Samir Okasha's
Darwinian Metaphysics: Species and the Question of
Essentialism
Key terms
Intrinsic property: the necessary property/ies that make a thing what it is. Usually it is considered to be part of the things microstructure but not necessarily. E.g. An intrinsic property of a car is that it has 4 wheels and an engine. An intrinsic property of water is that its chemical structure is H20.
Essentialism (about kinds): The idea that things have essences that make them what they are. This is often used interchangeably with intrinsic property. For example, what makes the kind 'dog' is that all dogs have the essence of "doggyness". This could also be explained by properties like characteristic behaviours, DNA, or form. Aristotle classically declared that the essence of 'man' is rational thought.
Introduction
In virtue of what is my wiener dog a
member of Canis familiaris? Is there some intrinsic
essential property that he and all other members of the species
posses? It seems like there is so much variation between 'breeds' that it'd be difficult to find a group of properties that all dogs have.
Okasha takes essentialism to be the position of Kripke, Putnam, and Wiggins—i.e., that the concept of kind essentialism used in the hard sciences also applies to biological kinds. Okasha suggests Putnam and Kripke (PK) are mistaken to apply their brand of essentialism to biological kinds because (a) of facts about evolutionary biology that don't apply to the hard sciences, and (b) the purpose of classification in biology is different from that in the hard sciences. However, it is not all doom and gloom for the PK model: if instead of demarcating kinds according to intrinsic properties we use relational properties, aspects of the PK model can be salvaged.
Okasha takes essentialism to be the position of Kripke, Putnam, and Wiggins—i.e., that the concept of kind essentialism used in the hard sciences also applies to biological kinds. Okasha suggests Putnam and Kripke (PK) are mistaken to apply their brand of essentialism to biological kinds because (a) of facts about evolutionary biology that don't apply to the hard sciences, and (b) the purpose of classification in biology is different from that in the hard sciences. However, it is not all doom and gloom for the PK model: if instead of demarcating kinds according to intrinsic properties we use relational properties, aspects of the PK model can be salvaged.
Philosophical roots and overview
of the Putnam/Kripke (PK) Essentialism
The philosophical origins of the PK essentialist model can be traced back to Locke's distinction between the nominal and real essence of a kind. The nominal essences are the macro-properties of an object that we pick out to determine the group to into which we place it. The standard example is that the nominal essence of gold is that it is shiny, metallic, yellow, and malleable. The particular properties we settle on to define kinds are conventional; that is, they are not dependant on anything intrinsic to the particular things we are classifying, rather they are selected based on utility and/or accidental facts about our perceptual system. For example, we could have grouped objects according to gross size and texture, but this would not have served any useful purpose.
In contrast to nominal essence there is the real essence which is the (intrinsic) hidden underlying microstructure which is causally responsible for nominal essences. If we could access the real essence of objects, we'd be able to group them according to their intrinsic properties/”hidden structure” and therefore their metaphysically real (as opposed to conventional) kinds. Locke didn't anticipate our having “microscopic eyes” to actually identify real essences, so he supposed all kinds would be nominal.
Of course, science has progressed to the point in the hard sciences where microstructure can be identified, as so objects can be classified according to (real) kind. PK natural kinds emerge out of this reality. The standard examples of PK essentialist kinds are “gold is having the atomic number 79” and “water is having the chemical composition H20”. So, determining whether something is a kind is an empirical matter. For example, to determine whether H20 is a kind is a matter of verifying that all samples of water have the molecular structure H201.
The philosophical origins of the PK essentialist model can be traced back to Locke's distinction between the nominal and real essence of a kind. The nominal essences are the macro-properties of an object that we pick out to determine the group to into which we place it. The standard example is that the nominal essence of gold is that it is shiny, metallic, yellow, and malleable. The particular properties we settle on to define kinds are conventional; that is, they are not dependant on anything intrinsic to the particular things we are classifying, rather they are selected based on utility and/or accidental facts about our perceptual system. For example, we could have grouped objects according to gross size and texture, but this would not have served any useful purpose.
In contrast to nominal essence there is the real essence which is the (intrinsic) hidden underlying microstructure which is causally responsible for nominal essences. If we could access the real essence of objects, we'd be able to group them according to their intrinsic properties/”hidden structure” and therefore their metaphysically real (as opposed to conventional) kinds. Locke didn't anticipate our having “microscopic eyes” to actually identify real essences, so he supposed all kinds would be nominal.
