ARISTOTLE AND THE UNITY CONDITION FOR SCIENTIFIC DEFINITIONS ALAN CODE [Discussion of DAVID CHARLES: ARISTOTLE ON MEANING AND ESSENCE] Like David Charles, I am puzzled about the relationship between Aristotle s account of definitional practice in the Posterior Analytics and his actual explanatory practices in the biological works. The former involves some kind of commitment to their being a single unifying explanatory feature for each natural kind, whereas the latter seems to offer explanations that do not conform to this commitment. What are we to make of this? Has Aristotle changed his mind in the about kinds and essences due to the results of sustained biological investigation? If so, does his departure from the Posterior Analytics model threaten to undermine the grounds he has for believing in the existence of natural kinds? If not, have we misunderstood the requirements on unity that arise from the Posterior Analytics? To address these issues we need first to get clear on exactly what that work requires for the unity of a scientific definition, and in our discussion today I do not propose to solve the problems connected with apparent differences between the two parts of the corpus, Rather we will use his excellent and nuanced discussion of the constraints on definition that Aristotle operates with in the Posterior Analytics as a framework for examining some foundational topics about scientific method. I will briefly discuss how they are at work in the biological treatises that he discusses in Chapter 12 of his book, but the main focus will be on understanding his treatment of scientific definition. His discussion of the Posterior Analytics helps greatly in formulating and investigating methodological issues about Aristotle s practices in the life sciences, and also leads to some puzzles as to how to understand that methodology. I would like to start with a brief, and I hope sympathetic, account of the main features of David s interpretation of the Posterior Analytics recommended procedures for arriving at scientific definitions. Only towards the end will I will turn briefly to a 1
consideration of the procedures that seem to be employed in the biological works. Does Aristotle practice there what he preaches elsewhere? As a part of what David dubs the DEFINITIONAL CONSTRAINT Aristotle in the Posterior Analytics commits himself to a UNITY CONDITION: UC: The appropriate method for arriving at definitions should legitimize the claim that the definiens is a unity. [p.192; 7.4] Definitions are arrived at by some procedure or other, and this is one of the constraints on a procedure if it is to be appropriate to the task at hand. There are, of course, other constraints as well, and the manner in which UC is exemplified in a method is not independent of other constraints and the way in which they work their way into the method in question. One such additional constraint that is also a part of the DEFINITIONAL CONSTRAINT is the PRIORITY CONDITION: PC: The appropriate method for arriving at a definition of C should legitimize the claim the some feature is prior to all other features of C. [p.191; 7.3] Like UC, this too is a constraint on appropriate procedure or method. However, since there are different types of priority the PC needs further clarification before we can see how it interacts with UC, and David provides the needed clarification with the SIMPLE CONDITION: SC: The appropriate method for arriving at definitions should give us good grounds for the claim that: (a) There is one feature which is a prior feature of the relevant kind, and 2
(b) That feature should explain why the kind possesses all the other necessary features specified in the definition. [p.195; 7.4] In light of SC we can see that the way in which a method of arriving at definitions can legitimize the priority of some feature is to give us good grounds for accepting that there is a single feature of the kind C that is prior to other features of C in that this one feature explains all of the other necessary features that are specified in the definition. Let us call this single feature the UNIFYING FEATURE. The SC commits Aristotle to the claim that an appropriate method for arriving at a definition must by itself provide good grounds for accepting the existential claim that there is a UNIFYING FEATURE kind C. It also presupposes that the definitions arrived at by the appropriate method will invoke a plurality of features. It does not require that the UF explains everything that can be explained about the kind, but rather embodies the more modest claim that it will explain all of the other features in the definition. That is, the UF is one component of the definition, but there are other components, and the fact that UF is characteristic of things of type C explains why Cs have the other features mentioned in the definition of that kind. The other features are not contingent accidents of C, but themselves necessary features, and the UF not only explains why the other features pertain to C, but also why the must. A Unifying Feature that plays this central role in a definition of C constitutes the essential nature of Cs, and an appropriate method for defining C must single out the essence of C as the core component of the definition of C and show how each and every necessary feature specified in the definition is indeed a necessary feature of one and the same subject, namely C itself. If the procedure is successful in producing this result, one will have simultaneously established that C is itself a Natural Kind and will have determined the essential nature of that Natural Kind. The essence of a C is what makes something a C, and by identifying the Unifying Feature that explains the other necessary, defining features of C, one has identified the essence. By showing how the Unifying Feature explains these other features one shows how it is the feature that makes a C what it is, namely a C. Additionally, be grounding the claim that there is an essence for C the 3
