Occam's Razor

Contents Updated: Thursday, August 05, 1999

Parsimony or Occam's Razor

Parsimony, parsimonious and Occam's Razor often appear in books on dinosaurs. What do they mean? Jeff Poling has offered a useful explanation of the principle of parsimony with some interesting illustrations he has devised, on which this item is based.

Parsimony is a principle that states:

when more than one explanation can be devised for certain observations always prefer that which gives the most complete explanation while requiring the minimum of assumptions, revisions or refinements of current knowledge.

This priciple does not dictate. It tells the student to prefer. There is no law that states that natural systems must behave parsimoniously. Nature is often not simple but, by sticking to parsimony, hypotheses are tested and rejected in a regular way beginning with the simplest then moving on to increasingly complicated ideas until the simplest one is found that will hold up.

Any cladogram is really a diagram of evolutionary steps, such as the loss of a feature, or the gain of another. When two cladograms for the same organisms are compared, the one with the least number of evolutionary steps should be preferred by parsimony. But the preferred cladogram can depend upon which features are considered and lead to contradictory conclusions. This is often at the heart of disagreement between phylogenetic taxonomists.

Skeletal features such as vertebrae and hips can easily be seen in the physical fossil record of an extinct organism. Soft features such as organs and skin can rarely be seen. When physical features can not be seen in the fossil record, scientists have to hypothesize about them. Parsimony suggests which of the alternatives is to be preferred.

Larry Witmer has shown that the anatomical features of extinct animals should lie between living (extant) phylogenetic groups. An extant ingroup (a group of organisms within the same clade as those organisms being considered) is compared with the closest living outgroup (a group of organisms closest to those organisms being considered but not in the same clade). Any hypotheses must be compatible with what is known about the two groups and the preferred one is indicated by the principle of parsimony.

Examples

In these examples, the extinct animal, non-avian dinosaurs, are compared with their closest living ingroup, considered to be modern birds (neornithine dinosaurs) and their closest living outgroup, the crocodilians. The criterion of parsimony adopted is the number of evolutionary steps each hypothesis requires.

Hypothesis 1: Non-avian dinosaurs had hearts

Non-avian dinosaur hearts are not preserved in the fossil record, but the soft tissue anatomy of both birds, the ingroup, and crocodiles, the outgroup, have hearts. And hearts in more distant outgroups such as lizards, mammals and fish strongly suggest that this is a primitive condition inherited from distant ancestors. Accepting the hypothesis that neornithines evolved from non-avian dinosaurs, dinosaurs without hearts would require two evolutionary steps: the loss of the heart in the first dinosaur, then a reversal to having a heart in the neornithines. One evolutionary step is required if one assumes that non-avian dinosaurs retained the primitive condition of having a heart. The hypothesis that non-avian dinosaurs had hearts is the most parsimonious.

Hypothesis 2: Non-avian dinosaurs built nests out of vegetation, and sang to their young still in their eggs

Most neornithines and most crocodilians build nests out of vegetation, whereas other amniotes primarily nest in sand or not at all. Among those that have been studied, most neornithines and crocodilians sing to their young still in the egg, at least near hatching. Other amniotes do not. As with soft tissue, behaviors are rarely fossilized, and one must formulate hypotheses based on the evidence. Again accepting the hypothesis that neornithines originated within the non-avian dinosaurs, dinosaurs not having made nests of vegetation and sung to their young would have required two evolutionary steps: the independent evolution of these behaviors within the two extant groups. One evolutionary step is required if one assumes that non-avian dinosaurs also shared these behaviors. The latter hypothesis is the most parsimonious. Further, parsimony suggests that the behaviors are a derived condition for Archosauria (the common ancestor of dinosaurs, pterosaurs and crocodilians, and all its descendants), so all extinct archosaurs, not just the non-avian dinosaurs, likely exhibited these behaviors, unless they were lost or modified later (the fossilized remains of what looks like vegetation nests further strengthens the hypothesis for nest building behavior above and beyond what is suggested by the Bracket Method).

Hypothesis 3: Predentatans had cheeks

Neither neornithines nor crocodilians have cheeks. Indeed, among extant animals, only mammals have true cheeks. Since both groups are cheekless, parsimony suggests that the primitive condition for the non-avian dinosaurs is cheeklessness (no evolutionary steps, never evolving cheeks, versus one, evolving cheeks). However, anatomical features do suggest that advanced predentatans (genasaurians) evolved a cheek condition independently of mammals, so this is an example where stepping outside the Bracket Method shows that parsimony does not hold in this case (cheeks evolved at least twice within the Amniota rather than only once).

Hypothesis 4: All dinosaurs were feathered

Neornithines are feathered while crocodilians are not, nor are any other vertebrates. This suggests the evolution of feathers after the crocodiles split off from the rest of the archosaurs, or their secondary loss in crocodilians. Parsimony suggests feathers arose after the split (one step, evolving feathers once, versus two steps, evolving then losing feathers), but when after the split? To reach a parsimonious hypothesis based solely on the Bracket Method would require a living outgroup for the neornithine clade within the Dinosauria. Since no such outgroup exists, parsimony based on the Bracket Method is neutral: feathers either evolved within the Neornithines, or within the Dinosauria (one evolutionary step versus one evolutionary step). As with Hypothesis 3, we must turn to other evidence for more information (if such evidence exists):

