Griffiths Chapter 8



In this Chapter, Griffiths expands on the ideas he developed in the previous chapter. He wants to apply the "natural kinds" idea to psychology. Before doing that, however, he feels it necessary to explore how the idea applies to biology.

You might expect that species will be natural kinds in the same way that chemical elements are. The causal homeostasis of chemical elements is their microstructural properties: those properties are what explain all the chemical elements other properties best. That holds true for any time, it seems (one might ask about before the big bang, or in the first few miliseconds, but let's not). Thus those properties are not historical.

Biological kinds, however, are historical entities. They are individuals in a long chain of individuals that are linked as parent and offspring. Each species is the result of a bifurcation: one species splits into two new species. Each species thus occupies a span of history and exists relative to its predecessors. Who its predecessors are goes a long way toward explaining what it is. This view is called "the historical view" of biological species.

A non-historical view of biological species is "intrinsic essentialism." Essentialism tries to identify a species by whether it has certain essential traits.

In its simplest form, intrinsic essentialism claims that in order to be an X, an individual must have certain essential traits. But for any given species, you can find individuals that seem as if they ought to be considered members of that species that do not have some essential trait: two-headed calfs, unstriped tigers, etc.

A more sophisticated form of essentialism: statistical essentialism. Statistical essentialism claims that an individual is a member of a species if it has "a sufficient number of appropriately weighted properties from the defining set" (of properties).

Griffiths objects to statistical essentialism:

  1. To be practical, an essentialist definition has to be limited, vague, cut down to a minimum. In that case, there will always be outliers that do not fit the definition but nonetheless seem as if they belong in the species.
  2. If the definition is complete, it might be so complicated that it is impossible for humans to apply.
  3. Species change over time, and so even a successful statistical definition would only apply at a particular time. As the species changes, you would get many definitions. Then you would have to ask what makes them all one species. What might get lost in all that is something biologists think is important: that there are specific changes to a species that are speciation events. Speciation events are the point where two new species replace a former single species.


Statistical essentialism is not projectable over time, and so it fails to identify natural kinds over time. If Darwinian evolution is right, there are natural kinds over time. Thus statistical essentialism is incompatible with Darwinian evolution.

The alternative is to think that species are not like chemical elements that are what they are because they have some essential definition that is their homeostasis. Rather, they are what they are in the same way that the Smith family is what it is, or New Zealand is what it is. Those are historically located individuals. Membership in a species is a matter of being descended from a common ancestor, not resembling other members of the species in intrinsic properties.

Griffiths calls the homeostasis of species "historical essences."

Sure, two species might have very great similarities, but those are homoplastic, not homologous. I.e. they have the same form but different explanations.

If you try to classify species by similarity (aka phenetically), you wind up with different classifications depending on what similarity you count as most important. That is an unstable situation. Phenetic classifications are similar to statistical essentialism. They both do not work, however.

Genealogical (aka historical) methods produce a stable classification.

There is a rival school of thought that has some validity. It is the ecological school. The ecological school looks at functional similarities of species and classifies them according to what functions they fulfill, regardless of their historical descent. Griffiths thinks that that sort of classification is useful, but only as a supplement to his historical classifications.

8.2 Cladistics
Cladistics is the technical term for the system shown in those branching diagrams that show descent of species, called cladograms. Each branching has only two branches. Each branch is a species. When there is a fork, one species ceases to exist and is replaced by two others. There is only one correct cladistic system (just as there is only one correct family tree for you). That means, however, that a species can evolve and change over time without becoming a new species. Without a speciation branching, there is no new species. Paleontologists nonetheless speak of different appearing animals as different species, even if it turns out to be the case that they are the same species which has just changed so drastically over time. Think of dogs and how they have changed so drastically from their ancestor dog.

This system of cladistics has little place for the traditional taxonomic rankings of kingdom, phylum, class, order, family, genus. Each line in a cladogram is a species. Those traditional rankings are not to be thrown out, but they are not basic any more. Aristotle has been rejected firmly here! Those traditional rankings are sort of like if you were to call yourself a member of the X family, which is itself a member of the Y clan, which is itself a member of the Z state, which is a member of the Q nation, etc. There is nothing that ties together all members of those higher traditional rankings beyond common descent: they no longer act as a unit. A species, however, is tied together and acts as a unit. It evolves together as long as it is a species.

In precise terms, the cladistic species are always provisional hypotheses: they are hypotheses about the underlying reality. The classification is always subject to revision, which is (in the case of progress) somehow caused by the underlying reality of what species are.

Traditionally, "natural kinds" cannot be restricted in time and place, but species are. So they are not traditional natural kinds. Traditionally, natural kinds share the same microstructure, but one single species can evolve its microstructure over time, and yet it is still the same species. So they are not traditional natural kinds again.

Griffiths' definition of natural kinds as kinds whose members have the same causal homeostasis as others of their kind makes it possible to speak of biological species as natural kinds.

So Griffiths is trying to change the concept of "natural kind."