Of course, science has progressed to the point in the hard sciences where microstructure can be identified, as so objects can be classified according to (real) kind. PK natural kinds emerge out of this reality. The standard examples of PK essentialist kinds are “gold is having the atomic number 79” and “water is having the chemical composition H20”. So, determining whether something is a kind is an empirical matter. For example, to determine whether H20 is a kind is a matter of verifying that all samples of water have the molecular structure H201.
Another important way the PK model
diverges from Locke is over the “semantic inertness” of
real essence. Whereas Locke thought that speakers when using kind
terms were only referring to nominal kinds, PK hold that even without
knowledge of real properties speakers imply that their kind-terms
refer to the real properties causally responsible for the observed
properties2.
There is little disagreement that this model applies to chemistry,
however, the debate between PK and Okasha is about whether we can
extend the “hidden microstructure” model of kinds to biological
kinds.
Arguments Against Using the PK Model for Biological Kinds
Okasha suggests two main lines of argument that lessen the probability that PK essentialist notion of kinds is applicable to evolutionary biology. The first line of argument is empirical which can be generally framed by referring back to Locke's observation that there is no principled way to distinguish between accidental and intrinsic qualities. Given that evolutionary theory doesn't restrict the possibility of changes (i.e., mutations, meiosis, and genetic recombinations) to any aspect of an group of organisms' phenotypic or genetic properties, it seems unlikely that a kind could have an immutable intrinsic essence. With no necessary enduring intrinsic property (that isn't also shared by other kinds), what type of property could defines the kind?
Arguments Against Using the PK Model for Biological Kinds
Okasha suggests two main lines of argument that lessen the probability that PK essentialist notion of kinds is applicable to evolutionary biology. The first line of argument is empirical which can be generally framed by referring back to Locke's observation that there is no principled way to distinguish between accidental and intrinsic qualities. Given that evolutionary theory doesn't restrict the possibility of changes (i.e., mutations, meiosis, and genetic recombinations) to any aspect of an group of organisms' phenotypic or genetic properties, it seems unlikely that a kind could have an immutable intrinsic essence. With no necessary enduring intrinsic property (that isn't also shared by other kinds), what type of property could defines the kind?
A loose analogy to illustrate the
problem would be to try to classify liquids according to their shape;
obviously, it will change depending on its environment. Another
problem with biological kinds that arises out of empirical
considerations is that often intra-species genetic and phenotypic
differences can be greater than (closely related) inter-species
differences. So, if there can be more differences within a kind than
without, on what grounds can we construct kinds that are based on
common intrinsic properties?
The conceptual argument again relates back to Locke's operationalism. Even if a set of (genetic/phenotypic/genotypic) properties were shared by all members of a kind and by no non-members, we would not consider having these internal properties necessary to membership. Suppose two members of a species produce an offspring lacking in one of the essential properties. We would probably still group the offspring with its parents.
So, it looks like the (internal) essentialist model for kinds doesn't fit well with the ephemeral nature of species in evolutionary biology. Should we then completely abandon the PK model in relation to biology? Okasha suggests that the PK model is still applicable to biological kinds so long as we relinquish the requirement that essential properties of kinds be intrinsic and instead replace them with relational properties.
Relational Kinds: Retooling the PK Model for Biology
A Relational property in this context means the essential property relates x to other xs. The relation is the property that tells us “in virtue of what organism x is a member of kind y”. In addition, an essential relational property cannot be shared by non-xs. Biologists use four basic relational properties to define species concepts: phenetic, interbreeding, ecological niche, and phylogenetic. While all 4 methods have their weaknesses, the phenetic concept is considered the weakest because it suffers from the same problems as internal essentialist concepts. Furthermore, a peculiarity of the relational species concepts is that two organisms could be molecule for molecule duplicates, but if they don't the share the relational kind property, they are considered to be of different species3.
The conceptual argument again relates back to Locke's operationalism. Even if a set of (genetic/phenotypic/genotypic) properties were shared by all members of a kind and by no non-members, we would not consider having these internal properties necessary to membership. Suppose two members of a species produce an offspring lacking in one of the essential properties. We would probably still group the offspring with its parents.
So, it looks like the (internal) essentialist model for kinds doesn't fit well with the ephemeral nature of species in evolutionary biology. Should we then completely abandon the PK model in relation to biology? Okasha suggests that the PK model is still applicable to biological kinds so long as we relinquish the requirement that essential properties of kinds be intrinsic and instead replace them with relational properties.
