The term teleology locates a series of connected philosophical questions. If we grant that there is such a thing as purposive or goal-directed activity (as we must, since, for example, a political campaign aimed at victory represents a clear, uncontroversial case), we may ask the following questions:
- By what criteria do we identify purposive activity?
- What is the nature of the systems that exhibit purposive activity?
- Does the nature of purposive activity require us to employ special concepts or special patterns of description and explanation that are not needed in an account of nonpurposive activity? And if we grant that there are objects and processes which perform functions (again, as we must, since no one would deny, for instance, that the human kidney performs the function of excretion), we may ask:
- By what criteria do we identify functions?
- What is the nature of the systems that exhibit functional activity?
- Does the description of functions require special concepts or special patterns of analysis?
|discussions of teleology|
These six questions have been formulated with the help of a distinction between purposive and functional activity. Although the distinction is not always drawn in discussions of teleology, it is desirable for a number of reasons.
It seems, at least prima facie, that the criteria of functional activity are quite distinct from the criteria of purposive activity: urine excretion, for example, seems to be a function by virtue of its role in the economy of a living organism, whereas activity seems to be purposive in virtue of the manner in which it is controlled.
Thus, it seems at least logically possible that a purposive activity could perform no function, and that a function could be performed without purposive activity. Moreover, in view of this fundamental conceptual difference between purpose and function, we should expect the analysis of purposive and functional activity to show differences in logical pattern.
|function and purpose|
On the other hand, it also seems clear that there are close connections between function and purpose; thus the final question: (7) What is the relation between ascriptions of function and ascriptions of purpose?
A number of writers have proposed definitions of “goal-directed” or “purposive” action that leave open the question whether the action is intentional or in any way involves consciousness. R. B. Braithwaite suggests, as a behavioral criterion of goal-directed activity that either may or may not be goal-intended, “persistence toward the goal under varying conditions.”
|R. B. Braithwaite|
This is a condensed version of very similar criteria offered by R. B. Perry, E. S. Russell, and A. Hofstadter. All presuppose that a goal may be identified and that both persistence and sensitivity to varying conditions may be located by reference to the goal.
E. C. Tolman adds the requirement that purposive activity show “docility,” that is, some improvement in reaching the goal in the course of successive trials. But docility, however important it may be in the total picture of biological purposiveness, is surely not part of the criterion of purposiveness. Any abilities that are in fact learned could, in logical principle, be innate.
This criterion, in Braithwaite’s form, is of course susceptible of considerable refinement; Braithwaite himself (in Scientific Explanation), for example, proposes a way of identifying variations in conditions as relevant variations for applying the criterion. Further possible refinements will be discussed in the next section.
|R. B. Perry|
The apparent circularity in the criterion—defining “goal-directed” in terms of a “goal”—is not serious. The location of persistence, sensitivity, and a goal may proceed together by a method of successive approximations.
For example, a pattern of animal behavior may appear persistent and lead to a tentative identification of a goal, and the identification may be checked by looking for sensitivity to conditions or further evidence of persistence. A hypothesis about any one of the three—goal, persistence, sensitivity—can be confirmed by investigating either of the other two.
It seems clear that there are behavioral criteria for identifying purposive action, not only of human beings but also of other animals and of artifacts such as self-guided missiles.
|E. S. Russell|
A pilot who watches a rocket approach in spite of his evasive maneuvers would rightly have no doubts about either the goal-directedness of the rocket’s movements or the identity of its goal.
No doubt the actual criteria of purposiveness that have been proposed suffer various shortcomings. In particular, they seem to lay down a necessary but not a sufficient condition. However, most philosophers would regard the program of seeking behavioral criteria as sound.
Nature of Systems Showing Purposive Activity
Is it possible for the philosopher, as distinct from the biologist, psychologist, or communications engineer, to say anything illuminating about the nature of the systems—men, mice, and missiles—that engage in purposive activity?
