Game Theory 101 (#78): Separating Equilibrium

This lecture introduces separating equilibrium, a form of perfect Bayesian equilibrium in signaling games. In a separating equilibrium, after observing an equilibrium strategy, the uniformed player can infer everything about its opponent. This makes solving for its optimal strategy straightforward.

Separating equilibria are normally the easiest type of PBE to solve for in signaling games. The cases will get a lot more complicated when we advance to pooling equilibria and semi-separating equilibria.

[gAmiRo]

Thanks for revealing all your information to us!

[SilverNight]

This helped me so much when studying for my exam! The explanation is very clear and concise, making it easy to follow and helps see why it is indeed logical. Thank you so much!!

[mohw]

Personal opinion. The player 1's actions being reveal or hide is kinda confusing, especially when adding the dot line. To make the question more intuitive, the actions can be changed to FIGHT STRATEGY 1 & 2.

[DanielVargas-yxux]

I wish I could see a similar explanation of this applied to the case of the Spencer model of education types, which was also an example.

[craigmagonna]

Thank you so much Mr. William Spaniel.

[PunmasterSTP]

PBE? More like PBJ, because my curiosity and this wonderful information go great together!

[nikki]

i think this video may have saved my life thank

[kalinagibson]

how would you solve this if there was a dotted line/info set connecting P2 on both sides

[wouterdegruijl]

This is a great explanation, thanks a lot

[sejalmitra]

I have a doubt. When you're discussing the profitable deviation for for weak type in the second case, you said that if player 1 would choose to hide its type rather than revealing it, player two would quit giving player 1 a better payoff than what he would get by revealing. But if player 1 does hide, then the probability with which player 2 assumes that whether player 1 is weak or strong would also change. Since player 1 is hiding, as explained before, player 2 would assume that player 1 is weak with the probability of 1 and hence fight and not quit as it did in the first case. But that is not what you have explained.

[flying_horse]

so is an assumption that the P is either 1 or 0 corect or not ?

[ianmcgettigan]

11:40 doesn't make sense based on what you said previously about the left side. If P1 deviates, that is reveals when weak, P2 should be under the false impression P1 is strong, and therefore would quit, not fight. This is the same reasoning you used on the left side so I don't see why it does not follow here on the right side. Am I missing something? Thank you

[MrNikehols]

why can Player 2 change his strategy in the first equlibrium from "quit" to "fight", but in the 2nd possible equilibrium he has to stay with his strategy "quit" even tough it would have been better for him to choose "fight" instead ?

Thank you

[AdelinaErad]

I think it should be strong hides, not reveals since 0.5 > 0.49 and 1 > 0.99

[naddar]

You should cover a model where information is not revealed also. Your chosen model is not used in our grad courses, nor highly used in Watson's Strategy book.

[jonavuka]

so wait, what about if the player always reveals or always hides?

[Scholarsbysimran]

Thankyou so much for this video, but i didnt get in end, so in conclusion which strategies are separating equilibrium can u list?

[rogermoody]

So pooling equilibrium = The same,  

separating equilibrium = something new!

[sags]

William in the end.. this is a simple game..

Me :- 😭