Covert signaling is an adaptive communication strategy in diverse populations

Abstract

Identity signals are those common components of communication transmissions that inform receivers of the signaler’s membership (or non-membership) in a subset of individuals. Such signals may be overt, broadcast to all possible receivers, or covert, encrypted so that only similar receivers are likely to perceive their identity-relevant meaning. Here we present an instrumental theory of covert signaling, based on the function of identity signals in social assortment. We argue that covert signaling is favored when signalers are generous toward strangers, when costs of being discovered as dissimilar are high, and when the ability to assort only with preferred partners is restricted. We further argue that covert signaling should be more common among members of "invisible" minorities, who are less likely to encounter similar individuals by chance. We formalize this theory with an evolutionary model to more rigorously explore the consequences of our assumptions. Our results have implications for our understanding of numerous aspects of social life, including communication, cooperation, social identity, humor, pragmatics, politics, hate speech, and the maintenance of diversity.

Strategic Identity Signaling in Heterogeneous Networks

“…We found a weak effect of cost in our experiment, which aligns with the theoretical prediction that higher costs of being identified as dissimilar should increase the relative frequency of covert signaling (Smaldino et al, 2018;Smaldino and Turner, 2021). In these previously published formal models, costs translated into observable differences in payoffs of agents, which affected their likelihood of becoming a target for imitation by others.…”

“…To select tweets that were most likely to be covert or overt, we used a set of criteria drawing on the theoretical framework of Smaldino and colleagues (Smaldino et al, 2018;Smaldino and Turner, 2021). Under the assumption that partisans with extreme views were likely to be highly engaged and therefore more likely to be attuned to covert co-partisan political signals, we expected covert signals to evoke strong responses from extreme partisans, but no consistent response from more moderate partisans.…”

“…Recall that the theory of covert signaling (Smaldino et al, 2018;Smaldino and Turner, 2021) predicts that covert signaling will be more prevalent among 1) individuals in more heterogeneous communities or with minority status, and 2) individuals who face higher costs from being recognized as dissimilar. The mathematical model we present above further predicts that in our specific experiment we should observe a large effect of outgroup proportion, whereby covert signaling increases with the proportion of the audience that consists of outgroup members, and and a small effect of cost, whereby covert signaling increases slightly when costs are high, but primarily when the proportion of the audience that consists of outgroup members is also high.…”

There is no generalizability crisis

“…The lack of specificity and resulting lack of alignment between theory and statistical model may be an instance of deceptive signaling (in the ecological sense, which doesn't imply intent to deceive), where a lack of theoretical rigor is covered up with statistical tests and a recitation of related observed correlations. This maps the generalizability crisis onto analogous problems for which models already exist as starting points, including models of signaling in collaborative environments (Smaldino & Turner, 2020;Smaldino, Flamson, & McElreath, 2018;Tiokhin et al, 2021), the evolution of scientific knowledge on networks (O'Connor & Weatherall, 2018Zollman, 2007Zollman, , 2010Zollman, , 2013, and the effect of prevailing social power on individual choices (Bergstrom, Foster, & Song, 2016;Henrich & Boyd, 2008;Higginson & Munafò, 2016;O'Connor, 2019). With some further development, these models could be used to conduct several "what if things are different" computational experiments under a variety of assumptions to understand what might happen if various interpersonal or institutional changes were instituted.…”

Mechanistic Modeling for the Masses

“…The lack of specificity and resulting lack of alignment between theory and statistical model may be an instance of deceptive signaling (in the ecological sense, which doesn't imply intent to deceive), where a lack of theoretical rigor is covered up with statistical tests and a recitation of related observed correlations. This maps the generalizability crisis onto analogous problems for which models already exist as starting points, including models of signaling in collaborative environments (Smaldino, Flamson, & McElreath, 2018;Smaldino & Turner, 2020;Tiokhin et al, 2021), the evolution of scientific knowledge on networks (Zollman, 2007;2010;2013;O'Connor & Weatherall, 2018;, and the effect of prevailing social power on individual choices (Henrich & Boyd, 2008;Higginson & Munafò, 2016;Bergstrom, Foster, & Song, 2016;O'Connor, 2019). With some further development, these models could be used to conduct several "what if things are different" computational experiments under a variety of assumptions to understand what might happen if various interpersonal or institutional changes were instituted.…”