Imagine yourself going to the cinema. You wonder what film to watch, and you notice that a new film by your favourite director has been released. However, you check the ratings online and another film, from a director you know nothing about, has great reviews. Thus, to make a choice between these two films, you need to arbitrate between two sources of information: past experience (remembering the pleasure you got from previous films of your favourite director) and explicit description (taking into account the extremely positive reviews for the other film).
Despite the fact that, as illustrated by our cinema example, most real-life decisions involve a tension between both components, decisions based on past experience or explicit description have been studied within separate scientific paradigms. The 'experience' paradigm originates in the study of learning by trial-and-error (i.e., reinforcement learning) in non-human animals, where descriptions (involving abstract symbols and language) can play no role. In the 'experience' paradigm, the only source of information is the (often noisy) feedback one receives after making a choice. In contrast, the 'description' paradigm is rooted in experimental economics where testing the assumptions of rationality demands immediate access to complete information. In other terms, the decision variables (namely probabilities and outcomes) are provided prior to the choice in an unambiguous manner.
Interestingly, these paradigms have in common to conceptualize decision-making as a two-step process: first individuals attribute a subjective expected value to each option, and second the options are compared and the one maximizing the subjective expected value is chosen. Coming back to our cinema dilemma, this two-step model would suppose that experience-based (i.e., previous films by your favourite director) and description-based (i.e., the positive ratings for the film of the unknown director) should be compared on the same scale. The 'common currency' hypothesis provides the theoretical justification for such a comparison. It posits that these two kinds (in fact any kind) of values are ultimately mapped into a unified internal scale. The common currency hypothesis, in addition to being a more or less tacit assumption of virtually all economic decision theories, also became dominant in the field of neuro-economics where claims have been made concerning its demonstration at the neural level.
Despite its generalized acceptance, we reasoned that decision problems similar to the cinema dilemma (i.e., mixing up experience to description) could provide a strong empirical test for the common currency hypothesis. To formally and empirically test this hypothesis, we designed a novel paradigm, where some option values were learnt by experience, and subsequently presented against options whose value was described symbolically (called then “hybrid” decisions) (Figure A).
We found that when making such hybrid decisions people largely underweight the information acquired from experience during the learning phase (Figure B). Numerous control measures were employed to ensure that this pattern was not due to insufficient learning, forgetting, or a lack of incentive (nine experiments were run in total). The results show that the tension between experiential and symbolic values is resolved by neglecting value information acquired by experience. The fact that experiential and described values are treated asymmetrically contradicts the common currency assumption. Indeed, if we assume that decisions are made based on values mapped on a unified internal scale, the comparison between experiential and described options should be frictionless.
Overall, this study represents yet another empirical demonstration that the way in which values are represented is inherently dependent on how they are built. Finally, by showing that hybrid decisions (i.e., decisions between experiential and described options) are made neglecting relevant information, this study documents a new possible source of suboptimal or maladaptive decisions in real life.