What I say now ought to be uncontroversial, but bears repeating: philosophy has a public relations problem. Specious criticism from unreflective popular figures has done its damage. Inquisitive laypeople are routinely exposed to philosophy in one of two contexts: as an activity that works at best as a pointless diversion, or as an activity directed towards defending itself from charges of pointless diversion. I have discussed this problem before, but the situation remains largely unchanged . I will therefore court controversy with a potentially helpful suggestion: we should start proclaiming, loudly and repeatedly, that one the most significant works of philosophy this year was “A specimen-level phylogenetic analysis and taxonomic revision of Diplodocidae (Dinosauria, Sauropoda),” and we should remind people that this is the work that brought Brontosaurus back .
I mentioned this paper in an offhanded comment made during a panel discussion earlier this year and members of the audience interrupted with applause. Politicians may be able to buy publicity like that, but we can’t. I believe that philosophers can harness the enthusiasm for this work in a way that not only improves the field’s public image, but also gives non-philosophers a better understanding of how philosophy works and why philosophy is useful.
In the discussion below, I will make clear what I mean by “philosophy,” lay out the history of debate over the name “Brontosaurus,” explain how Tschopp et al 2015 resolved that debate, and demonstrate how their resolution fits my model of philosophy.
What I do not intend is to denigrate any of the more clearly philosophical work done by my colleagues, nor do I intend to misrepresent the rigorous quantitative research compiled by Tschopp et al. We can all get along. My intention is to demonstrate why we should.
What is philosophy?
We have a running joke in philosophy: that when you ask n philosophers to define what they do, you will receive n+1 answers. Humor is clearly not our best tool for improving public relations.
It is telling, I think, that the words that I speak most often while teaching are “by that logic.” I doubt that I’m alone in this: other philosophers certainly use that phrase very often in conversation. I say these words so often that I now include a full lecture at the beginning of each semester that explains what those words mean. “By that logic…” statements are common tools in philosophical practice. These tools are a philosopher’s shorthand for conceptual analysis.
Concepts, being abstract entities, do not suffer the same sort of poking and prodding that scientists use for testing observations. If we mean to test a concept, then we must probe the concept’s limits through thought experiment. Philosophers do this by considering apparent contradictions among applications of the concept’s definition. If the apparent contradictions can be resolved, then the concept passes the test; if they cannot, then the concept ought to be rejected as false.
For example, here is a conceptual analysis of one common definition of the term “dinosaur”:
Dinosaurs may be defined as gigantic reptiles that have been dead for a very long time. By that logic, a cloned T. rex like the one in “Jurassic Park” wouldn’t be a dinosaur. But a cloned T. rex like the one in “Jurassic Park” should be considered a dinosaur; therefore, the suggested definition of the term “dinosaur” must be false.
Whether or not you disagree with the details of the example is irrelevant; what matters is how the analysis works. Definitions are universal principles, meant to apply to all possible instances of the term. If the definition cannot apply to one possible instance of the term, then there is an apparent contradiction that must be resolved. The resolution happens in one of two ways: either the suggested instance is reconsidered in such a way that the definition would in fact apply or the definition is discarded.
I’m not suggesting that philosophy is just semantics. Yes: we are trying to figure out what words mean, but that only tells half the story. I am reminded of a quote from Epictetus: “How then did Socrates act? … If the adversary had defined envy, he did not say, ‘You have defined it badly, for the terms of the definition do not correspond to the thing defined’” . Socrates wanted to show how definitions must be constrained by reality. It is not just that we’ve found inconsistency in the application of the words “dinosaur” or “envy”; the reason that we’ve found those inconsistencies is that dinosaurs and envy are real things and our understanding of those things needs improvement. Conceptual analysis is intended to improve our understanding of these real things.
This goal is reflected in how we resolve apparent contradictions in conceptual analysis. If one considers the definition to be truly universal, then the possible instance will be reconsidered; if one considers the possible instance to be a true misapplication, then the definition will be rejected. Using traditional logic, the goal in either case is to preserve truth. Unlike empirical truths discovered through scientific observation, however, the truth preserved by conceptual analysis is reasoned truth, discovered through philosophical argument.