"The traditional conception of a natural kind supposed that natural kinds would enter into exceptionless empirical generalizations of "natural laws." This conception reflects a philosophy of science in which the sciences that do not yield exceptionless laws are seen as inadequate and awaiting replacement by some more exact science. The logical empiricist philosophy of science of the first half of this century assumed that physics gave a privileged description of reality and looked forward to the day in which science could be "unified" on the foundations provided by physics. The last half century has seen the abandonment of this dream. The "unity of science" has dwindled to a minimum notion of supervenience-the world studied by economics or population biology does not change independently of the world studied by molecular biology or by microphysics. In this new philosophy of science the exception-ridden generalizations of many life and social sciences are seen as the only way to uncover some of the regularity inherent in natural processes." (213)

"Natural kinds are no longer conceived as the subjects of the fundamental laws of nature. They are simply nonarbitrary ways of grouping natural phenomena. These kinds are nonarbitrary (or "natural") because they have some degree of projectability." (213)

8.4 Cladistics and biological traits

So species are natural kinds and their causal homeostasis is that they each have a particular historical placement in the cladogram.

Now what about traits of species? Should we say that natural kinds of traits are things like "flying" or "egg-bearing" or "furry"? They seem like good ways to classify traits. Are they good ones for epistemic purposes (i.e. for explanation, for science)? No, says Griffiths. Natural kinds of traits are homologues. "A homologue is a trait uniting a natural clade (a clade is all the species that are after any given bifurcation in a cladogram). Every species in the clade either has the trait or is descended from a species that had the trait." (213)

"Homology can be identified with resemblance due to descent from a common ancestor. Analogy is identified with resemblance due to parallel evolution." (214) But resemblance is not the key to either of them, really. They key is descent and parallel evolution to solve the same problem, respectively.

This way of classifying traits by clade is fairly common in biology, but psychologists are still not altogether convinced. Some psychologists want to classify traits by whether they are aimed at the same goal, whether they fulfill the same function. In other words, they are sticking to classification by analogy rather than homology.

If you classify by function (that is, by analogy), you can make cross-species generalizations, which makes the classification useful, and is why people resist jettisoning functional classifications.

A functional classification is useful, but it has less precise predictive and explanatory power than cladistics, however. For example, specifying the particular adaptive problem that organisms need to solve, that is the function it needs to fulfull, does not tell you much about how a particular organism will solve it. On the other hand, knowing that crocodiles and birds are both archosaurs, that is knowing their cladistic classification, will help you explain why they both lay eggs, have complex vocalisation, build nests, and care for a brood (things which non-crocodile reptiles don't do). Based on that, the claim is that you can predict other similarities: the cladistic classification gives you a heuristic device that can lead to rich investigative results.

8.5 Griffiths' "explanatory levels"

Evolution is "Genealogical actors in ecological roles." (Hull's phrase: 217).

So each species is an actor that plays a role in something like a play. When the environment changes, the species has to take on a new role, a new function.

Griffiths maintains that there are several levels at which biology wants to explain species.
There are:


There is the most abstract level. Griffiths says disappointingly little about this level, but it is clear that it is the level at which what one is trying to explain is not some particular species, but rather one is trying to explain larger cross-species patterns of adaptation via generalizations about adaptation strategies and some generalizations of population dynamics. At this level, what is being discussed could well apply to chimpanzees or dung beetles, and could apply to any function that biological things need to fill: that is, what is being explained is so abstract that it is not specific to any clade or species or even trait or function. It is rather about how any solution(s) to any problem in any species come to be selected for and the patterns of how it is distributed in any population. No historical facts are used to explain outcomes.

The next level is one of "general ecology." It uses functional classifications and is far from explanation of any particular species too. On the other hand, it is not the most abstract level, because it IS explaining via particular functions. An example of a general ecological explanation is one that explains signal detection. That is, it explains a particular function and the strategies that can fill it. No historical facts are used to explain outcomes.

The general historical descriptions have particular realizations in particular species. When these particular realizations of general historical descriptions are what is to be explained, Griffiths calls it the natural historical level. At this level, a given species' trait is explained. At this level, particular historical facts are relevant and used to explain outcomes.

The lowest level is that which explains anatomically and physiologically how things get done. So how the whale's fin works. Particular historical facts, particular species, and particular traits are considered here.

Now here comes something really interesting: those different levels use different natural kinds. The homeostatic mechanism of the things which the natural historian is trying to explain are those which explain species. The general ecologist, however, needs different natural kinds!

You can think of these different explanatory levels as nested disciplines. The Abstract scientist takes the things that the general ecologist has explained and creates a classification of those things and tries to explain that classification. The general ecologist takes the things that the natural historian has explained and creates a classification of those things and tries to explain that classification. The natural historian takes the things that the anatomist/physiologist has explained and creates a classification of those things and tries to explain that classification.

Each asks questions of the other and provides a valuable heuristic tool for the other.

That explains why the cladistic classification explained so much about the crocodile and the bird, but the functional did not. The higher levels of explanation explain less and less detail and are more and more general.

The table on page 221 is a comparison of the explanatory levels of biology with those of psychology.

Psychologists are of different stripes. Some want to explain the nitty gritty level of how a particular psychological function happens in mechanical ways. Some want to explain what the particular function does: what information does it use, how does it process it, etc. Still others want to explain something yet more general: what does the trait do for the organism?

8.7 The payoff of chapters 7 and 8
Why summarize this? It is a summary itself of the results of the work that Griffiths has laid out in chapters 7 and 8.

The second full paragraph on P. 225 is fascinating. So is the first full paragraph on P. 226 which considers the question of what "pain" "is".

Griffiths wants to ask now "how should we talk about emotions?" Which natural kind classification is the one which we should default to when we speak?