John Bargh  is now pushing the idea of "The Selfish goal" where our mind is composed of various half-independent goals each of whihc is actually caring on its own, and that our actions and behavior is the composite of all this.
It made me thinking about "selfish opinions" implying that we have multiple competing opinions perceptions etc. which create the final opinion by aggregation. Just like Bargh's goals theory.
That way overstated "irrational" intuitions may combine to an optimal aggregated opinion.
Compatible with what we know that people are quite overreacting with specific questions (like stereotypes or heuristics generally). But these opinions are usually reality compatible (most stereotypes are true says literature).
When one makes a decision his mind sum up a bunch of overstated discrete-like opinions, and the decision is their (kind of) sum. An extreme statement when mixed with twenty other opinions may turn out quite representing reality. (The overstated opinion is actually diluted to 5% and in this amount it corresponds to the real values).
Generally this is compatible with the idea of ecological rationality of Greg Gigerenzer and dan Goldstein. In this view, heuristics that may not be "correct" in laboratory settings (like in Kahneman & Tversky's experiments) may still be very efficient for real life situations, because in the complexity of life there are various parameters etc. that make our seemingly incorrect intuitions reasonably efficient.
 (father of the research showing the enormous power of the unconsciousness in making us do things taht look exactly the same as if we did them consciousely + showing how easy it is to make people do things by unconscious means like showing flags (Hassin) putting a suitcase mnear the door or giving them to play with words)