procedure also establishes that C is indeed a natural kind, for it shows that the plurality of features that go to make up the definition of C are not a weak or accidental conglomeration, but rather are unified by a single explanatory nature. So much for the elucidation of the Priority Condition. Now, back to the Unity Condition, the requirement that an appropriate method for arriving at a definition should legitimize the claim that the definiens is a unity. The Priority Condition, or strictly speaking its elucidation through the Simple Condition, shows how on Aristotle s own recommended account of definitional practice the UC is honored. The way in which the method of definition legitimizes the claim that the definiens is a unity is to that the deployment of the method by itself provides good grounds for the claim that the components of the definiens are, each and every one of them, intrinsic features of one and the same subject. The reason for this is that (i) one of the features is the essence of the subject of definition, and (ii) this one unifying feature explains how all of the other features in the definition are necessary features of that same subject. In short, the Simple Condition requires that when a definition is arrived at there is a unifying feature, and a unifying feature does just that it unifies. It is the existence of the unifying feature that guarantees that the definiens constitutes a unity. Although the conditions that we have been exploring here are conditions on the practice of definition, it should be obvious from the fact that the Simple Condition assigns an explanatory role to the Unifying Feature that on Aristotle s view explanatory practices are not independent of the definitional practices. As analyzed by David, the procedure for arriving at definitions that Aristotle advocates in the Posterior Analytics has three stages, each of which interacts with explanatory proof procedures. Although nothing in the conditions examined so far requires a syllogistic structure for explanation and proof, that is in fact what Aristotle does require. Both the UC and the SC are aspects of a general condition of adequacy for definitional procedures that David calls the DEFINITIONAL CONSTRAINT (DC): 4
the appropriate method for coming to know a definition should give us grounds for the claim that this definition is correct, and do so in a way which essentially involves the use of deduction (of some kind) [p.189; VII.3] The more specific UC and SC make explicit some of the criteria for the correctness of a definition. The definiens must be a unity, and the unity is guaranteed by its containing a unifying feature that explains the necessary connection of the defining feature of a kind to the kind itself. Furthermore, the procedures utilized must give good grounds for thinking that the definition meets these criteria, and hence is correct. This general DEFINITIONAL CONSTRAINT tells us that the procedures that give us these good grounds must make essential use of deduction. In Aristotle s hands this means that the procedures must make essential use of syllogisms. The interpretation of the stages involved in Aristotle s proposed method has been the subject of a great deal of fruitful scholarly discussion and controversy. Here I will not be entering into the controversy, but simply outlining the interpretation for which David argues at considerable length, and with considerable ingenuity. The procedure itself has three stages. At the first stage of this three stage method the one inquiring into the definition of some kind already has a definition of a sort, a definition of what a name (or name-like expression) signifies. At this stage one need not know whether or not the kind in fact exists. Even if there is no kind corresponding to a term, one can nonetheless have an account of what that term signifies. However, such nominal definitions are not definitions of kinds. If the kind does not exist, then there can be no definition of the kind. This is because a real definition of a kind must contain a Unifying Feature which explains why all of the features invoked in the definition are necessary features of one and the same subject, the kind that serves as the object of the definition. If one can show that the features specified in the definition are necessary properties of some kind C, then C must exist. At the second stage one establishes the existence of something corresponding to the nominal definition. At this stage one knows that whatever it is that is signified by the 5
name (or name-like expression) in fact exists. At this second stage one would not yet know whether this is in fact a kind, because simply showing that something exists corresponding to the nominal definition does not establish that whatever this is it has an explanatory, unifying essence that makes it a kind. Finally, at the third stage one comes to know what the essence of the kind C is. On David s interpretation, the nominal definitions are not definitions of a kind, and there are three different but related types of definitions of kinds. One of these is a complex formula of the following sort: Thunder is noise in the clouds caused by the quenching of fire. [p.43] Let us call this the scientific definition This definition not only specifies the cause, but also contains two other elements that themselves are both necessary features of thunder. In fact, it corresponds to a proof of the existence of thunder of the following type: Fire being quenched belongs to the clouds. Noise belongs to fire being quenched. So: Noise belongs to the clouds. Given that the definition of the signification of the name thunder is noise in the clouds, a demonstrative syllogism that has as its conclusion the proposition that noise belongs to the clouds is a proof that thunder exists. The scientific definition is, according to Aristotle, the same account as this proof, but differs in its arrangement. Notice that the phrase fire being quenched functions as the explanatory middle term in this proof. This is the efficient cause of thunder, and in this case the efficient cause is the essence of the phenomenon defined. The existence of the kind is desmonstrated by a syllogistic proof that makes use of the essence of the natural kind as the explanatory 6
