Periodic Reporting for period 4 - STONECULT (Do early stone tools indicate a hominin ability to accumulate culture?)
Période du rapport: 2021-10-01 au 2022-09-30
What is culture? Culture requires at least one variant of a “social learning” mechanism to influence at least the frequency – or even the form (the specific design) – of behaviours or artefacts in a population. Previous research has argued that similar social learning mechanisms underlie modern human and early hominin technology. However, early hominin culture shows periods of stasis, where modern human culture is instead cumulative – even explosive. This cumulative culture is powered and enabled by special variants of social learning mechanisms – namely those that can copy forms. Form copying even necessarily leads to cumulative culture - leading to “path-dependent” form changes and, eventually, to a complete dependency of the resulting form variants on copying (copying-dependent forms). It follows that the assumption of the early presence of these specific copying skills therefore is at odds with the observed patterns of stasis for early stone tools.
Instead of copying even copying-dependent forms from others (as modern humans do), contemporary ape cultures are based on non-copying variants of social learning. These variants still produce cultural patterns as they mediate the relative frequencies of forms between populations. Yet, ape cultures are based on and fuelled by "socially mediated individual reinnovations" instead of form copying. These reinnovations consist of forms that apes can in principle recreate on their own (“latent solutions”; Tennie et al. 2009, 2020). Again, unlike copying, such learning do not lead to culture-dependent forms. This latent solutions approach is thus a core candidate to account for early hominin stone tools because it provides an actual explanation for the stasis observed in the early stone tool archaeological record.
The STONECULT project tested whether early stone tools may be most parsimoniously regarded as manifestations of cumulative culture or whether they are best accounted for by the latent solutions model. STONECULT evaluates whether early stone tools were more similar either to modern ape (i.e. restricted to latent solutions) or modern human technologies (i.e. form-copying-dependent cumulative culture). In this way, STONECULT aimed help pinpoint when cumulative culture first started in our lineage.
Using a general triangulation of the question, STONECULT tested for latent solutions underlying early stone tools across three objectives/approaches. Objective 1 tested apes, while Objective 2 tested modern humans, for early stone tool reinnovations. Objective 3 was designed to tested for the unintentional, automatic production of more specific, later stone tool forms via the unsupervised outputs of an automatic "virtual knapper" software program that the STONECULT project was supposed to create.
Instead of copying even copying-dependent forms from others (as modern humans do), contemporary ape cultures are based on non-copying variants of social learning. These variants still produce cultural patterns as they mediate the relative frequencies of forms between populations. Yet, ape cultures are based on and fuelled by "socially mediated individual reinnovations" instead of form copying. These reinnovations consist of forms that apes can in principle recreate on their own (“latent solutions”; Tennie et al. 2009, 2020). Again, unlike copying, such learning do not lead to culture-dependent forms. This latent solutions approach is thus a core candidate to account for early hominin stone tools because it provides an actual explanation for the stasis observed in the early stone tool archaeological record.
The STONECULT project tested whether early stone tools may be most parsimoniously regarded as manifestations of cumulative culture or whether they are best accounted for by the latent solutions model. STONECULT evaluates whether early stone tools were more similar either to modern ape (i.e. restricted to latent solutions) or modern human technologies (i.e. form-copying-dependent cumulative culture). In this way, STONECULT aimed help pinpoint when cumulative culture first started in our lineage.
Using a general triangulation of the question, STONECULT tested for latent solutions underlying early stone tools across three objectives/approaches. Objective 1 tested apes, while Objective 2 tested modern humans, for early stone tool reinnovations. Objective 3 was designed to tested for the unintentional, automatic production of more specific, later stone tool forms via the unsupervised outputs of an automatic "virtual knapper" software program that the STONECULT project was supposed to create.
The theoretical foundation to test the suitability of the latent solutions model for early stone tools (the target of STONECULT) was published as a Forum Article at the start of the project (Tennie et al. 2017 Current Anthropology). This article made clear that the latent solution model provides a parsimonious alternative to a currently widespread null hypothesis for early stone tools (the former null being that early stone tool forms were and had to be copied in their behavioural and/or artefact forms).
The empirical tests were mainly performed according to the triangulation approach taken by the STONECULT project:
In Objective 1 we tested apes for their spontaneous abilities with regard to stone tool use and production know-how. We found - for the first time ever - that untrained, unenculturated apes (in this case, orangutans) can spontaneously make stone tools and that they can also spontaneously use stone tools as tools for cutting (Motes-Rodrigo et al. 2022, PLOS ONE). Chimpanzees tested failed to spontaneously make or use stone tools (Bandini et al. 2021, ORE).
In Objective 2, we tested humans for their spontaneous abilities with regard to stone tool use and production know-how. No previous study has ever tested humans in such a pure "baseline" condition (i.e. in the entire absence of any type of a form model). We found that naive human participants can spontaneously make and use stone tools. These form-naive participants re-innovated all four early stone tool production know-hows on the spot (Snyder et al. 2022, Sci Adv).
