Great post as usual, Lionel. I hadn’t considered that there are also epistemic benefits to social comparison (gaining knowledge about what’s possible for me to achieve) independent of status concerns—that’s a good insight. But it still leaves the puzzle unsolved as to how we determine the reference group. It seems like the optimal way to…
Great post as usual, Lionel. I hadn’t considered that there are also epistemic benefits to social comparison (gaining knowledge about what’s possible for me to achieve) independent of status concerns—that’s a good insight. But it still leaves the puzzle unsolved as to how we determine the reference group. It seems like the optimal way to choose a reference group—at least to fulfill the epistemic function—is to select people who are as similar as possible on whatever dimensions are most causally relevant to achieving the goal in question. We may even have different reference groups for different goals. I may not care if my tennis equal (in terms of skill level), gets a promotion at his job, but it stings if he wins a tournament and I don’t. So there may not be one, singular reference group, but it may be tailored to the specific goal under consideration, focusing on whatever variables are most relevant for the achievement of that goal.
Hi David, thanks! The insight about the epistemic benefits was presented by Luis Rayo and Gary Becker in 2007 (https://www.journals.uchicago.edu/doi/abs/10.1086/516737). In their paper, they looked at the case where you have one group of peers whose achievements are informative. In our recent paper with Greg, cited in the post, we consider any type of information that would be informative. Implicitly, we have in mind exactly what you say: in terms of peers, we should identify the people most relevant for us, along the lines of the prediction of Festinger and the quote from Russell. We are working on some aspects of this process, assuming agents are Bayesian, as an extension.
Great post as usual, Lionel. I hadn’t considered that there are also epistemic benefits to social comparison (gaining knowledge about what’s possible for me to achieve) independent of status concerns—that’s a good insight. But it still leaves the puzzle unsolved as to how we determine the reference group. It seems like the optimal way to choose a reference group—at least to fulfill the epistemic function—is to select people who are as similar as possible on whatever dimensions are most causally relevant to achieving the goal in question. We may even have different reference groups for different goals. I may not care if my tennis equal (in terms of skill level), gets a promotion at his job, but it stings if he wins a tournament and I don’t. So there may not be one, singular reference group, but it may be tailored to the specific goal under consideration, focusing on whatever variables are most relevant for the achievement of that goal.
Hi David, thanks! The insight about the epistemic benefits was presented by Luis Rayo and Gary Becker in 2007 (https://www.journals.uchicago.edu/doi/abs/10.1086/516737). In their paper, they looked at the case where you have one group of peers whose achievements are informative. In our recent paper with Greg, cited in the post, we consider any type of information that would be informative. Implicitly, we have in mind exactly what you say: in terms of peers, we should identify the people most relevant for us, along the lines of the prediction of Festinger and the quote from Russell. We are working on some aspects of this process, assuming agents are Bayesian, as an extension.
Very cool, thanks! Will check out those papers.