He can at least examine more closely the behavioral criteria of purposiveness, in order to see whether there might be covert reference to the nature of the system in the criterias’ actual application.
A critic of the behavioral criteria might remark that a river is persistent in reaching the sea and is sensitive to the conditions necessary for reaching the sea—it detours all obstacles—but we would not call the flowing of a river purposive, nor would we call the sea or reaching the sea its goal. In short, the critic might say, a river is not the sort of thing to which we ever ascribe purposiveness.
Directive correlation. A number of philosophers, including Braithwaite, Ernest Nagel, George Sommerhoff, and Morton Beckner, have proposed ways of avoiding the difficulty about rivers and the like.
Although there are differences in their accounts, they all adopt the strategy of regarding an activity as purposive only when its goal seeking character is the outcome of relatively independent but dovetailing processes.
Sommerhoff, for example, defines “purposive behavior” with the help of a concept he terms “directive correlation.” Two variables, such as the position of a moving target and the direction in which an automatic target-tracking mechanism points, are said to be directively correlated with respect to a goal state (in this case, the state in which the mechanism points at the target) whenever:
- The two variables are independent in the sense that any value of one is compatible with any value of the other;
- The actual value of both, at a given time, is at least in part causally determined by the prior value of a “coenetic” (steering) variable (in the example, the coenetic variable is the same as one of the directively correlated variables, namely, the position of the moving target); and
- the causal determination is such that the actual values of the directively correlated variables are sufficient for the realization of the goal state. Sommerhoff then defines “purposive behavior” as directively correlated behavior in which the coenetic variable is identical with one of the directively correlated variables.
Stipulations (2) and (3) make the notion of two processes dovetailing so as to achieve a goal as precise as the notion of causal determination; and stipulation (1) specifies that the processes must be independent. The requirement of independence rules out such cases as the river, for the direction in which a river flows is not independent of the lay of the land.
Sommerhoff ’s analysis is not without difficulties (see Nagel and Beckner), but it is undoubtedly correct in general approach. A system S that could exhibit directive correlation would satisfy a number of prior conceptions about purposive behavior; for instance, that S would employ information about its environment, particularly about an aspect of the environment associated with the goal, and that the behavior of S would be dependent upon a specialized physical hookup, such as some sort of circuitry.
It is now possible to suggest a schema for constructing a criterion of purposive activity that includes both a necessary and a sufficient condition and that incorporates some reference both to the empirical character of the activity and to the nature of the system that engages in it. Activity is purposive if and only if it exhibits sensitivity and persistence toward a goal as a result of directive correlation.
Need for Special Concepts or Patterns of Description and Explanation
Purposive activity, in the analyses of Braithwaite and Sommerhoff described above, does not involve a special kind of causality but only a special organization of ordinary causal processes. If these analyses are correct, both living organisms and artificial machines are capable of purposive activity.
If, therefore, special concepts or patterns of description and explanation are not needed in the case of purposive machines, it would appear that they are equally unnecessary in the case of organisms. Many philosophers have drawn this conclusion, and it must be admitted that accounts like Braithwaite’s and Sommerhoff ’s constitute powerful arguments in its support.
There is room for some doubt, however. Even if we grant that purposive activity can be defined in terms that are equally applicable to organic and inorganic systems, it does not follow that all purposive activity can be explained on the model of inanimate activity.
The most serious doubt concerns those purposive activities that may be described as the acts of agents, such as acts deliberately undertaken for the sake of a consciously envisaged end. Suppose, for example, that some or all of these acts of agents are in principle unpredictable—a view accepted by some philosophers.
Then, if they can be explained at all, their explanation is essentially post hoc. The pattern of such explanation is not yet properly understood; nevertheless, there is at least some doubt that it can dispense with the conception of following a rule. But these considerations raise questions that cannot be pursued here.