Conceptual analysis may not be necessary for all philosophy, but I do think that it is sufficient for philosophy. This kind of analysis is not science because it is not empirical; it is not linguistics because it is about more than just words; it is not semantics because it examines more than just meaning. Here is where philosophy steps in: to examine meaning as it is related to reality itself.
What is Apatosaurus?
For over a century now people have believed that Brontosaurus was a real thing, only to be told by dinosaur enthusiasts that it was not. No matter how many times dinosaur enthusiasts pointed this out, however, the real name — “Apatosaurus” — seemed to be outside the lay public’s memory capacity. This is why Tschopp et al 2015 was so welcomed by the public despite being a 298-page discussion of the minutiae of caudals and distals and such: because the purported “real thing” had such a comparatively disappointing name.
The history of the layperson’s disappointment gives important context to the new paper. It is also important because it reinforces a point made above: that we are debating whether or not something is real, not simply whether or not we should prefer one name to another.
Brontosaurus’ story began in 1877 when a team led by Othniel Charles Marsh excavated a partial sauropod skeleton in Colorado. The individual specimen, now named YPM 1860, became the type specimen (i.e., reference point) for a new genus and species that Marsh named Apatosaurus ajax. Two years later another of Marsh’s teams unearthed parts of a larger sauropod in Wyoming. Citing the differences in size, as well as some differences in vertebral anatomy, Marsh used the second specimen — now named YPM 1980 — as the type specimen for another new genus and species, Brontosaurus excelsus. Shortly thereafter everyone became very, very confused.
In 1903 paleontologist Elmer Riggs wrote a review of then-known sauropods, similar to the work that would be done a century later by Tschopp et al. In that review Riggs argued that the differences between YPM 1860 and YPM 1980 could be explained by growth and development: in effect, that YPM 1860 should be regarded as a teenager and YPM 1980 as an adult within the same genus. The name “Apatosaurus” was coined earlier and so Riggs renamed the Brontosaurus species “Apatosaurus excelsus.” Riggs’ paper was published while the American Museum of Natural History was planning to mount a restoration of the larger specimen; since the museum’s curator preferred the name “Brontosaurus,” the skeleton was so labeled when it was unveiled in 1905. Thus began 110 years of separation between received scientific wisdom and popular perception .
Marsh is often criticized for sloppy scholarship in his zeal to name more dinosaur species than his hated competitor, Edward Drinker Cope. Much of that criticism is justifiable. Nevertheless, Marsh wasn’t sloppy or overzealous when he distinguished Brontosaurus from Apatosaurus.
Darwin once likened the fossil record to a “history of the world imperfectly kept, and written in a changing dialect … only here and there a short chapter has been preserved; and of each page, only here and there a few short lines” . The classification of dinosaurs depends on inferring information about whole species and genera from isolated, scattered, and incomplete skeletons. Sauropods in particular have been problematic. It’s rare that paleontologists find a specimen that includes more than half the animal’s skeleton. It’s even more rare that the skeletons include the part most useful for classification: the skull. Still more rare are multiple specimens found in a single location. These are severe handicaps in determining classifications.
Of over 100 described genera in the Sauropoda, many (if not most) are known only from an isolated humerus, femur, or collection of vertebrae. Sauropod classification therefore tends to lean heavily on five kinds of trait: vertebral anatomy; limb anatomy; presumed adult size; geographic location; and geological period. Many living species, much less genera, vary widely in these particular traits. Riggs’ point was that paleontologists should try to use the same naming standards adopted by other taxonomists. For his part, Marsh was just using the information he had available.
Paleontologists didn’t have much more to work with even after another hundred years of research. Consider this diagnosis of the Diplodocidae, the family that includes Apatosaurus, from McIntosh’s 1990 study of sauropods:
“Long snouted skull with superior nares and weak peg-like teeth confined to the front of the jaws, cervicals and anterior dorsals opisthocoelous, posterior dorsals amphicoelous, short cervical ribs, deeply divided V-shaped neural spines in and on both sides of the shoulder region, very long tail with elaborate wing-like transverse processes on anterior caudals, caudal centra gentle proceolous with weak chevron pacets, elongated middle and distal caudals, fore and aft expanded forked chevrons in the mid-tail region, tail ending in a whip-lash, short metacarpals, and metacarpal II or III:humerus = .32 to .37.” 
