Note that in our experiment, participants could ride the bubble, but not directly influence its formation, due to the nature of the experimental design. However, this situation is analogous to real financial markets in which the action of a single trader very rarely has a detectable impact on the whole market. We then sought to clarify the role played in this process by participants’ attempts to forecast the
intentions of other players or of the market as an intentional agent. In fact, while standard financial theory assumes that competitive markets are nonstrategic, it is Trichostatin A concentration not uncommon for people to assign intentionality to markets. Financial commentators often say, anthropomorphically, that “markets are panicking” or “markets are losing confidence.” Assigning intention or agency is a natural way for humans to model and interpret complex behavior (as in the case of simple societies in which human-like gods are thought to control natural processes such as the weather). Humans
live in social environments and therefore usually benefit from ToM abilities that allow them to forecast check details the intentions of others and take preventive actions (Fehr and Camerer, 2007, Frith and Frith, 1999, Gallagher and Frith, 2003 and Sanfey, 2007), an ability instantiated in medial prefrontal cortex (dmPFC) (Amodio and Frith, 2006 and Frith and Frith, 2006). Using an independent ToM task (Baron Levetiracetam Cohen et al., 2001), we showed that the increase of activity isolated during the bubble markets correlates with the individual ability in ToM. Furthermore, we showed that the functional coupling between dmPFC and vmPFC was increased during bubble markets.
We interpreted these results by proposing a putative mechanism that produces the increase in value sensitivity that we observed in vmPFC while participants traded in the context of bubble markets. These data suggest that during financial bubbles, participants are taking into account the intention of other players in the market (or of the market as whole) while updating their value estimates, and that this effect is mediated by the interaction between dmPFC and vmPFC. This interpretation fits with previous studies that have highlighted the role of dmPFC in shaping value computation by showing that social signals change the way in which values are updated through reinforcement learning (Behrens et al., 2008, Hampton et al., 2008, Behrens et al., 2009 and Suzuki et al., 2012). For example, activity in dmPFC correlates with the likelihood that participants playing a “work-or-shirk” strategic game learn the value of an action using a model that takes into consideration the intentions of the other players in the game (Hampton et al., 2008). A recent study by Nicolle and colleagues (Nicolle et al.