  1. Functional arguments for the flight mechanism in extinct volant birds suggest that feathers must be present to allow flight. This takes feathers down through clade Ornithothoraces to the base of clade Aves.
  2. The presence of feathers in Archaeopteryx at the base of Aves suggests that feathers are at least primitive for Aves (one step, feathers evolving once, versus two steps, feathers evolving independently, or three steps, feathers evolving, being lost, then re-evolving for flight). Witmer calls this aspect the Lagerstatten modifier to the Extant Phylogenetic Bracket Method: well-preserved fossils (Lagerstatten) can make some fossil material effectively "extant" in terms of the evaluation of the soft tissue structure.
  3. The probable feathers of Compsognathus primus (=Sinosauropteryx prima) pulls feathers down to the base of Coelurosauria, assuming that C. primus was a basal coelurosaur, C. primus's integument really was feathers, and that the most parsimonious explanation for feathers is a single origin (one step, feathers evolving once in the Dinosauria, versus two steps, feathers evolving twice within the Dinosauria).
  4. Carnotaurus, as far as can be seen, lacked feathers (at least as an adult). This is the only non-coelurosaurian theropod for which much of the integument is known. It suggests by parsimony that, unless its scales were derived feathers, feathers (or at least feathered adults) developed somewhere between the base of clade including Ceratosaurus up to and including the base of clade Coelurosauria (one step, evolution of feathers in a later Dinosaur, versus two steps, evolution then loss of feathers). The presence of feathers in such dinosaurs as Torvosaurus and Allosaurus, for example, is therefore problematic in cladograms where allosaurids are more primitive members of clade Theropoda than Carnotaurus.
  5. The absence of feathers in preserved integument in any sauropodomorph or predentatan is consistent with the hypothesis that feathers evolved within the Theropoda and is not primitive for the Dinosauria as a whole. However, the confirmation of feathers in a small hyspilophodont, for example, in a lithographic limestone would question this part of the argument, as well as that used above for non-coelurosaurs. Such a find would suggest that the most parsimonious hypothesis is that feathers are common to all dinosaurs (one step, feathers evolving once, versus the multiple evolution of feathers in different branches of the Dinosauria).

Adding the known fossil record to the Bracket Method results in feathered coelurosaurs being the most parsimonious hypothesis (one step, feathers evolving once in the Coelurosauria, versus two steps, feathers evolving independently in at least two separate branches of the Coelurosauria, Aves and whatever branch contains C. primus).

Hypothesis 5: All dinosaurs were warm-blooded

For similar reasons, the debate about dinosaur warm-bloodedness (endo/exothermy, homeo/poikilothermy and tachy/bradymetabolism cannot be solved by the Bracket Method alone. Like with feathers, parsimony and the Bracket Method suggests that warm-bloodedness evolved somewhere within the Dinosauria but does not tell us where. Direct fossil evidence of warm-bloodedness is extremely hard to come by, and interpretations of the evidence that does exist are highly controversial. If one assumes that bone histology and other evidence is truly indicative that some predentatans were warm-blooded, and the fact that the ingroup is warm-blooded and the outgroup is not, then the most parsimonious hypothesis is that all dinosaurs were warm-blooded (one evolutionary step, warm-bloodedness evolving once, versus two steps, warm-bloodedness independently evolving twice). The case is further complicated by the fact that the first members of the outgroup, crocodilians, were swift, wolf-like predators with an erect stance,7 suggesting that today's semi-sprawling, cold-blooded lifestyle is a secondary adaptation to their ecological niche. Further, pterosaurs, as active fliers, likely were warm blooded as well. Parsimony would therefore make all archosaurs warm-blooded, with warm-bloodedness evolving somewhere between the first diapsids and the first archosaurs. Warm-bloodedness will probably remain a contentious subject moreso than the issue of feathers, for which there is direct evidence in some non-avian dinosaurs.

As in cladistics, multiple workers can disagree on what constitutes the most parsimonious hypothesis.

Conclusion

Parsimony is a principle in science where the simplest answer is always the preferred. In cladistics, the preferred hypothesized phylogeny, or cladogram, is the one that requires the fewest evolutionary steps and does not violate the tenets of logic and science. Similarly, parsimony can be used to test hypotheses, formed using the "Extant Phylogenetic Bracket Method" with consideration of other evidence, on what anatomical features an extinct organism might have had. Although there is no rule that requires nature to follow the simplest path, and results can vary based which pieces of evidence are used, it is nonetheless a strong basis for the scientific work of paleontologists.

These notes are partly abstracted from the website, Dinosauria On-line, to which readers wanting more detail are referred. It has a wealth of dinosaur information for dinosaur amateurs and enthusiasts alike, aims to give the reader a broad exposure to dinosaur science and provides a forum for topical discussion. Discover the links between Archaeopteryx and modern birds and find out why the dinosaur DNA find claim may be a mistake. There is a link to the Dinosaur Electronic Mailing List and real enthusiasts can order a replica oviraptor egg for their mantelpiece. Send e-mail to Jeff Poling.

The Dinosaur Mailing List is an e-mail based newsgroup, owned by Mickey Rowe, for scientific discussions about dinosaurs. To subscribe to the mailing list send e-mail to listproc@usc.edu with no subject and SUBSCRIBE DINOSAUR your-real-name-here in the body of the message.

To unsubscribe from the mailing list send e-mail to listproc@usc.edu with no subject and UNSUBSCRIBE DINOSAUR in the body of the message.