While there is mention of some general features of skull anatomy, most of the diagnostic traits of the taxon are subtle, technical features of vertebral anatomy. Just as you should make lemonade when life only gives you lemons, you should diagnose a fossil taxon by backbones when the ground only gives you backbones.
But just as a master chef can make astounding and surprising concoctions from lemons, so too can paleontologists draw astounding and surprising information from relatively small collections of fossils. The increasing sophistication of statistical analysis allows modern researchers to test hypotheses upon which Marsh and Riggs could only speculate.
What is Brontosaurus?
The stated purpose of Tschopp et al 2015 was not to comment on the validity or invalidity of “Brontosaurus” as a genus name. Instead, the authors set out to clarify the evolutionary relationships within the family of dinosaurs that includes Apatosaurus. Doing so required specification of conditions for species-level and genus-level distinctions, and so the commentary on “Brontosaurus” became almost inevitable.
What made the authors’ commentary different from Riggs’ or McIntosh’s is that Tschopp et al employed one of the sophisticated modern statistical methods of classification. It is called numerical taxonomy.
Numerical taxonomy is a method of drawing distinctions between biological taxa, first proposed by Sokal and Sneath in 1963 , which intends to make classification more objective. The method suggests that biologists should tally all of an organism’s traits and compare those traits against variations found in other organisms; through statistical analysis taxonomists could then designate a taxon according to which organisms are more similar to each other than they are to members of other taxa .
In the 2015 paper Tschopp et al compared variations of 477 traits found among 81 well-described sauropod specimens (2015, 162-164). Many of those traits are relevant to vertebral anatomy; many are not. Previous attempts at classification put undue influence on the body parts of sauropods that have preserved best in the fossil record, but by tallying so many other traits the authors could potentially mitigate the effects of preservation bias.
Buried about 2/3 of the way through the authors’ 298-page analysis is the passage that has so captured the public’s interest:
“… the clade comprising A. ajax YPM 1860 + A. louisae type is separated from its sister clade Brontosaurus excelsus YPM 1980 + mdA by eleven changes … the analysis with implied weights suggests the presence of two different genera, whereas only specific separation is supported with equal weighting. As mentioned above, also mean pairwise dissimilarity between specimens of Apatosaurus and those of Brontosaurus (0.2606) supports generic distinction: intrageneric mean pairwise dissimilarity is lower (0.1831 for Apatosaurus, and 0.2149 for Brontosaurus) than what is found between the two groups.” (2015, 195-196)
Obviously, when I say “the passage that has so captured the public’s interest,” what I mean is “the passage that has been reduced to a three-word social-media-friendly sound byte and lost most of its context in the process.” Allow me to restore some of that context.
When the authors compared uncontroversial members of the genus Apatosaurus — YPM 1860, the reference specimen for A. ajax, and specimens in the species Apatosaurus louisae — with other specimens in taxa that evolved later, they found consistent differences in at least eleven traits. Narrowing the comparison down, the authors found that specimen YPM 1980 — the one originally named “Brontosaurus” by Marsh — differed from members of the genus Apatosaurus in about 26% of its traits (the particular differences may have varied in specimen-by-specimen comparisons). By contrast, members within the genus Apatosaurus tended to differ from each other in only 18% of their traits.
To simplify further: the specific dinosaur skeleton originally named Brontosaurus” (YPM 1980), later reclassified as “Apatosaurus,” differs from other specimens originally named “Apatosaurus” in just over one quarter of its measured traits. Uncontroversial specimens in the genus named “Apatosaurus” differ from each other by less than one quarter of their measured traits.
From these particular facts the authors drew the conclusion that “Brontosaurus and Apatosaurus should therefore be considered valid genera” (2015, 196). Cue fanfare.
What is dinosaur philosophy?
Tschopp et al did not observe a distinction between the genus “Apatosaurus” and the genus “Brontosaurus.” Dinosaur fossils do not come out of the ground with nametags. Their observation, if it can indeed be called an “observation,” was made indirectly, through application of a generalized principle.
