Vaesen, K. & Houkes, W. Biol Philos (2017).
Publisher Name: Springer Netherlands
Print ISSN: 0169-3867
Online ISSN: 1572-8404
AbstractThe consensus among cultural evolutionists seems to be that human cultural evolution is cumulative, which is commonly understood in the specific sense that cultural traits, especially technological traits, increase in complexity over generations. Here we argue that there is insufficient credible evidence in favor of or against this technological complexity thesis. For one thing, the few datasets that are available hardly constitute a representative sample. For another, they substantiate very specific, and usually different versions of the complexity thesis or, even worse, do not point to complexity increases. We highlight the problems our findings raise for current work in cultural-evolutionary theory, and present various suggestions for future research.
The evidence for and against version (c) of the technological complexity thesis (i.e., ‘overall increase’) appears to be scarce and inconclusive. This finding is hard to reconcile with what many of us believe, often quite strongly, about progress in science and technology.
So, where does this consensus view come from? McShea (1991) offers several hypotheses regarding the origins of the consensus view in biology, and these seem to be sensible also with respect to technology. Perhaps we possess a cognitive algorithm that correctly outputs increasing complexity, but employs a conception of complexity which differs from those used in current empirical work. Perhaps items simply look more and more different from contemporary ones as we go further and further back in time; if contemporary items are assumed to be complex, less familiar items would mistakenly be judged simpler. Or perhaps we are biased towards a few spectacular cases (e.g., the Apollo mission), and accordingly get the impression of a universal, long-term trend.
To the extent that one finds plausible the analogy between biological and technological evolution, our findings are also hard to reconcile with theoretical work in biology. This work has suggested that even if complexity in every lineage follows a simple random walk (i.e., complexity decreases are as likely as complexity increases), and if there is a lower bound to complexity, the mean complexity of all lineages will rise (Fisher 1986; McShea 1991, 1994); and it has suggested that in the absence of constraints or forces, complexity will increase in a lawful manner (McShea and Brandon 2010).
There are obvious reasons to mistrust the analogy between biological and cultural evolution. McShea and Brandon (2010, p. 132), for instance, find it plausible that in cultural evolution quite a few homogenizing forces are at work. Consequently, trends in technological complexity plausibly will not conform to a law that is developed for cases in which no forces are at work. Likewise, even if Fisher’s work would theoretically establish the complexity thesis in biology, it would only apply to technological evolution if none of the following defeating conditions obtains: decreases being more likely than increases, decreases being larger in size than increases, and there not being a lower bound to complexity. Since none of these conditions can be ruled out empirically, theoretical work in biology certainly leaves open the possibility of finding conclusive evidence against the complexity thesis in technology.
What Fisher’s paper does show, however, is that complexity, at least theoretically, can increase even in the absence of high-fidelity transmission. Even when cultural learners never copy the exact behavior of their cultural parent (e.g., complexity increases and decreases have equal probabilities of 0.5), Fisher’s work predicts that, in the absence of defeating conditions, complexity will increase. Hence, from the mere fact that human culture is characterized by the accumulation of complexity one should not infer, as some authors seem to do (e.g., Tomasello and Call 1997; Tomasello 1999; Boyd and Richerson 1996; Richerson and Boyd 2008; Mesoudi et al. 2013), that such accumulation is the result of imitation, i.e., faithfully replicating actual behavior of a mentor.
An alternative for advocates of the technological complexity thesis is to abandon version (c), and focus instead on versions (a) (i.e., ‘Presence: human cultural evolution has resulted in complex technologies) and (b) (i.e., ‘Threshold’: human cultural evolution has resulted in technologies that are so complex that they cannot be the result of individual learning). As indicated above, at least one of these—version (b)—is still associated with the salient explananda regarding humaniqueness. Yet as we made clear in Sect. 2, endorsing version (b) does not reduce the need for evidential support. On the contrary, it requires a non-arbitrary specification of a threshold value—and version (a) requires clear application criteria for the labels ‘simple’ and ‘complex’. Moreover, as argued in the previous section, substantiating claims of humaniqueness would require an unbiased test of non-human cultural traits with regard to whichever application criteria or threshold value has been established. Both McShea’s hypotheses concerning the origins of the consensus view in biology and associations with the problematic distinction between ‘primitive’ and ‘advanced’ cultures give reasons for being cautious regarding version (a). Although—or perhaps precisely because—the ‘presence’ claim seems intuitively obvious, there is a real risk that the claim merely expresses perceptions of readily observable but superficial differences, or ignorance on the part of the person claiming complexity for her own cultural traits.
In this regard, operationalizing version (a) through the ‘beyond-individual-invention-from-scratch’ version (b) may look like a promising way forward. For this version, however, our discussion of Dean et al. (2014)’s claims regarding humaniqueness points out some of the complications: as obvious as (b) might seem for Jumbo Jets or smartphones, comparison between human and non-human culture needs to be on an equal (i.e., artifact vs artifact or behavior vs behavior) footing. Moreover, and in closing, the evidential basis for (b) should not be overrated. For most technologies, re-invention from scratch is practically unnecessary, but perhaps not beyond individual capacity. Not coincidentally, cultural evolutionists typically provide indirect evidence, in the form of ‘lost European explorer’ narratives (e.g., Boyd et al. 2013): historical anecdotes of expeditions of well-trained, presumably intelligent people failing in environments where indigenous people manage to survive; here, a combination of motivation, available resources and cognitive capacities was presumably insufficient to re-invent local subsistence technologies from scratch. A more rigorous search for evidence would need to establish which technologies lie outside what has recently been called the ‘zone of latent solutions’ (Tennie et al. 2016): behavioral patterns that may be triggered in sufficiently motivated individuals by social and environmental cues. Scenarios can be imagined in which this may be tested for various technologies. One is the “Island Test” proposed by Tomasello (1999): a scenario in which someone is separated at birth and raised alone on an island, but provided with sufficient motivation and raw materials to trigger and facilitate re-invention of a technology in case it is in the ‘zone of latent solutions’. As the difficulties in realizing this scenario should make clear, currently even our knowledge regarding this form of technological complexity is easily overstated.