Relational Kinds: Retooling the PK Model for Biology
A Relational property in this context means the essential property relates x to other xs. The relation is the property that tells us “in virtue of what organism x is a member of kind y”. In addition, an essential relational property cannot be shared by non-xs. Biologists use four basic relational properties to define species concepts: phenetic, interbreeding, ecological niche, and phylogenetic. While all 4 methods have their weaknesses, the phenetic concept is considered the weakest because it suffers from the same problems as internal essentialist concepts. Furthermore, a peculiarity of the relational species concepts is that two organisms could be molecule for molecule duplicates, but if they don't the share the relational kind property, they are considered to be of different species3.
Is this a big problem? I'm not sure.
We could defend the relational view and say that there is an
extremely low likelihood of there being two or more co-existing
molecule for molecule duplicates that don't bear the same kind
relation so we shouldn't worry about this problem. On the other
hand, maybe the existence of the logical possibility of this
counter-intuitive outcome of relational kinds is
evidence of a problem. The reply to this worry is that just because
the logical consequence of the relational-concept kind is
counter-intuitive doesn't mean there's a problem—it only tells us
something about our intuitions. There is no logical problem
with relational kinds so long as we can accept the counter-intuitive
consequence, so we shouldn't worry.
It seems that with the substitution of relational for intrinsic properties, the PK model can be applied to biological kinds after all. The PK model can maintain the semantic role of kinds terms because now we can say that speakers who use the terms are intending to refer to some essential property beyond superficial appearances. Unfortunately, the applicability ends here because the PK model also implies that the essential property of a kind (its hidden structure) is also causally responsible for its superficial properties. While this is true of chemical kinds, this isn't necessarily the case for biological kinds. An organism's belonging to a particular chunk of the genealogical nexus (or occupying an ecological niche, etc...) isn't the proximal cause of its superficial properties—there is only an indirect causal relationship.
For these reasons Okasha concludes that the PK model is only half-right when applied to biological kinds. On the PK model essences play both a semantic and causal/explanatory role, but “there is no a priori reason why the same thing should play both of these roles4.” While I agree with Okasha here I think there is something a little disconcerting about decoupling these two roles. It appears we lose a degree of objectivity and predictive power.
It seems that with the substitution of relational for intrinsic properties, the PK model can be applied to biological kinds after all. The PK model can maintain the semantic role of kinds terms because now we can say that speakers who use the terms are intending to refer to some essential property beyond superficial appearances. Unfortunately, the applicability ends here because the PK model also implies that the essential property of a kind (its hidden structure) is also causally responsible for its superficial properties. While this is true of chemical kinds, this isn't necessarily the case for biological kinds. An organism's belonging to a particular chunk of the genealogical nexus (or occupying an ecological niche, etc...) isn't the proximal cause of its superficial properties—there is only an indirect causal relationship.
For these reasons Okasha concludes that the PK model is only half-right when applied to biological kinds. On the PK model essences play both a semantic and causal/explanatory role, but “there is no a priori reason why the same thing should play both of these roles4.” While I agree with Okasha here I think there is something a little disconcerting about decoupling these two roles. It appears we lose a degree of objectivity and predictive power.
The Purpose of Kinds and Worries
About Decoupling Essences from Causal/Explanatory Roles
The fact that there are so many different species-concepts can raise worries about conventionalism. Why should we consider one species-concept over another? Consider that mammals are often grouped according to either phylogenetic or breeding or ecological niche concepts but bacteria are grouped according to degree of variability in section (?) 16 of sRNA. If one strand of bacteria has greater that 1% variability from its “parent” strand then it is considered to be a new species of bacteria5. Clearly, biologists are picking and choosing their species concepts based on what is useful to their research aims (and contingent upon the sophistication of their measurement techniques). If biologist were really classifying according to essential kinds, why isn't there just one concept of species?
The fact that there are so many different species-concepts can raise worries about conventionalism. Why should we consider one species-concept over another? Consider that mammals are often grouped according to either phylogenetic or breeding or ecological niche concepts but bacteria are grouped according to degree of variability in section (?) 16 of sRNA. If one strand of bacteria has greater that 1% variability from its “parent” strand then it is considered to be a new species of bacteria5. Clearly, biologists are picking and choosing their species concepts based on what is useful to their research aims (and contingent upon the sophistication of their measurement techniques). If biologist were really classifying according to essential kinds, why isn't there just one concept of species?