Indirect observation is a combination of direct observation with philosophical argument. Consider another example of indirect observation in paleontology: evidence for feathers in the species Citipati osmolskae. Several specimens of the species have been found associated with a nest of eggs in a posture that birds assume while brooding . We therefore have direct evidence that C. osmolskae assumed a brooding posture while sitting on its nest. This itself is not evidence for feathers, but as a general principle ornithologists and paleontologists believe that brooding is an effective means of nest incubation only when the brooder is feathered. As an application of that principle, then, we conclude that C. osmolskae must have had feathers. Neither the empirical observations nor the conceptual argument alone would be sufficient to show that the species bore feathers. When the two are combined, however, that truth can be reasoned through.
If I am correct, then Tschopp et al needed a general principle from which they could conclude that Brontosaurus is a valid genus. That principle is given shortly before their analysis of Apatosaurus and Brontosaurus:
“Within Diplodocinae, specimens considered to belong to the same genus exhibit values below 0.181, whereas different genera show values of 0.222 and higher. Two generally accepted species within a single genus (Diplodocus carnegii and Diplodocus hallorum) have a value of 0.1195.” (2015, 171)
In other words: two groups in the family Diplodocinae (a subgroup of the titular Diplodocidae) should be considered distinct genera if members of one taxon tend to differ from members of the other taxon in at least 22% of their traits. This principle was generalized from analysis of comparisons between members of species that are “generally accepted” to belong within the same genus and members of species that are generally accepted to belong in different genera.
The principle also shows why the authors’ conclusion about Brontosaurus cannot qualify as a purely empirical observation. If the conclusion were a direct observation, then when I write, “YPM 1980 differs from Apatosaurus specimens in 26% of its traits on average,” the most natural inference on the part of a reasonable reader would be, “I guess that Brontosaurus is a valid genus, then.” That inference would be invalid because “22% pairwise difference” doesn’t always mean “different enough to qualify as distinct genera.”
Tschopp et al admit that “the present analysis was designed for the study of diplodocid intrarelationships, and is not suitable for inferring the phylogeny of clades outside Diplodocoidea” (2015, 180). The principle “differences between at least 22% of two specimens’ traits is enough of a difference for distinct genera in diplodocids” is one they derived from painstaking observation. If the reality of the genus Brontosaurus was observed at all, then it was observed indirectly through application of that principle.
Here, then, is the argument for restoring “Brontosaurus” as a valid genus name:
- The sauropod specimen YPM 1980, currently classified in the genus named “Apatosaurus,” tends to differ from other members of the genus Apatosaurus in about 26% of its traits.
- Among members of the dinosaur group Diplodocinae, which includes the genus Apatosaurus, specimens that differ from each other in at least 22% of their traits are different enough to be classified in different genera.
- Therefore, YPM should be classified in a genus distinct from Apatosaurus; since YPM 1980 was originally classified in the genus named “Brontosaurus,” the distinct genus should have that name.
What we have here is a general principle being tested against a controversial particular instance. Tschopp et al established what it means for taxa to be different enough to qualify as distinct genera in the family Diplodocinae. The particular genus Apatosaurus seems to violate that principle: one of its purported members, YPM 1980, should be different enough to be in a distinct genus. Since the classification of YPM 1980 was originally controversial, and since their principle has been demonstrated empirically, the authors reasoned that the particular instance should be reconsidered to fit the principle. What we have here is conceptual analysis of what it means for Brontosaurus to be “different enough” from Apatosaurus.
This is why I consider Tschopp et al 2015 — or at least the part that argues for the reality of a genus named “Brontosaurus” — to be such a vital work of philosophy. The authors have employed an important philosophical method to advance an ongoing discussion that has significant public appeal. Their argument about Brontosaurus simultaneously puts the lie to two common misconceptions about philosophy: that philosophy fails to make progress and that the progress philosophy fails to make is in irrelevant or esoteric topics.
