I agree that fatalism was an unnecessary distraction. The point about energy vs. willpower is that you cannot point to any precise biological mechanisms for wi

I agree that fatalism was an unnecessary distraction. The point about energy vs. willpower is that you cannot point to any precise biological mechanisms for willpower, but the claim that willpower competes with cognitive processing for processing resources suggests that there is a precise biological mechanism, in the same way that the coexistence of aerobic and anaerobic energy processing suggests a precise biological mechanism. The box doesn't yield measurements of willpower and cognitive processing. It yields measurements of memorizing a phone number and choosing to eat cake. We bind those things to our terms in order to generalize and create a model. Models are fine. But the problem is taking the model and saying the model actually maps to reality inside the brain. Even the authors of the paper in question hedge on this mapping with the words "seems like". Models are unfalsifiable because they are models and we made them up. The only thing that is falsifiable are hypotheses about behavior derived from the model. My irritation is when models of the brain that we do not possess the tools to validate in the brain are taken as neurological truth. I can suggest other models that fit the experimental results, and there's no way to say whether mine are worse or better. As far as I can tell, we agree about the utility of things outside of the box. But I'm not sure if we agree that the way the box works on the outside doesn't generate testable claims about how it works on the inside. As an example, if you write me a program that converts one list of numbers into another list of numbers and give me a binary (without the GPL...), it is actually not possible for me to tell what algorithm your program uses. I would have to disassemble it to figure things out.