> "Bluff" has a very specific definition, and it requires that the person bluffing knows they cannot successfully follow through on the bluff.
From a game-theoretic perspective, this is incorrect (even though it accurately describes everyday use of the word).
Think about a poker bot which bluffs. It makes no sense to talk about "knowing" here. The GTO strategy to poker is just a solution to a gigantic series of equations.
My point was that talking about "thinking" and "knowing" is missing the deeper point. You can talk about bluffing without those concepts.
I'm not sure you understood my point at all.
What if the shrimp genuinely thinks he can take the big guy, because he's just so angry he gets stupid? That's not a bluff, that's a suicidal threat.
"Bluff" has a very specific definition, and it requires that the person bluffing knows they cannot successfully follow through on the bluff.
And from reading this, I'm not sure the researchers accounted for that distinction.