This isn't necessarily a problem, but
seems there is a lot more lateral flexibility in relational kinds
than there is with intrinsic kinds, such as in chemistry. That is
to say, organisms can be grouped into kinds various different ways at
the same “level of grain”, whereas the microstructure of chemical
kinds is more restrictive in respect to method of classifying at that
grain. Furthermore, it seems that organizing according to intrinsic
properties provides a greater ability to make predictions about the
properties of kind than you might make with relational kinds in
biology. However, given the nature of evolutionary biology, I'm not
sure there is a choice in the matter since--as has been
shown--defining biological kinds according to intrinsic properties is
a non-starter.
This brings us head to head with the final issue I wish to discuss: what is the purpose of a classificatory system? On the PK model it seems that the implicit answer to this question is that kinds are meant to be scientifically useful; viz, “to provide the greatest possible predictively useful generalizations6”. Predictively useful generalizations usually require we know something about the kind's causal structure. But in biology we don't have this information, nor do we need it because it has a different purpose of classification.
The purpose of classification (i.e., species concepts) in biology is to identify “units which we believe play an important role in the evolutionary process7”. Of course, knowing an organism's species (defined in one of the relational concepts) does provide the ability to make predictions about behaviour and morphology, but, again, this is not the primary purpose of having the classification.
Concluding Thoughts
To conclude I think it's interesting to consider how species concepts are defined and why one would be used over another in a particular situation. There seems to be an interesting reciprocal relationship between the empirical and the analytic concepts. The species concepts are analytic but they are informed by empirical considerations. Consider the breeding concept. Certainly, whether one group breeds with another is a matter of empirical observation but the decision to define a kind based on a particular concept of inter-breeding rather than another is analytic. However, that particular kind concept will also have been shaped by past empirical observations. At some point, the biologist has to “break in” with a an imperfect concept. The relationship between the analytic and synthetic in the context of natural kinds is something I'm interested in exploring more, so I encourage comments on this issue!
This brings us head to head with the final issue I wish to discuss: what is the purpose of a classificatory system? On the PK model it seems that the implicit answer to this question is that kinds are meant to be scientifically useful; viz, “to provide the greatest possible predictively useful generalizations6”. Predictively useful generalizations usually require we know something about the kind's causal structure. But in biology we don't have this information, nor do we need it because it has a different purpose of classification.
The purpose of classification (i.e., species concepts) in biology is to identify “units which we believe play an important role in the evolutionary process7”. Of course, knowing an organism's species (defined in one of the relational concepts) does provide the ability to make predictions about behaviour and morphology, but, again, this is not the primary purpose of having the classification.
Concluding Thoughts
To conclude I think it's interesting to consider how species concepts are defined and why one would be used over another in a particular situation. There seems to be an interesting reciprocal relationship between the empirical and the analytic concepts. The species concepts are analytic but they are informed by empirical considerations. Consider the breeding concept. Certainly, whether one group breeds with another is a matter of empirical observation but the decision to define a kind based on a particular concept of inter-breeding rather than another is analytic. However, that particular kind concept will also have been shaped by past empirical observations. At some point, the biologist has to “break in” with a an imperfect concept. The relationship between the analytic and synthetic in the context of natural kinds is something I'm interested in exploring more, so I encourage comments on this issue!
1Okasha,
p. 193.
2Ibid,
p. 195.
3Ibid,
p. 201.
4Ibid,
p. 203.
5Chana
Palmer-Davis, world famous geneticist and my sister.
6Ibid,
p. 209.
7Ibid.
hi arlene. thanks for taking the time to comment. the question is, in virtue of what is a dog, a dog? That is, what is the thing (ie, property or properties) that we can find in every dog and not in other species. For example teacup Chihuahuas and great danes are both dogs. why do we say they are both dogs when a Chihuahua probably has more in common with a rodent, while the great dane probably has more in common with a small horse. It it that they both have 4 legs? But there are other animals that have 4 legs and we don't say they are dogs. Is it that they have a great sense of smell? But there are other animals that have a great sense of smell, and we don't say they're dogs. So, maybe a dog is a dog because of a certain collection of properties; lets say a dog has 4 legs, great sense of smell, and a tail. But this also describes lots of other animals that we don't consider to be dogs. So, that's just one problem. The other problem is that species evolve over time. It's not like all of a sudden there were dogs and lots of different kinds of dogs. Also, the proto-dog (the wolf) from which the dog evolved is considered to be different species. At some point a population of wolves "became" dogs. How and where do we draw the line? On what basis?
ReplyDeleteHelllo mate great blog post
ReplyDelete