 Now, Shep Stuff is really interesting. He does some fantastic work on drugs, placebos, anticipation. He wrote a really nice paper recently on the wisdom of the body, where he unpacks each of these issues. And I'd encourage people to check out that uncut or the extended conversation that we had with him, which is at the end of this episode. But he mentions two mechanisms, two reasons for the placebo effect. One is the response bias, the other is regression to the mean. And it's worth unpacking this idea of response bias a little bit more. If we go back to his example that he provided with appendicitis, so if you're a patient, you're complaining, oh, my pelvis hurts, and you go to the clinic and you see a physician, they're either going to decide to operate or not on the basis of the information they have in front of them. So there are four possible outcomes here, and we're representing it using a two by two table, two by two contingency table. There's four cells, it's nothing fancy. But essentially, you either have appendicitis or you don't. And the physician's either going to decide to operate or not. So there are two ways of being right and two ways of being wrong. So these have labels. So if you, in fact, have appendicitis and they decide to operate, that's good. That's a correct decision, that's called a hit. But if you have appendicitis and they decide not to operate, that's pretty severe, that's a miss. So you have an inflamed appendix and they decide, oh, you're fine, go home. You could die if this were serious enough. That's a really costly mistake. If they decide, if you don't have appendicitis and they decide to operate anyway, that's also a mistake. That's called a false alarm. And that's severe. That's essentially unnecessary surgery. But it's not as severe as a miss. And finally, we have a correct rejection. That's where they correctly decide to take out an inflamed appendix. So those are the four possible cells. The four possible outcomes here. And when we're talking about a response bias, all we're talking about is a tendency, a bias, a tendency to say, in this case, operate. Now the reason that you would want to err on the side of caution and operate is because of that very severe cost of a miss. So somebody could die as a result of missing an inflamed appendix. And so if a patient comes in and they have a complaint of a sore pelvis, yeah, you take that very seriously. And you would err on the side of caution and you might decide to operate. And as a result, you might take out the odd healthy appendix. But on the same note, you're also going to take out a lot of inflamed appendices. So that's this idea of a response bias and deciding more often to operate than you may ordinarily. Yep, the exact same thing is happening with the placebo effect. So again, there are two ways of being right and two ways of being wrong. Four possible outcomes that can happen. You can take a pill that has no active ingredient, a placebo. You can take a pill that has an active ingredient, a drug. And you can either report feeling better or report not feeling better at all. Now a response bias in this case is just a tendency to report feeling better regardless of whether you took a drug or a placebo. Just as in the case with the physician, they make mistakes. We're going to make mistakes as well. Think what you're doing here. You have to reflect on your own internal states about how much pain you're feeling. You have to think back with your fallible memory about how you felt before and report whether you felt better or not. Just as we saw in episode three, we are not very good at internally reflecting on our own physical and mental states. It's a very difficult task and we, just like the physicians, are going to make mistakes. That's right. And again, thinking about the nature of the sort of decisions that you're making here, it's things like pain or depression, which are very fuzzy, very multifaceted and ambiguous as we talked about before. But I think it's really great, this idea of representing things using this 2x2 table. So it doesn't seem like much, but it is. I mean, if somebody has a claim about something, if they say that this diet is the best thing under the sun, or this thing is going to heal you, that's a claim. And often what's happening is it's an anecdote, it's a testimonial. So all you're hearing about is that positive, positive cell. The number of times that you did the thing, took the diet, did the treatment, and felt better as a result, felt more energetic, whatever. But all you're hearing about are the hits. All you're hearing about are those positive instances. Hopefully, if we get anything out of this course, of this episode, if not the course, is what about the other three cells? Now I want this just to be at the tip of everyone's tongue. When they see any sort of claim that anyone makes, what about the other three cells? What information do I have? What information am I missing? And if you represent it using these two by two tables, it becomes immediately obvious. Well, hang on, what about the number of times that I took the treatment and I didn't feel better? Or if the claim is about someone else, how many times did someone else take the thing and didn't feel better or more energetic? How many times did they not take anything and they felt better? That's right. Or better, a placebo, right? If they took the thing without an active ingredient in it and didn't feel any better, then that's also worth knowing. And so you might demand that sort of evidence from the person who is making the claim. Well, has it been done with placebos? What sort of information do you want? And that makes it really immediately apparent. So what I want to do next is I hope people will go to the next section and answer a few questions about two by two tables and placebos and really make sure that they understand what's going on before we move on.