feature that relates the other two components in the scientific definition to each other, showing why one is an attribute of the other. However, it is important to point out that this existence proof is a demonstration of the existence of thunder, but the subject of the conclusion is not thunder, but rather the clouds. The explanatory essence of thunder is supposed to do more than explain the existence of the phenomenon. It is also supposed to explain why that phenomenon has the necessary features that are specified in its definition. The explanatory essence (i.e., fire being quenched in the clouds) also explains why thunder is a noise in the clouds. Here the explandum is not the existence of thunder, but rather a proposition about thunder, that it is a noise in the clouds. Although Aristotle doesn t give us the proof, he tells us that this is the conclusion of a demonstration of what it is (94a7-9). That is, the proposition that thunder is a noise in the clouds is both a definition and the conclusion of a proof. This kind of definition is distinct from what I have been calling the scientific definition. It contains the terms of the scientific definition minus the explanatory essence. The proof of this definition is what shows that the other elements of the scientific definition are necessary features of the phenomenon of thunder. We should now turn briefly to a consideration of a problem that David raises in reconciling this account of definition with the actual practices of definition and explanation in the biological works. In 12.5 he that many of Aristotle s biological explanations about fish do conform to our expectations, and explain various necessary features of fish as due to their nature as swimmers. In general, Aristotle holds that in zoology final causes are prior in explanation to efficient causes, and hence if there is an element of the definition of an animal that is prior to and explanatory of the others it must be a teleological cause. In the case of fish it is their characteristic mode of locomotion, swimming. However, in 12.6 he argues that Aristotle cannot account for all of the features that are part of the nature of fish in this accordance with the Posterior Analytics model. 7
For instance, fish are gluttons. This is part of their characteristic mode of nourishment. The reason for this is that their digestive system is terribly inefficient and most of the food passes through undigested. So they need to take in a lot of nourishment. Since they are gluttons it is a good thing for them (well, some of them) that they have mouths underneath their tip. This makes it a little harder to get the food in, and helps prevent them from killing themselves from overeating. As David notes, this train of thought makes no appeal to the nature of fish qua swimmers, and does not even attempt to explain why they have such lousy digestive systems. Turning from mode of nourishment to mode of reproduction, David cites further examples of explanations that do not make use of the essential nature of fish as swimmers, but instead appeal to such factors as habitat or material constitution. The upshot is that Aristotle is definitely not employing a model of explanation that traces all of the necessary features of fish to their essential nature as swimmers. Indeed, there is no one, single feature that figures in all of the various explanations of the necessary features of fish. The point is not that there are features of fish that are explained in non-conforming ways. The problem is that there are features that are part of the distinctive, characteristic nature of fish that are not explained by a privileged, teleological cause. David s examples are numerous and compelling. What are we to make of this? It would be grasping at straws to suggest that Aristotle thinks that in cases where he doesn t explain some characteristic activity of fish by appeal to their nature as swimmers that are simply because he hasn t yet found the true explanation yet, and what he is giving us is provisional and incomplete. David asks Are we, at this point, witnesses to the collapse of a brilliant research programme? [p.336] The stakes are high. Without a unifying teleological feature Aristotle cannot utilize the ideas of the Posterior Analytics to mark out the essence of a biological kind, and the system of animal classification would go without a foundation. The interpretation that David favors is a kind of middle road between outright rejection of the Posterior Analytics model and a valiant attempt to hold onto it at all costs. The central idea of the middle road approach is that of INTERACTIVE UNITY. This concept will be the focus of the remainder of my comments. On this approach one 8
