Comments on: Psychology and Game AI Gavan: Of course the logic you're talking about is one method of delivering the behaviour, so its useful, but my point really is that you shouldnt focus too heavily on the logical structure as the important part of the system, so much as on the psychology of the effect of that structure on the player (via the character or system you implemented). Behaviour is far less "crisp" than a lot of logical techniques would imply. We arent in a "state", but as you say are handling competing motivations. Plus I'd argue that psychology isnt based on "logic" in the programming sense at all. We rely a lot on emotion to colour our decision making (see the case of Phineas Gage and how losing emotions will cripple your decision making). I highly recommend reading "Descartes Error" by Antionio Damasio if you're interested in this. So yes, we should use logic as decision making, but then apply emotional and personality modelling within that process to produce appropriate psychological behaviour. Of course, the counter argument is that some subtle behaviour is impossible to tell from a notional "state" so why not just model the state in its simplest form. Which is a fair approach to take, but I think ultimately loses a lot of the finer detail of real behaviour, almost akin to the difference between a solid coloured polygon and a texture mapped one. Gavan: Of course the logic you’re talking about is one method of delivering the behaviour, so its useful, but my point really is that you shouldnt focus too heavily on the logical structure as the important part of the system, so much as on the psychology of the effect of that structure on the player (via the character or system you implemented).

Behaviour is far less “crisp” than a lot of logical techniques would imply. We arent in a “state”, but as you say are handling competing motivations.

Plus I’d argue that psychology isnt based on “logic” in the programming sense at all. We rely a lot on emotion to colour our decision making (see the case of Phineas Gage and how losing emotions will cripple your decision making). I highly recommend reading “Descartes Error” by Antionio Damasio if you’re interested in this. So yes, we should use logic as decision making, but then apply emotional and personality modelling within that process to produce appropriate psychological behaviour.

Of course, the counter argument is that some subtle behaviour is impossible to tell from a notional “state” so why not just model the state in its simplest form. Which is a fair approach to take, but I think ultimately loses a lot of the finer detail of real behaviour, almost akin to the difference between a solid coloured polygon and a texture mapped one.

]]>
By: Gavan Woolery/2011/04/19/psychology-and-game-ai/#comment-3010 Gavan Woolery Wed, 20 Apr 2011 07:44:32 +0000 ). Psychology is based largely on motivations, and motivations can be defined in logic (even if they are “irrational”), i.e. propositional logic. Backwards-chaining can also be used to determine the links between many psychological phenomena, which are often not direct 1:1 correlations.

I think that perhaps what you might be trying to say is that interactions should not be explicitly defined (as with your typical behavior tree), but instead driven implicitly by deeper psychology?

]]>
By: Phil Carlisle/2011/04/19/psychology-and-game-ai/#comment-2991 Phil Carlisle Tue, 19 Apr 2011 19:24:43 +0000 Great experiment and post! You got me thinking about your topic. Great experiment and post! You got me thinking about your topic.

]]>
By: Nick/2011/04/19/psychology-and-game-ai/#comment-2986 Nick Tue, 19 Apr 2011 17:51:22 +0000 Did you check for confounding variables? Did you take into account reflexivity i.e. how do you know they didn't affect you - which in turn affected their response. Haha just kidding. Did you check for confounding variables? Did you take into account reflexivity i.e. how do you know they didn’t affect you – which in turn affected their response.

Haha just kidding.

]]>
By: 3boyutlu/2011/04/19/psychology-and-game-ai/#comment-2977 3boyutlu Tue, 19 Apr 2011 15:00:42 +0000