Figure 5. Orthogonal views of the 3D MDS results for the Jaccard distances using the SO character subset. Clustering with the HN subset of characters was easily the worst of all character subsets for both simple matching and Jaccard distances. Distance correlation with simple matching distances revealed four poorly-defined clusters with an average silhouette width of -0.01 (Figure 22). Homo sapiens clusters with chimpanzee and gorilla. Neandertals cluster with Par. aethiopicus, Par. boisei, Sahelanthropus, and Au. africanus. Medoid partitioning, while better than the distance correlation clusters, still produced poor results. Two-cluster medoid partition had an average silhouette width of 0.15, and three-cluster medoid partition had an average silhouette width of 0.26. In both partitions, H. sapiens clustered with Au. afarensis and two Paranthropus species. Fuzzy analysis produced a two-cluster hard partition with an average silhouette width of 0.25 and a three-cluster partition with an average silhouette width of 0.26. In both fuzzy partitions, H. sapiens clusters with Par. boisei. Three-dimensional MDS of these simple matching distances revealed no obvious clusters (Figure 23). Clustering using Jaccard distances from the HN characters was not notably different from the clustering based on simple matching distances (Figure 24). Distance correlation again revealed a four-cluster partition with an average silhouette width of 0.02, and the cluster partition of taxa was identical to that derived from simple matching distance correlation. Medoid partitioning and fuzzy analysis were again quite poor with no average silhouette width exceeding 0.25. Three-dimensional MDS of these simple matching distances revealed no obvious clusters (Figure 25). A summary of all sixty cluster partitions is shown in Figure 26. Leaving aside the spurious results of the HN character subset, we see that despite repeated observation of odd clusterings (e.g., Homo sapiens with Paranthropus), we do find a core of taxa that tend to cluster with Homo sapiens. These taxa include Neandertals, H. heidelbergensis, African H. erectus, Georgian H. erectus, and H. naledi, all of which cluster with Homo sapiens exclusively using all character subsets except the HN characters. In addition, Asian H. erectus, H. habilis, H. rudolfensis, and H. antecessor nearly always cluster with Homo sapiens. Of the fifty cluster analyses involving the SO, LC, EAO, FA, and PO character subsets, Au. sediba clusters with H. sapiens 46 times. In contrast, H. floresiensis clusters with Homo sapiens only eight times, four times using the SO character subset, once using the LC subset, and three times using the PO subset. Among the more clearly ape-like taxa, only Ardipithecus and Au. anamensis cluster apart from H. sapiens using every clustering method and character subset except the HN characters. Every other ape-like taxon occurs rarely in a cluster with Homo sapiens, with the surprising exception of the Paranthropus species. Par. robustus clusters with Homo sapiens in twenty of the fifty clusterings (not counting the HN character subset). Par. boisei and Par. aethiopicus cluster with Homo sapiens seventeen and eight times respectively, but never apart from Par. robustus. Of the fifty non-HN taxon partitions, Homo sapiens clusters with a nonhuman ape (i.e., Au afarensis, Au. africanus, Au. garhi, Paranthropus, Kenyanthropus, or Ardipithecus) 26 times (52%). We must remember, of course, that these sixty clusterings are not of equal quality, but the frequency with which nonhuman apes occur in the same cluster as Homo sapiens gives us reason to be cautious and skeptical of these character sets. DISCUSSION Essentialism may be scientifically or philosophically unverifiable, but it still remains a very likely true description of God’s created kinds, especially the human kind. Here, I’ve explored a much simpler and more tractable question: Can we adopt a naïve, character-based essentialist approach to distinguishing humans from nonhumans by using very simple lists of character traits? The lure of character-based essentialism would seem to be simplicity: a quick and easy test for placing God’s creatures into well-defined categories. Here, character sets derived from essentialist assumptions generally fail to distinguish human from nonhuman. The fifteen bipedalism characters listed here certainly do not distinguish human from nonhuman. The character subsets derived from the larger sample of 391 craniodental characteristics likewise fail uniformly to delimit a sensible human cluster. SO characters, defined as character states in living humans that differ from those in chimpanzee and gorilla, would seem to be the safest set of characteristics that could be reliably said to define human beings, but they exclude Neandertals, which creationists widely accept as being human. Expanding the set of “humans” reduces the number of character states they share that differ from those of extant apes. The Lubenow Core (H. sapiens, Neandertals, and African and Asian H. erectus) share only 33 characteristics not found in living apes, compared to 104 found in living humans only but not in living apes. Every one of the LC character states are shared by at least one fossil hominin that creationists agree are not human. Similar results were seen when trying to define character states essential to living apes (EAO characters) WOOD Essentialism and Human Kind 2023 ICC 98
RkJQdWJsaXNoZXIy MTM4ODY=