Theoretical Population Biology 76 (2009) 189–196 Contents lists available at ScienceDirect Theoretical Population Biology journal homepage: www.elsevier.com/locate/tpb Simulated Trust: A cheap social learning strategy Dieter Vanderelst a,∗ , René M.C. Ahn b , Emilia I. Barakova b a Active Perception Lab, University Antwerp, Antwerp, Belgium b Designed Intelligence Group, Eindhoven University of Technology, Eindhoven, The Netherlands article info Article history: Received 3 March 2009 Available online 19 July 2009 Keywords: Social learning Agent-based model Directed social learning abstract Animals use heuristic strategies to determine from which conspecifics to learn socially. This leads to directed social learning. Directed social learning protects them from copying non-adaptive information. So far, the strategies of animals, leading to directed social learning, are assumed to rely on (possibly indirect) inferences about the demonstrator’s success. As an alternative to this assumption, we propose a strategy that only uses self-established estimates of the pay-offs of behavior. We evaluate the strategy in a number of agent-based simulations. Critically, the strategy’s success is warranted by the inclusion of an incremental learning mechanism. Our findings point out new theoretical opportunities to regulate social learning for animals. More broadly, our simulations emphasize the need to include a realistic learning mechanism in game-theoretic studies of social learning strategies, and call for re-evaluation of previous findings. © 2009 Elsevier Inc. All rights reserved. Throughout the animal kingdom, individuals exploit informa- tion that has been gathered by others. Animals ranging from in- vertebrates (Reviewed in Leadbeater and Chittka, 2007; Leadbeater et al., 2006; Fiorito, 2001) to great apes and humans (Tomasello, 1999; Whiten et al., 2007; Bonnie et al., 2006, for example) exhibit forms of social learning. 1 The widespread use of social learning among the taxa is caused by its enormous ecological advantages under many circumstances (See for example Kendal, 2005; Coolen et al., 2005; Bonnie and Earley, 2007, and references therein). Evolution favored social learning because it allows individuals to be flexible and adaptive learners while avoiding the dangers as- sociated with individual exploration (Boyd and Richardson, 1988; Zentall, 2006). Ecologists typically stress the fact that individu- als benefit from copying behavior from others because it saves them the costs of asocial learning (Laland, 2004). Indeed, Zentall (2006) remarked that the behavior of others has often already been shaped by its consequences and might therefore be assumed to be safe to copy. Unsurprisingly, social learning comes in many flavors. Various forms of social learning have been identified (Zentall, 2006) and the underlying physiological mechanisms range from fairly simple to thoroughly complex (Noble and Todd, 2002). However, when studying the dynamics and ecology properties of social learning one can ignore the differences in implementations and consider ∗ Corresponding author. E-mail address: dieter.vanderelst@ua.ac.be (D. Vanderelst). 1 Here, on theoretical grounds, taken to include the use of public information. See Bonnie and Earley (2007) for a discussion. the exchange of information only (Coussi-Korbell and Fragaszy, 1995). This makes it possible to evaluate the advantages of social learning in theoretical studies focusing on the game-theoretic and computational aspects (e.g. Voelkl and No, 2008; Schlag, 1998). This theoretical line of research, supported by empirical find- ings in animal behavior, has shown that the advantage of social learning is by no means universal. In contrast to intuition, so- cial learning is advantageous only if one takes certain precautions (Laland, 2004; Galef and Laland, 2005; Ihara, 2008). Social learn- ing can support the spread, the acquisition and the persistence of maladaptive behavior (Giraldeau et al., 2002). This is because social learners re-use information gathered by others but do not collect new information themselves. Therefore they are implicitly assum- ing that the information they gather from others is reliable. There are circumstances under which this assumption does not hold (Giraldeau et al., 2002; Laland, 2004; Leadbeater and Chittka, 2007; Laland et al., 2005; Galef and Laland, 2005). Second hand informa- tion can be a.o. incomplete, outdated, biased, utterly wrong or al- ready overexploited by others. Theoretical and empirical studies have identified a number of strategies in animals and humans to determine when and from whom to learn socially (for a review see Laland, 2004). These strategies allow to use social learning in a more adaptive fashion, avoiding its potential pitfalls. In this paper we focus on the type of strategies that regulate from which conspecifics animals learn socially (directed social learning strategies or ‘whom’ strategy according to Laland, 2004). These strategies are heuristics that allow animals to determine which animals have information that is interesting to copy. One class of these heuristics try to evaluate the value of the 0040-5809/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.tpb.2009.07.001