countenances a plurality of basic explanatory factors, and does reject the idea of a single, common unifying feature for all of the intrinsic, necessary features contained in the account of the nature of an animal. We give up the simple model according to which just one feature in the definition is basic, and all others are derived. On the new model, there are many derived features that cannot be derived from any one single basic feature, but can be derived from a set of such basic features. The unity of a biological kind is due to the fact that a plurality of basic features interact in such a way as to give rise to the derived features that are also a part of the characteristic mode of enlivening activity for that type of organism. This Interactive Unity of the nature of an animal accounts for the unity of the basic, non-derivative features of the animal s nature as well as the derivative elements, and stems from the way in which the explanatory structure of the science is itself a unified system. According to David this approach retains quite a bit of the Posterior Analytics account. It still allows one to distinguish the basic from the derived features in the nature of an animal, and even allows within the class of basic features for some to be essential (as opposed to merely necessary). One might add that this account retains the interdependence of definitional and explanatory practices. However, he also thinks that there are definite respects in which this new model is incompatible with the old. In particular, he now treats both a teleological nature and given particular material natures of an animal as basic explanatory principles whose presence is not otherwise explained. This makes it impossible for him to treat the prior, teleological nature as prior because it explains all of the other necessary features specified in the definition, and hence deprives him of the idea that the unity of a kind is due solely to an essence that explains the other features in the nature. Let us for present purposes treat the nature as composite in the sense that each kind of animal has both a formal and a material nature. Although Aristotle also at times appeals to habitat to explain distinctive characteristics of a species, he doesn t think that animals also have an environmental nature. Nonetheless, habitat too is a part of the natural world and makes its own contribution to the lives of animals. Some features are due to form. 9
For instance, the fact that animals reproduce is due to their form, that is to say, their soul. However, the precise manner in which the various species reproduce is not due solely to form. There are facts about the material constitution of the animal that are not due to the form, and both they and environmental factors are determinants of the specific ways in which the reproductive functions of animals are instantiated. Examples would be the fact that fish produce a large number of eggs because the inhospitable environment results in most of them are destroyed, and the fact that their white part is indistinct is due to the cold and earthy nature of the matter. [p.333] Additionally, even the unity of the form itself (i.e., the soul) cannot be explained by singling out a single, privileged essential activity and deriving the other functions of soul from it. I agree with this assessment of what one finds in the biological works, and also agree that this doesn t conform to the strictures of the Simple Condition: SC: The appropriate method for arriving at definitions should give us good grounds for the claim that: (a) There is one feature which is a prior feature of the relevant kind, and (b) That feature should explain why the kind possesses all the other necessary features specified in the definition. [p.195; 7.4] However, in light of the fact that the biological works so clearly do not conform to this condition, I would like to suggest that we examine once again its credentials. As you will remember, it was introduced as a clarification of the Priority Condition. The Priority Condition does not invoke the idea of a Unifying Factor, but rather states: PC: The appropriate method for arriving at a definition of C should legitimize the claim the some feature is prior to all other features of C. [p.191; 7.3] 10
The concept of Integrative Unity that David has suggested for the biological works is compatible with the idea that a single feature is prior to the others, much as form can be prior to matter, without having to be the sole factor that explains them. The more specific version of PC, the SIMPLE CONDITION, was motivated by a consideration of Aristotle s rejection of the method of division as a method of definition on the grounds that it cannot satisfy the PC and the UNITY CONDITION: UC: The appropriate method for arriving at definitions should legitimize the claim that the definiens is a unity. [p.192; 7.4] However, if Integrative Unity were indeed a form of unity sufficient to underwrite the unity of a natural kind in the biological works, then it would be sufficient here as well. The complaint he lodges against the method of division need not be taken as support for the idea that there must be a single UNIFYING FEATURE, but rather as support for the weaker claim that the elements of a definition must be unified by showing how they are interconnected in an explanatory nexus. The problem with the method of division is not that it doesn t have a common cause for being terrestrial and being two-footed. The problem is that he doesn t have any explanation as to how they are related to a single subject, the species man. If this is along the right track, then although the methodological practices of the biological works go considerably beyond the Analytics framework, perhaps are not at odds with it. In any case, the issue is a fundamental one, and goes to the heart of Aristotle s views on kinds and essences. The book that we are all discussing today has advanced the scholarly discussion of these issues on a number of fronts, and has enabled us to formulate and examine this topic on unity in a clearer and more articulated way. 11