The tools of New Caledonian (NC) crows offer some of best evidence for cumulative cultural evolution in non-human animals (Hunt & Gray Reference Hunt and Gray2003). These birds manufacture tools from a range of materials for extractive foraging and other purposes (Hunt Reference Hunt1996; St-Clair et al. Reference St Clair, Klump, Sugasawa, Higgott, Colegrave and Rutz2018; Taylor et al. Reference Taylor, Hunt and Gray2012a; Reference Taylor, Miller and Gray2012b; Wimpenny et al. Reference Wimpenny, Weir and Kacelnik2011), and show considerable diversity in tool design across different populations, which cannot be explained by ecological conditions alone (Hunt & Gray Reference Hunt and Gray2003). Some wild NC crows incorporate hooks into their tool designs, a level of technology not seen in our closest relatives, the chimpanzee, and that appears to have only emerged in our species around 100 kya (Yellen et al. Reference Yellen, Brooks, Cornelissen, Mehlman and Stewart1995). In particular, NC crows incorporate naturally occurring barbs into the leaf tools they make from pandanus plants (Hunt Reference Hunt1996; Reference Hunt2000). These pandanus tools are particularly impressive because of their stepped design: crows cut steps into the edge of pandanus leaf tools as they fashion them, with the number of steps cut into the leaf edges showing large variation between populations. It is this variation in tool designs that provides strong evidence for cumulative cultural. These tool designs show evidence of cumulative improvement between populations, are not created by naïve juveniles and the variation between populations is not explained by ecological correlates (Hunt Reference Hunt2000; Hunt & Gray Reference Hunt and Gray2003; Kenward et al. Reference Kenward, Weir, Rutz and Kacelnik2005).

However, work on this bird species has failed to find evidence for many of the social cognitive mechanisms that have been claimed to be required for cumulative technological culture (CTC), such as imitation, teaching, and language (Jelbert et al. Reference Jelbert, Hosking, Taylor and Gray2018; Logan et al. Reference Logan, Breen, Taylor, Gray and Hoppitt2016). For example, Logan et al. (Reference Logan, Breen, Taylor, Gray and Hoppitt2016) presented a classic two-action task to NC crows, but found evidence only of stimulus enhancement and individual trial and error learning, rather than evidence of imitation. There is also currently little evidence for metacognition or theory of mind in this species (Taylor Reference Taylor2014).

Instead, recent work suggests that CTC in this species is best explained by a form of reverse engineering (or goal emulation), which we termed “mental template matching” (Jelbert et al. Reference Jelbert, Hosking, Taylor and Gray2018; Logan et al. Reference Logan, Breen, Taylor, Gray and Hoppitt2016). We have suggested that crows form a mental template of a particular tool design from observing or using tools made by conspecifics, and then reverse engineer this design when they manufacture their own tools. A recent paper-ripping study provides good evidence that NC crows have the cognitive capacity for mental template matching (Jelbert et al. Reference Jelbert, Hosking, Taylor and Gray2018). Crows that learned that either pre-made large or small pieces of card could be inserted into a vending machine to obtain food rewards, subsequently manufactured pieces of card that matched the sizes of the previously rewarded examples. This behaviour occurred even though subjects were rewarded at random for the items they manufactured at test, and there was no available tool design to copy, meaning only the crow's mental representation of the correct tool design could guide manufacture. The presence of this cognitive capacity, coupled with observations showing adult-like pandanus manufacture develops slowly over the first year of life (Holzhaider et al. Reference Holzhaider, Hunt and Gray2010b), that juveniles regularly borrow their parents tools (Holzhaider et al. Reference Holzhaider, Sibley, Taylor, Singh, Gray and Hunt2011), and the parallels between template matching and song bird learning (Jelbert et al. Reference Jelbert, Hosking, Taylor and Gray2018; Slater Reference Slater1983) provide strong support for the hypothesis that NC crows pass on and update tool designs through reverse engineering in the absence of advanced social cognition.

Importantly, although NC crows do not appear to have remarkable social cognition (Jelbert et al. Reference Jelbert, Singh, Gray and Taylor2015a; Reference Jelbert, Taylor and Gray2015b; Logan et al. Reference Logan, Breen, Taylor, Gray and Hoppitt2016), their technical reasoning abilities are substantial and have been the focus of much research. In line with Osiurak and Reynaud's (O&R) technical reasoning hypothesis, there is good evidence for a range of different types of technical intelligence in this species. NC crows show evidence for the ability to transfer causal knowledge between perceptually distinct problems (Taylor et al. Reference Taylor, Hunt, Holzhaider and Gray2007; Reference Taylor, Hunt, Medina and Gray2009), though it is not clear if this is via functional generalisation or true analogical reasoning. They also can mentally preplan solutions to problems (Gruber et al. Reference Gruber, Schiestl, Boeckle, Frohnwieser, Miller, Gray, Clayton and Taylor2019) and make inferences, such as reasoning by exclusion (Jelbert et al. Reference Jelbert, Singh, Gray and Taylor2015a; Reference Jelbert, Taylor and Gray2015b) and inferring the weight of novel objects from observing their movements in the wind (Jelbert et al. Reference Jelbert, Miller, Schiestl, Boeckle, Cheke, Gray and Clayton2019). There is also some evidence that NC crows might be capable of representing hidden causal mechanisms (Taylor et al. Reference Taylor, Hunt and Gray2012a; Reference Taylor, Miller and Gray2012b; but see Boogert et al. Reference Boogert, Arbilly, Muth and Seed2013; Dymond et al. Reference Dymond, Haselgrove and McGregor2013; Taylor et al. Reference Taylor, Miller and Gray2013a; Reference Taylor, Miller and Gray2013b), but there is currently no conclusive evidence that they form representations of unobservable ones (Neilands et al. Reference Neilands, Jelbert, Breen, Schiestl and Taylor2016; Taylor et al. Reference Taylor, Hunt, Medina and Gray2009), nor that these birds possess the ability to make causal interventions (Taylor et al. Reference Taylor, Cheke, Waismeyer, Meltzoff, Miller, Gopnik, Clayton and Gray2014; but see Jacobs et al. Reference Jacobs, von Bayern, Martin-Ordas, Rat-Fischer and Osvath2015; Taylor et al. Reference Taylor, Cheke, Waismeyer, Meltzoff, Miller, Gopnik and Gray2015).

This pattern of results provides strong general support for the technical reasoning hypothesis raised by O&R. NC crows show a profile of high technical reasoning abilities, less developed social intelligence, yet strong indications of CTC. This profile is in line with the hypothesis that CTC initially arises because of technical intelligence, rather than social intelligence. However, it also suggests that not all the technical reasoning skills of humans are necessary and sufficient for CTC. Work on NC crows suggests that CTC can emerge not only in the absence of imitation, teaching, and language, but also in the absence of analogical reasoning, reasoning about unobservable causal mechanisms, and possibly also the ability to make causal interventions. Instead, it appears that a more limited causal understanding, along with functional generalisation, inference, and mental simulation, might be a sufficient suite of technical intelligence skills for cumulative cultural evolution to emerge via template matching. Clearly though, far more work is needed to test this hypothesis further. This is not to say that many other abilities are not required for the full complexity of the cumulative cultural evolution we currently see in humans, but that the necessary and sufficient cognitive requirements for this process to start in a species might be simpler than the full suite of technical reasoning skills O&R suggest.