Objective 3 was designed to test for the unintentional, automatic production of early stone tool forms via the unsupervised outputs of a "virtual knapper" 3D software program under various "know-how" rule settings. STONECULT did not achieve this, as the necessary software program remained in a prototypical - proof-of-principle – stage (Orellana Figueroa et al. 2021, Sci Rep). However, this proof-of-principle created a path that can now be realistically followed to create a functioning virtual knapping software
The empirical tests were mainly performed according to the triangulation approach taken by the STONECULT project:
In Objective 1 we tested apes for their spontaneous abilities with regard to stone tool use and production know-how. We found - for the first time ever - that untrained, unenculturated apes (in this case, orangutans) can spontaneously make stone tools and that they can also spontaneously use stone tools as tools for cutting (Motes-Rodrigo et al. 2022, PLOS ONE). Chimpanzees tested failed to spontaneously make or use stone tools (Bandini et al. 2021, ORE).
In Objective 2, we tested humans for their spontaneous abilities with regard to stone tool use and production know-how. No previous study has ever tested humans in such a pure "baseline" condition (i.e. in the entire absence of any type of a form model). We found that naive human participants can spontaneously make and use stone tools. These form-naive participants re-innovated all four early stone tool production know-hows on the spot (Snyder et al. 2022, Sci Adv).
Objective 3 was designed to test for the unintentional, automatic production of early stone tool forms via the unsupervised outputs of a "virtual knapper" 3D software program under various "know-how" rule settings. STONECULT did not achieve this, as the necessary software program remained in a prototypical - proof-of-principle – stage (Orellana Figueroa et al. 2021, Sci Rep). However, this proof-of-principle created a path that can now be realistically followed to create a functioning virtual knapping software
All three objectives of the STONECULT grant achieved outcomes beyond the state of the art - though to varying degrees (see below).
Objective 1: Our results changed the status quo on ape abilities to produce and use sharp stone flakes (i.e. corresponding to early stone tools; henceforth stone tools). Formerly, based on earlier ape stone tool studies, it was sometimes thought that apes are able to do so regularly. Our experiments have shown instead that apes rarely if ever make or use stone tools (e.g. Bandini et al. 2021; Motes-Rodrigo et al. 2022). believe that differences in training and enculturation overall best explain the observed differences in performances (Bandini et al. 2021; Motes-Rodrigo et al. 2022). Overall, we therefore favour the hypothesis that human demonstrations alone (as in previous studies) do not suffice to make apes reliable stone tool users or makers.
Objective 2: Here we found that the often-held assumption that humans require form models of some type(s) to copy early stone tool production know-how ("early knapping techniques") is untrue. When we tested knapping-naive human participants - naivety established via post-test questionnaires - many proved spontaneously able to knap in the entire absence of such models (Snyder et al. 2022). Together with primate data (our orangutan data above, but also capuchin test data from the 1990s by Westergaard and Suomi) this data strongly indicates that knapping forms do not logically require form copying - indeed, none of the earliest knapping techniques logically requires such copying. This is direct evidence for the latent solution model of early stone tools (sensu Tennie et al. 2016, 2017).
Objective 3: We failed to produce a fully functioning, realistic virtual knapping software - which meant that we could not then use it to test various ways in which various later stone tool shapes came about. However, we were still successful in achieving the creation of a feasible software production pipeline for a future virtual knapper software (Orellana Figueroa et al. 2021).
Also in many other ways the STONECULT project further advanced the state of the art (see project website for these).
Objective 1: Our results changed the status quo on ape abilities to produce and use sharp stone flakes (i.e. corresponding to early stone tools; henceforth stone tools). Formerly, based on earlier ape stone tool studies, it was sometimes thought that apes are able to do so regularly. Our experiments have shown instead that apes rarely if ever make or use stone tools (e.g. Bandini et al. 2021; Motes-Rodrigo et al. 2022). believe that differences in training and enculturation overall best explain the observed differences in performances (Bandini et al. 2021; Motes-Rodrigo et al. 2022). Overall, we therefore favour the hypothesis that human demonstrations alone (as in previous studies) do not suffice to make apes reliable stone tool users or makers.
Objective 2: Here we found that the often-held assumption that humans require form models of some type(s) to copy early stone tool production know-how ("early knapping techniques") is untrue. When we tested knapping-naive human participants - naivety established via post-test questionnaires - many proved spontaneously able to knap in the entire absence of such models (Snyder et al. 2022). Together with primate data (our orangutan data above, but also capuchin test data from the 1990s by Westergaard and Suomi) this data strongly indicates that knapping forms do not logically require form copying - indeed, none of the earliest knapping techniques logically requires such copying. This is direct evidence for the latent solution model of early stone tools (sensu Tennie et al. 2016, 2017).
Objective 3: We failed to produce a fully functioning, realistic virtual knapping software - which meant that we could not then use it to test various ways in which various later stone tool shapes came about. However, we were still successful in achieving the creation of a feasible software production pipeline for a future virtual knapper software (Orellana Figueroa et al. 2021).
Also in many other ways the STONECULT project further advanced the state of the art (see project website for these).