When we assert truly—for example, that a function of the kidney is the excretion of urine—precisely what relations must hold between the kidney and excretion? It has been proposed, for example by Nagel, that such teleological terms as purpose and function can be eliminated in the following way:
An expression such as “A function of the kidney is the excretion of urine” is translated into the nonteleological expression “The kidney is a necessary (or necessary and sufficient) condition of urine excretion.”
In general we may interpret Nagel as proposing a translation schema—For “F is the function of A,”write “A is a necessary (or necessary and sufficient) condition of F”—that dispenses with teleological language and that also provides part of a criterion (a necessary condition) for identifying functions.
At best, however, Nagel’s schema must be modified, for the possession of kidneys is neither a necessary nor a sufficient condition of urine excretion. It is obviously not sufficient; but it is also not necessary, since urine can also be excreted by various artificial devices. (If it is objected that these devices are themselves a sort of kidney, then the statement that a kidney is necessary for excretion reduces to a tautology.)
Moreover, the translation schema is much less plausible when applied to organic functions that are ordinarily accomplished in distinct ways. Temperature regulation, for example, is a function of man’s body hair; but hair is not necessary for heat regulation, since the function may be performed by other physical and physiological mechanisms.
When we ascribe a function to the kidney or to body hair, we seem to be saying no more than that these structures contribute to certain processes; we leave open the question whether they are necessary or sufficient for the processes. The relation “contributing to” may be defined without employing teleological language.
Let F be a process, some or all of which takes place in system S; and let A be a part of, or a process in, S. Finally, let the terms “S-like,”“F-like,” and “A-like” refer, respectively, to all those entities that answer to the definition of the terms employed in specifying S, F, and A. (In the example “A function of the kidney in vertebrates is the excretion of urine,” all vertebrates are S-like, all cases of urine excretion are F-like, and all kidneys are A-like.)
Then “A of S contributes to F” if and only if there exist S-like systems and states or environments of these S-like systems in which F-like processes occur and the possession of A-like parts or processes is necessary for the occurrence of F-like processes.
On this definition, we may say that in general a man’s kidney contributes to the excretion of urine and that body hair contributes to heat regulation. And if we adopt the translation schema “For ‘F is the function of A in S,’ write ‘A contributes to F in S,’” we may say, even in the case of a man whose bad kidneys have been bypassed to an artificial kidney, that the function of his flesh-and-blood kidneys is still the excretion of urine; they merely fail to perform it.
Nature of Systems Showing Functional Activity
Nagel’s translation schema and the above modification of it provide a way of translating a teleological statement T1 into a statement T2 that does not employ explicitly teleological terms. Therefore, the satisfaction of T2 by a given A, F, and S is a necessary condition of F’s being a function of A.
It is, however, not a sufficient condition; we may not in general translate T2 into T1. We would not say, for example, that the function of the ground is to hold up the rocks even though, in our technical sense, the ground contributes to the holding up of rocks. It would seem that out of the whole set of “contributing” cases, only a very restricted subset could be regarded as functions.
How may this subset be specified? We ordinarily attribute functions to two sorts of systems, artifacts and living things. We may consider first a simple artifact such as a cooking pan. We ascribe a function to the whole pan: cooking.
Moreover, we also ascribe functions to parts and properties of the pan insofar as they contribute to its usefulness in cooking. For example, it is natural to think of the handle as providing a grip, of the rivets as fastening on the handle, and so on.
In short, whenever we are prepared to acknowledge a single function F, we are also prepared to acknowledge a hierarchy of functions, with F at the top and the functions at each lower level contributing to all those above them.
The assignment of functions to living organisms proceeds on the same principle. There are two organic processes that are regarded as fundamental, the maintenance of life and reproduction. Alternatively, these two processes may be thought of as contributing to a single process, the maintenance of a species, which stands at the top of all functional hierarchies.
