People's sense of control over actions is reduced when angry or afraid: study

A game-theoretic threat is a statement that, if you do something that hurts me (and might benefit you), then I will do a thing that hurts both of us. It's clear why "hurting you" is a critical part of the threatened action; the reason "hurting me" is included is that if it benefited me, then we'd assume I would do it anyway. The puzzle of the threat is then, how can I convince you I'll do something that will hurt me? Well, if I could modify my brain such that I will be determined to do something violent to you even at great risk to myself, that would be a great solution. Anger is an emotion that does exactly this. Even better, if it is generally understood that any violations of my body (or whatever I want to protect) will trigger an anger response in me, then I don't have to make specific statements towards people I come across.

One can imagine variations on this: (a) if you know someone else has an anger response, and you want to do something that you're worried might trigger it, then maybe you can do a light-touched version of it and see if you can detect the beginnings of their anger response; (b) if you get into fights too often, you might give a "warning" (growl) response first.

It's interesting to think how these things might have evolved, because there are two sides to them. Anger is useless—counterproductive, in fact—if no one else understands before they act that you might become angry. And if anger doesn't exist, then detecting anger wouldn't be a useful trait. What came first, then? I would guess that some animals evolved a fight-or-flight response to deal with predators (because that increases survival chances), and the triggering mechanism was probably clumsy enough that it happened in non-life-threatening scenarios and caused needless violence (with negative expected value to all parties), and that led to animals either being able to see the early stages of fight-or-flight, or having a sense of what will trigger fight-or-flight and avoiding doing those things (unless they plan to win the fight). (The fight-or-flight response is probably reasonably obvious to observe—muscles tensing, freezing or making sudden movements, etc.—so at least predators should have gotten good at noticing it, to become more efficient.)

I don't know what research has been done on the evolutionary origins of anger. But it makes so much game-theoretic sense—and I have the impression that mammals in general have an anger response; quick googling gives mixed results about reptiles—that I'm very confident that what we would call anger evolved because it was useful for backing up an implicit threat of "don't violate my body or other things important to me, or else I will become violent". Yet I didn't see any discussion of evolution or game theory in either the article or the study. Those would seem to be important for forming hypotheses and interpreting results.

Anyway, for the threat to work, the "I will become violent" must be a strong compulsion (taking away the choice to not carry out the threat) and/or value-adjustment (so carrying out the threat now looks like the best thing to do). Either might subjectively feel like a loss of control, or not. Given that evolution has successfully implemented this feature, the subjective feeling seems like an implementation detail to me—maybe useful in learning to manage your own anger response.