Personal blog of Harshvardhan Joshi

Yesterday I attended a presentation titled "Can a robot see the way we see?". It was an interesting presentation, made even more interesting by the presenter's enthusiastic delivery (and the fact that he looked a little like Jackie Chan). The case studies involving some rare illusions, though long, were very well received; however they left very little time for the central ideas of consciousness, one's concept of self and whether we can know anything with cetainty. But the most important thing is: the food at the end was good...
In the presentation the case studies were about some of the illusions patients have had about their body, imagining body parts which weren't existent or denying the existence of their own body parts. Interesting as these illusions may be, they are quite rare. Fortunately, most of us don't experience our body in a way that conflicts with other's concept of it (or external reality). However, its quite common for people to experience emotions and not be quite sure about them. In fact (some) psychologists believe that people experience emotional arousal and then search the environment for an explanation. Thus specific emotions are differentiated and labeled based on external cues. (Read more about it and an interesting experiment on my website)
This means that the labels we give to the emotions are selected in face of uncertainty, and the decisions are largely unconscious, with chances of error. For example, everytime I am excited and enjoying a conversation and it turns to food, I feel hungry. And I've known someone who spends some time with a girl in, what he thinks, romantic situations and he is convinced he is in love with her. Like the men in the Dutton & Aaron study, he (mis?)labels his emotions as love. Fortunately for him, he can easily switch labels and move on, but others are not that fortunate and become unhappy. On the other hand, there are people who inspite of strong feelings for someone refuse to call it love and try to trivialize the emotions. Very likely they have acquired a fear of rejection through classical conditioning (which serves an important function of protecting their emotional stability).
So the problem is: the labels we give to emotions are influenced by external factors and we would like to minimize the effects of these external factors to reduce errors in labelling; we cannot identify the emotion with certainty - we can only hope to achieve high confidence levels; we are especially concerned about not making errors with important emotions like love.
Thus we are faced with a typical problem of statistical inference with two possible types of errors. alfa: accepting a hypothesis(H0) when it is not true and beta: rejecting a hypothesis(H0) when it is true. Now, which type of error should we try to reduce (we cannot reduce both). Should we reduce the error of labeling an emotion love when it is not in fact so? Or should we reduce the error of refusing to recognize love? Again statistics can provide the answer. Typically in statistics we select (no change) as the null hypothesis(H0) and its complement as the alternative hypothesis(H1). Thus in this example H0 whould be : the emotion felt is not love. And H1 would be: the emotion experienced is indeed love. And typically in statistics we tend to reduce alpha errors, i.e. accepting H0 when it is not true: concluding that emotion felt is not love when it is so. As a Sanskrit subhashit says: Vachane kah daridrata? - why be parsimonious with words/labels.

------------------------------------------------
An interesting implication of the above theory is that a third person can identify and analyze an emotion with as much accuracy as the person experiencing the emotion - as both use external environment for that purpose. The only information exclusive to the person experiencing the emotion is the extent of emotional arousal or the strength of the feeling (though even this may be measured externally using autonomic response!!).