I have a lot of mixed feelings about this essay, probably too many to put in a comment. I really enjoyed it and they raise a lot of good points to think about,

I have a lot of mixed feelings about this essay, probably too many to put in a comment. I really enjoyed it and they raise a lot of good points to think about, and appreciated the bit at the end about causal mechanism, even though it came out of nowhere a bit. I think they're making a bit of a strawman argument in an attempt to make an untenable argument for homo economicus, though. This shows up in a few places. For example, the author sets up this argument that ego depletion "is the foundation for priming-style automaticity and in fact most of the phenomena that I label as examples of automaticity." This makes no logical or theoretical sense at all. Ego depletion is a resource depletion theory of cognitive and behavioral control; priming and automaticity has to do with dual-process and fast-slow processing theories. They intersect but are not dependent on one another at all. The author might be interested to know that there is a respectable literature arguing against dual-process / automaticity models on more sophisticated grounds. But they don't argue against the idea that humans are biased or that decisions can be made outside of conscious awareness, they argue against the idea that automatic and controlled processes are distinct, that it makes sense at all to separate the two types of processes. There's several other examples of this. Evidence for cognitive biases is massive; the priming literature is just a small slice of it. There's all sorts of biases that operate outside of primes, and might not even be thought of as automatic, but are thought of as operating outside of awareness. EEG/psychohysiological studies have also documented pretty clearly that decisions can occur before we are aware of them. It's pretty clear that humans are not always rational and unbiased, unless you stretch the definition of rational to some point where it ceases to have any meaning outside of "not automatic". I had trouble with the placebo effect section also. The meta-analysis on the magnitude of the placebo effect on depression the author cites itself recognizes the difference between active and passive placebo, which provides evidence the placebo effect isn't a simple regression to the mean or reversion via homeostasis. There have also been studies suggesting that iatrogenic placebos do maybe happen (in waitlist conditions for example), but they happen in such a way that outcomes are worse than passive placebo. Most people know they will be given treatment at the end of a trial regardless of their condition, so the explanation that they have to "fake bad" to get treatment seems plausible but just as likely as someone with the treatment who believes they've been assigned to control. Are placebo effects small? It depends on the placebo effect. That average is just an average: depending on the nature of the control, and the patient population, it can be small or large. Sometimes, the placebo effect is largest for precisely the patients where it matters the most. Anyway, this all sounds like I disliked the essay which strangely I didn't. I think it comes across to me as raising some strong arguments (that maybe placebo effects are smaller than we think, or maybe automaticity maybe needs to be revisited), and I agree about the causal mechanism part. But a lot of it comes across as sort of hand-wavy; I think it could have left half of it out, or even aquiesced on some points, and it still would have been persuasive. The phantom in the room is the lack of a definition of "rational" provided by the author. It seems as if a lot of the essay hinges on somewhat loose definitions: that goes for attacks on automaticity, but it could go the other way too. What's "rational" versus "biased"? Can someone learn something rational in one scenario, but have it become irrational due to circumstances changing in a way they don't recognize? Is that irrational?