When modelling emotion expression, can "felt" and "expressed" emotions be distinguished?

This is a question in current ECA systems.
comments: Cause and Effect type descriptions --RichardCraggs, Thu, 19 Feb 2004 12:40:46 +0100 reply
I believe that this question is quite closely related to Roddy Cowie's notion of `Cause' and `Effect' type descriptions of emotions, given in "Describing the emotional states expressed in speech" in SpeechEmotion 2000. The distinction made there is between emotion as experienced by the speaker and the affect that this has on the listener. The applications of studies into the two are likely to be quite different. Perhaps the question could be rephrased as "is it possible to see past a person's external expression, and recognise the underlying emotion". I suggest that the answer is "unlikely" but more importantly I believe that is often not a particularly useful thing to do. I've discussed this point in a little depth in an upcoming paper - http://www.cs.man.ac.uk/~craggsr7/papers/sss404CraggsR_2D_scheme.pdf