I think the element in the existing game-theoretic corpus that comes closest to being "emotional" is the idea of the "friendliness of promises and the unfriendliness of threats"

http://econfaculty.gmu.edu/klein/PdfPapers/GameTheoreticRenderingPromisesandThreats.pdf

In mainstream  game theory, players are supposed to be rational. Hence there is no space for emotions.  The Benabou-Tirole theories could be a strting point for your work.

Dear Özgür, thank you for the remarks and references. I found Zizzo (2007) quite interesting.

However, my interest is more into strategic learning (I should state that in the question). I view the topic from an individualistic perspective: from the point of view of the agent, who deals with emotions that affect his strategic behaviour. 

Of course, there is a lot of literature on strategic learning (http://ukcatalogue.oup.com/product/9780199269181.do), but I am wondering what's the best place to look for ideas. Extensions to existing learning algorithms (e.g., no-regret, hypothesis testing, bayesian), for example, seem a good place to start.

I am not sure what exactly you wish to model but a few suggestions:

1. Game theory cannot work without empathy (an emotion!) -- each player HAS to either know or have some beliefs about all other players for game theory to work . So it is not true at all that there is no room for emotion in game theory.

2. You may want to think about emotions as part of the payoff structure. For example utility represents satisfaction or happiness. Game theory payoffs are of course each players utility as a function of another players actions. In other words you happiness depends on what other people do. So it is a great way to capture emotional states.

I am not directly answering your question but adding a new perspective that goes beyond game theory. I am interested in the interplay between game theory and social simulation. More information can be found on the "Behavioural Economics Meets object oriented Simulation" Special Interest Group (BEMooSSIG) website. The presentation section might give you some good ideas.

http://www.cs.nott.ac.uk/~pos/bemoossig/

We also work on a simulator to support peace building activities in Africa where we model the interplay of anger and fear using an agent based approach.

http://www.cs.nott.ac.uk/~pos/projects/simpb/

Have a look at the screenshots (see link at bottom of page). The last screenshot features an agent that considers changes in levels of anger and fear.