. This may be. Indeed, many studies have shown that when drugs that block the action of opioids are administered immediately subsequent to placebos, analgesia does not occur. But a huge mystery remains: how does a mere expectation (no matter how strong or confident) lead to the complex neurophysiological activity associated with analgesia? Put another way, how exactly does expectation "turn on" the brain's built-in analgesic systems? If expectation can induce analgesia, why can't we merely "will ourselves" to feel less pain?

Here's another scenario: You're in the hospital. Everyday a nurse comes to your room and gives you a shot of morphine. Shortly afterwards you feel your pain ease. This repeats multiple times a day over a series of days. But then one day the nurse plays a trick on you. Without your knowing it she substitutes saline for the morphine. Shortly afterwards your pain eases. Thank goodness, you think, for morphine.

But once again, we are left with a mystery. How is it possible that to feel less pain when you received nothing more than salt water? Is morphine one big placebo effect? Is it a scam? No, morphine is truly a powerful painkiller. Placebo researchers would say that in this case, the repeated pairing of morphine with subsequent pain relief conditioned your nervous system to predict/expect pain relief much as dogs learn to salivate at the sound of a bell following multiple pairings of a bell followed by food. So this conditioning is somewhat similar to the situation in which your are led to the expectation of pain relief through verbal suggestion by a doctor. The difference is that in the case of conditioning, you are not just given a suggestion, you actually learn from experience that A (injection of drug) leads to B (pain relief).

But again, the mystery remains: how on earth can conditioning produce pain relief? You see, during the conditioning (learning) phase, you're actually receiving a painkiller (e.g., morphine) so it is the specific pharmacological action of the drug that is producing the analgesic effect. Then you're given a saline solution, yet you still feel less pain. How is this possible in the absence of the drug? It makes one wonder whether it was the drug that was relieving pain after a few administrations.

So the exact mechanisms underlying these two methods of provoking placebo effects remain a mystery.

Just because it's a mystery doesn't mean there aren't any theories out there.

One particularly interesting one suggests that the placebo effect is nothing more than an issue of signal detection. What the heck does that mean?

I'd like to use a great analogy introduced by Lorraine Allan and Shepard Siegel (2002) in their interesting paper on the application of signal detection theory for understanding placebo effects. Suppose that you're a physician and a patient presents with abdominal pains. You need to diagnose the cause of the symptoms. So you perform a physical exam, order blood tests, x-rays and an ultrasound. The task before the you is to identify a diagnosis that best fits the evidence. You suspect appendicitis. There are 4 possible outcomes:

	Decision
	Operate	Do not operate
Appendicitis	Correct positive	False rejection
No appendicitis	False positive	Correct rejection

1. The patient has appendicitis and you choose to operate (correct positive).
2. The patient has appendicitis but you fail to detect it and therefore do not operate (false rejection).
3. The patient does not have appendicitis, yet to believe you've detected and therefore operate (false positive).
4. The patient does not have appendicitis and you reject the possibility of appendicitis, and therefore do not operate (correct rejection).

So, there are two ways that you could be right: the signal (appendicitis) could be present amid the noise and you are able to detect it or the signal could be absent (no appendicitis) and you correctly conclude it is not there.

In a situation in which the evidence is unambiguous and a diagnosis is clear, it is easy to conclude correctly whether a signal is present or not. But most real life situations are filled with ambiguity. In this situation, it is quite possible to make an error. So in making your diagnosis you must decide which of the two errors is the worst evil. Is it better to mistakenly identify a signal (false positive) or miss the presence of a signal (false rejection). I would hazard a guess that most doctors would say the latter is the worst of the two evils and so they would adopt a "liberal" criterion.

Well let's say you decide the patient has appendicitis and you perform an appendectomy. During surgery no evidence of appendicitis is found. The patient apparently suffers from some other condition with indications that are similar to appendicitis. You made a false positive error -- you decided that a signal (infected appendix) was present amid the ambiguous information, when in fact it wasn't.

Since it has been determined that appendicitis is not the cause of the symptoms, you give the patient medication to provide symptomatic relief. Now the patient faces a task that is actually quite similar to the physician. The patient does not feel the same level of pain or types of sensations all the time. Signals from the body are ambiguous just as there was ambiguity in the information available to the physician. Yet, like the physician, the patient must detect a signal (attenuation of abdominal pain) is present amid the context of noise (variable levels of pain and other sensations over time). The patient reports a decrease in pain.

But this patient has been given a placebo -- a pill with no active ingredient. In effect the patient applied a liberal criterion, as you did. Was your patient highly suggestible, gullible? No. As Allan and Siegel (2002) put it,

"The patient's mistake is as understandable as the physician's mistake. Both are inevitable results of deciding on the presence or absence of an ambiguous signal. In the case of the physician, the mistake is called a false positive. In the case of the patient, the mistake is called a placebo effect." (Allan & Siegel, 2002, p. 415).

In essence, signal detection theory states that placebo effects can arise because of a search for signs amongst a noisy information that a treatment is working can lead to the misidentification of improvement when it has in fact not occurred.

Interesting, eh?