An attentional theory of emotional influences on risky decisions

It is well known that choices between gambles can depend not only on the probabilities of gains or losses but also on the emotional richness of the items to be gained or lost. Rottenstreich and Hsee (2001) demonstrated that overweighting of low probabilities is magnified if the possible events are emotionally rich, such as a kiss versus an amount of money. Ramirez (2010) showed that persistence in the face of comparable numerically presented losses is greater when the scenario involves taking care of a pet (emotionally richer) versus a business (emotionally poorer). Much of this phenomenon is captured in a neural network model of the Rottenstreich-Hsee data (Levine, 2012). The model is based on interactions among the orbitofrontal cortex, amygdala, cingulate, striatum, thalamus, and premotor cortex that implement categorization of multiattribute vectors representing choice options, in a manner consistent with the gists of fuzzy trace theory. Before categorization, the vectors are weighted by selective attention to attributes that are either emotionally salient or task relevant, with increasing emotional arousal shifting the attentional weights away from numerical attributes such as precise probabilities. This interpretation is supported by the data of Hsee and Rottenstreich (2004) showing that how much participants would pay to save endangered animals is not influenced by the number to be saved if they see pictures but is influenced by the number if they are given verbal descriptions. The theory suggests a few open questions. How are the selective attentional signals represented in the interactions between prefrontal cortex and subcortical areas? Would the salience of numerical attributes still be reduced with high arousal in highly numerate participants? Would the differences between the pet and business scenarios be altered if the positive or negative feedback participants received were shown via pictures rather than numbers?