I do not mean to say that Tschopp et al 2015 is not a work of paleontological science, nor do I intend to claim any credit for their work. The authors have painstakingly accumulated a great deal of quantitative data towards the end of answering empirical questions about extinct organisms. But the passages quoted above put the lie to a third common misconception about philosophy: that philosophy is wholly distinct from other disciplines. Paleontology and philosophy are not mutually exclusive pursuits; they are complementary.
It’s easier to see this relation once we accept conceptual analysis as sufficient for philosophy. Considered as the testing of definitions against the constraints of reality, conceptual analysis should be part and parcel of the sciences. While Tschopp et al may use scientific methods for deriving a definition of “different enough to be distinct diplodocid genera,” they must wear a second hat — the hat of a philosopher, in addition to the pith helmet of the paleontologist — in order to test that definition. It is possible and in many cases necessary to act as a philosopher while working as a scientist.
Why, then, do we need philosophers at all? If a scientist must also be a philosopher, and if one of our best works of philosophy can’t even really be categorized as such, then haven’t I already conceded too much to the critics I mentioned earlier? No. I am reclaiming conceptual ground that philosophers never should have ceded in the first place. Scientists are trained in empirical and mathematical methods; that is their expertise. What I have shown is that they sometimes borrow from the expertise of philosophers — from our work in logic, semantics, and metaphysics — as is to be expected in an academic community. Whether they do it consciously or not is irrelevant. My point is that it happens. We would all benefit from conscious acknowledgement of that fact: the philosophers’ image would be improved and the scientists’ work could be given added depth.
Sometimes the depth is already there. Without philosophy, we could gather exactly this much from the research done by Tschopp et al: that the fossil taxon including YPM 1980 is quantifiably different from the fossil taxon including YPM 1860. To demonstrate that the taxa are different enough to qualify as distinct genera, however, the authors had to engage in philosophical reasoning. That philosophical reasoning alone wouldn’t be sufficient to show that YPM 1980 should be classified as “Brontosaurus,” but neither was the empirical work. Tschopp et al have done all this with aplomb and deserve credit as such; my goal is only to call things what they are. I am very much a philosopher in that sense.
Leonard Finkelman is an Assistant Professor of Philosophy at Linfield College (OR). He received his PhD in Philosophy from the City University of New York Graduate Center in 2013, under the supervision of Massimo Pigliucci. His research focuses on issues in the philosophy of biology, particularly those related to paleontology and classification. In addition to this research, Leonard has written on topics in ethics, possible-world semantics, and human nature. He will also occasionally indulge interests in astronomy, prehistoric art, science fiction, and graphic novels.
 The Value of Public Philosophy to Philosophers, by M. Pigliucci and L. Finkelman, Essays in Philosophy, 2014.
 A specimen-level phylogenetic analysis and taxonomic revision of Diplodocidae (Dinosauria, Sauropoda), by E. Tschopp et al., PeerJ 3:e857, 2015.
 Discourses, Book II, by Epictetus. Thanks to Massimo Pigliucci for bringing this passage to my attention through his daily social media postings of Stoic meditations.
 Among those confusions: the idea that Brontosaurus was discarded as a valid genus name because the American Museum of Natural History restored YPM 1980 with an incorrect skull. It is true that the original mount included the skull of Camarasaurus supremus, a sauropod that had been described by Edward Drinker Cope in 1877. This was a practical move: the museum did not want to display a headless skeleton. Nevertheless, this mistake had no bearing on the controversy over the name “Brontosaurus.” One story holds that Cope engineered the mistake by hiring a mercenary to plant a Camarasaurus skull close to the quarry from which YPM 1980 came, but this is neither substantiated nor relevant.
 On the Origin of Species, by C. Darwin, pp. 310-311.
 Dinosaur Systematics: Approaches and Perspectives, ed. by K. Carpenter and P.J. Currie, Cambridge University Press, 1992.
 Principles of Numerical Taxonomy, by R.R. Sokal and P.H.A. Sneath, W.H. Freeman & co., 1963.
 Detractors highlight a number of limitations inherent in the numerical method. In particular: the designation of traits and variations thereof must often come down to the researcher’s choice. That, too, is a place where philosophy can make significant contributions to the life sciences. But that is an essay for another day.
 A nesting dinosaur, by M.A. Norell et al., Nature 378:774-776, 1995.