By integrating over personal and social information sources, uncertainty can be reduced (Morgan et al., 2011, Rendell et al., 2011 and Toelch et al., 2009). The behavior of competitors could thus serve as a proxy for the common value (Beggs and Graddy, 2009, Campbell-Meiklejohn et al., 2010, Hayes et al.,

1995, Nicolle et al., 2012 and Suzuki et al., 2012), particularly when uncertainty is high, social sources and social dynamics are used to update private values (Berns et al., 2010, Rendell et al., 2011, Toelch et al., 2010 and Toelch et al., 2009). Despite the recognition of competition as a social process, the interplay between competition and changes to private value estimates has received little attention. One reason is that many competition experiments Roxadustat chemical structure are common value auctions where signals about the common value are induced (Rutström, 1998) and symmetrical (Kagel & Levin, 2008). In

common value auctions, social cues (competitor SB203580 bids) carry no information, a case rarely occurring under non-laboratory conditions with auctions mainly being private value auctions. Here, we investigate an important interaction between differences in (ex ante) private values and the effect of subsequent competition on individuals’ (ex post) private value estimate. We specifically test how private values for real items are influenced by the bidding behavior in a two player multiple item repeated all-pay auction game. Crucially, we manipulated auctions such that participants encountered real competitors with lower, approximately equal, or higher private value estimates. As participants bid repeatedly and possibly opted out of the auction by bidding nothing, bids during these auctions potentially deviated from private value estimates. To account for this, we used preference 1 statements as a proxy for participants’ private value estimates ( Warren, McGraw, & Van Boven, 2011). We specifically investigated how preference ranks of the auction items changed because of both the overall level of competition and the dynamics of the auctions across the session. For this, participants ranked items by preference before and after the game.

We then linked behavioral parameters from the bid progression within auctions to participants’ propensity to change their preference for a particular item. Participants were Interleukin-2 receptor recruited from a local participant pool via email invitation. In total 42 (17 male) participants played the game in pairs of two with a maximum of four players per session (10 same gender pairs and 11 mixed gender pairs; sample size calculations can be found in the SI). After the experiment, participants answered a questionnaire where we collected background information like age and gender. Additionally we asked participants to give verbatim description of their strategies during the game. All procedures comply with APA guidelines and were approved by the Ethics board at Charité University hospital (EA1/212/11).