The dynamics of the transition between late Neanderthals and anatomically modern humans (AMHs), also referred to as the Middle Paleolithic to Upper Paleolithic transition (MUPT), is still controversial. The location and duration of the coexistence of the two human populations, as well as their possible interactions and cultural exchanges, are still debated in the scientific community (d'Errico, 2003; Hublin, 2015; Talamo et al., 2020; Djakovic et al., 2022). The precise chronological position of the different cultural facies and the human remains associated with them delineate the period within which Neanderthals and AMHs could have interacted. While evidence of interbreeding between Neanderthals, Denisovans and/or AMHs have been documented elsewhere (Fu et al., 2015; Slon et al., 2017; Massilani et al., 2020; Hajdinjak et al., 2021; Prüfer et al., 2021), northern and western Europe seem to have been populated only by Neanderthals until around 45,000–42,000 cal BP (Nigst et al., 2014; Hublin, 2015; Devièse et al., 2021). Moreover, genetic analyses show the absence of genetic flow from early AMH to late Neanderthal populations (Hajdinjak et al., 2018) as well as the absence of Neanderthal genes in early Upper Paleolithic modern humans in northern Europe (Posth et al., 2016). However, this interpretation is based on a limited number of hominin specimens because of their scarcity in the archeological record. Mousterian and Aurignacian industries, associated with Neanderthals and AMHs respectively, are found in much larger quantities in the archeological record, and can also be used to define the timing of both occupations.

Numerous sites excavated in Belgium since the 19th Century (Supplementary Online Material [SOM] S1) are attributed to the Mousterian (Ulrix-Closset, 1975; Di Modica, 2010; Di Modica et al., 2016) and the Aurignacian (Otte, 1979; Dinnis and Flas, 2016). There are also two occurrences related to the Lincombian-Ranisian-Jerzmanowician (LRJ) transitional technocomplex (Flas, 2011). However, most of these historical collections lack reliable contextual data (Pirson et al., 2012; Di Modica et al., 2016; Abrams, 2018). Before the present study, the most recent dates for the late Mousterian industries in Belgium came from recent excavations at Scladina and Walou caves, suggesting that the Mousterian ended around 42,000 cal BP (Fig. 1). Unfortunately, the dates were not produced on anthropogenically modified bones, which date human presence unequivocally compared to unmodified bones from the archeological contexts which rely on the interpretation of site formation processes (Pirson et al., 2012). The oldest Aurignacian occurrence in Belgium was established with a bone point from Spy Cave dated to 38,100–36,500 cal BP (Pirson et al., 2012; Flas et al., 2013). This data suggests a 4000-year hiatus between the end of the late Mousterian and early Aurignacian. However, given the high C/N atomic ratio (3.6) of the bone point from Spy Cave, this date should be considered at best as a minimum age (Pirson et al., 2012; Flas et al., 2013; Dinnis and Flas, 2016). Radiocarbon dates obtained by Semal and colleagues on the Spy Cave Neanderthals partially filled this 4000-year hiatus (Semal et al., 2009). The survival of late Neanderthals beyond the Mousterian allowed their hypothetical association with the LRJ, for which archeological evidence was found in the historical archeological record (Semal et al., 2009; Flas, 2011). Based on this data, the proposed scenario for the MUPT in Belgium was that the Mousterian ended around 42,000 cal BP, was followed by the LRJ (≈40,000 cal BP) and that the Aurignacian started around 38,000 cal BP.

To refine the chronology of the transition between the Mousterian and Aurignacian cultures and to discuss the occupation of northwestern Europe during this period of cultural and human transition, Neanderthal remains were first (re)dated. The model highlighted their possible disappearance between 44,200 and 40,600 cal BP (at 95.4% probability; Devièse et al., 2021). In this study, we now focus on modified bones, which constitute, along with human remains, the best material to directly date human activities. Even though the use of bone retouchers persists without major modifications throughout the entire Paleolithic (Tartar, 2012; Jéquier et al., 2018; Toniato et al., 2018), the use of hard animal materials was witness to profound changes during the MUPT, highlighted by the production of ornaments (d'Errico, 2003; Caron et al., 2011; Arrighi et al., 2019; Hublin et al., 2020) and weaponry, such as the bone points (Tartar and White, 2013; Kitagawa and Conard, 2020).

Not only the choice of the material to radiocarbon date is important, but also the method to prepare the samples before measurement. This is illustrated, for example, by the recent radiocarbon dating study on Belgian Neanderthals including those from the caves of Spy, Engis and Fonds-de-Forêt. Using the compound specific radiocarbon analysis (CSRA) approach targeting the amino acid hydroxyproline, it was demonstrated, as suspected, that some of the previous dates produced on collagen for the Neanderthal specimens from Spy Cave were inaccurately young (by up to 10,000 years) due to the presence of unremoved contamination including conservation materials (Devièse et al., 2021).

In this study, we report a series of unpublished radiocarbon dates on bone tools that were obtained on collagen and a new set of dates on modified bones obtained by compound specific radiocarbon analysis (Table 1). Some of the artifacts were cross-dated using both pretreatment methods, which allows us to discuss the reliability of all the results obtained on the bone artifacts, such as the bone point from Spy Cave.