The fundamental processes thus play a defining role in the identification of functions. The following schema lays down a necessary and sufficient condition of functional activity: F1 is a function of A in S if and only if A contributes to F1 in S; and F1 is identical with or contributes to F2 in S, where F2 is either a purpose for which the artifact S is designed or the process of maintenance of the species of which S is a member.
The concept of an artifact may be interpreted quite broadly in order to include not only things like cooking pans but also all cultural products, such as works of art, language, and legal institutions.
It makes sense, for example, on the above analysis and on this interpretation of artifact, to ask “What is the function of Ophelia in Hamlet?” and “What is the function of verb inflections in Japanese?” The justification for regarding maintenance of the species as a fundamental function, serving a logical role in functional analysis, is examined below.
Need for Special Concepts or Patterns of Analysis
The definition of functional activity offered above provides a way of interpreting ascriptions of functions without using explicitly teleological expressions. However, there is a sense in which many of the concepts that are employed in the ascription of functions are implicitly teleological.
Consider, for example, the concept of an “escape reaction.” It is applied to a great variety of animal movements, such as flying up, forming dense schools, withdrawing into burrows, jumping into water, and gathering under the mother.
These diverse reactions probably have no relevant feature in common other than a functional one; they all, in the technical sense, contribute to the avoidance of death by predation.
Such functional concepts are common in the theory of animal behavior, in all branches of natural history, in physiology, and indeed in everyday language. The terms that we most commonly use, for example, in describing machines are defined functionally.
The view that teleological language can be eliminated from the language of science may be true; again, the most difficult cases concern human agency. But the program of eliminating teleological expressions even from biological theory must involve more than the elimination of such terms as function, purpose, goal, and in order to.
If there is any point in eliminating these terms, there is just as much point in eliminating all concepts that are defined functionally, for “The function of this movement is to escape from a predator” is equivalent in asserted content to “This movement is an escape reaction.”
It is obviously true that the movement in question can be described, without employing the term escape reaction, as a movement that contributes to the avoidance of a predator. But if we eliminate the term escape reaction, we have excised from the language the term that applies not only to this movement but to all the diverse movements, in a variety of taxonomic groups, that serve this function.
The ascription of functions, therefore, does not require either an explicit or an implicit teleological vocabulary. It should be recognized, however, that the elimination of implicitly teleological expressions (concepts that are defined functionally) would result in a language for biological theory that would bear very little resemblance to the existing language.
Moreover, the difference would not be superficial; the rejection of functional concepts would amount to the rejection of a powerful and fruitful conceptual scheme. Our picture of living organisms as organized functional hierarchies is an essential part of the theory of natural selection; it is the foundation of physiology and morphology; and it is the basis of the medical view of disease as derangement of function.
It is the fruitfulness of this conceptual scheme, embodied in a network of connected functional concepts, that constitutes the justification for assigning to maintenance of the species its central logical role in the ascription of functions.
Relation between ascriptions of function and of purpose
|Relation between ascriptions of function and of purpose|
We have drawn a sharp distinction between functional activities, which contribute to a “fundamental” process, and purposive activities,which are persistent, flexible patterns of directively correlated behavior. It is clear, however, that function and purpose are closely connected—so closely, indeed, that many writers have failed to see the distinction. These connections may be described as follows:
- Whenever we construct an artifact as an aid to our own purposive activities, we are willing to ascribe functions to the artifact and to its parts and properties.
- Many but by no means all organic functions are served by purposive activities. For example, temperature regulation in the mammals involves directive correlation, whereas the excretion of urine does not.
- Conversely, every organic mechanism that provides an organism with the means of purposive activity serves the function of maintenance of the species.
This is an empirical fact. It does not mean, however, that each case of purposive activity, when it occurs, performs a function. A purposive activity that is ordinarily adaptive (functional under normal circumstances) can lead to disaster when the circumstances are abnormal.
For example, the homing of a male moth on a female, directed by the attractant secreted by the female, is ordinarily both purposive and functional. But it can lead the moth to his death when the attractant is placed on a surface covered with an insecticide.