The replicable, independent observations required for empirical science are possible only if different scientists are capable of making the same observations (Kuhn,
1970). Distance illusions are among the most common devices for testing perceptual capacity across individuals (Biggs, Stey, Davoli, Lapsley, & Brockmole,
2014; Jackson,
2009; Tinker,
1938). Although illusions depart from reality, the constancy of their experience across individuals reveals the underlying brain mechanisms that humans employ (Chua & Enns,
2005; Eagleman,
2001; Goodale, Milner, Jakobson, & Carey,
1991). From the study of distance illusions, researchers regularly draw conclusions about a singular human nature and consider distance illusions to be broadly constant, universal cognitive mechanisms within our species (Eaglemen,
2001; Henrich, Heine, & Nornzayan,
2010; Wu, Ooi, & He,
2004).
There is little evidence of universal distance illusions. The majority of perception research has used nonrepresentative samples of participants from Western, educated, industrialized, rich democracies (Henrich et al.,
2010). Of the cross-cultural data available, distance illusions either occur at vastly different magnitudes or fail to replicate (Caparos et al.,
2012; Henrich et al.,
2010; Pollnac,
1977; van der Kamp, Withagen, & de Wit,
2013; Wober,
1970). This failure of cross-cultural replication includes the most well-known illusions tested over the last 110 years (Rivers,
1905). These investigations typically use theories that do not identify species-level evolutionary forces and thus have little predictive power for identifying species-typical behavior (Barkow, Cosmides, & Tooby,
1995).
In the following study, we tested the predictions of an evolutionary theory across disparate human groups to identify a universal distance illusion. No single study can settle the broader issues of replicability. Here, our goal was to identify the capacity for replicability in a largely unreplicated research area by using an ultimate-level theory. Ultimate-level theories are especially important for generating predictions and specifying phylogenetic, developmental, and environmental contexts for phenomena (Barkow et al.,
1995; Marr,
1982; Tinbergen,
1963). Part of the lack of previous replications in this area (see Hutchison & Loomis,
2006; Wagner,
1977) likely stems from mechanistic explanations that do not specify the environments in which phenomena evolved, and thus, they fail to make explicit predictions across the settings of which culture is a part.
We selected participants who could provide a nearly diametric comparison to the populations tested in the social sciences and in distance perception research in particular. Our participants were Ixil Mayan individuals from the area surrounding Nebaj, El Quiche, Guatemala. Ixiles have experienced little exposure to Western media, have little to no formal education, live in relative poverty, and have a very different experience of the natural world than do members of highly industrialized nations. The likelihood of coincidental replication across cultures, particularly across Ixil and the United States, is exceedingly small—far smaller than the replication likelihood seen in psychology experiments tested within the same culture (Open Science Collaboration,
2015).
We tested participants for the
descent illusion, predicted by evolved navigation theory (Jackson,
2005; Jackson & Cormack,
2007). Evolution researchers propose that selection in the form of navigational mortality resulted in distance overestimation of vertical surfaces, especially from above, where the risk of falling is greatest (Jackson & Cormack,
2007,
2008,
2010). This illusion occurs such that observers overestimate the height of a vertical surface and to a greater extent while standing at its top than bottom. The illusion gives positive values for (
et –
eb)/
eb, where
et is the observer's estimate of the height of a vertical surface while standing at its top and
eb is the same estimate while standing at its bottom. The magnitude of the descent illusion in U.S. samples is roughly 27% (Jackson & Cormack,
2007). Illusion magnitude is as high as 84% when we compare
et to the actual distance, making this the largest known real-world distance illusion (for comparison, see Chapanis & Mankin,
1967; Higashiyama & Ueyama,
1988; Howe & Purves,
2005).