Reference

2. Three arguments against description theories of proper names

Description theories dominated for half a century until challenged by three arguments around 1970: the unwanted necessity and rigidity arguments, both due largely to Kripke (1980), and the argument from ignorance and error, due to Kripke and to Donnellan (1972).

Unwanted necessities were one of the obvious problems for the classical description theory. If ‘the last great philosopher of antiquity’ is synonymous with ‘Aristotle’, then ‘Aristotle is a philosopher’ should be necessarily true (provided Aristotle exists). Yet it is not: Aristotle might have died young, long before his philosophical fulfilment. The cluster theory avoided this version of the problem. The description ‘the last great philosopher of antiquity’ is just one among many in the cluster that expresses the meaning of ‘Aristotle’. Aristotle need not have any particular one of the many properties specified by the cluster. The cluster theory does require, however, that Aristotle have most of the properties specified by the cluster. Kripke points out how implausible this is. Aristotle might not have had any of the properties commonly associated with him: he might not have been a pupil of Plato, taught Alexander the Great, and so on. So ‘Aristotle’ cannot be synonymous with the cluster of associated descriptions (see Kripke, S.A. §2).

The rigidity argument deploys the notion of ‘rigid designation’. This is explained as follows: for a term a to be a rigid designator is for it to designate the same object in every possible world (in which it designates at all), or, less picturesquely, for it to be such that ‘a is F’ would truly characterize some non-actual situation if and only if the object that the term actually designates were F in that situation. Kripke argues that names are rigid designators whereas the descriptions alleged to be synonymous with them are not. So description theories are wrong. Consider:

Suppose that Aristotle had indeed died young. Then Plato, not Aristotle, would have been the last great philosopher of antiquity. In those circumstances the truth of (4) would depend on whether Plato was fond of dogs. But the truth of (3) would still depend, just as it does depend in the actual world, on whether Aristotle was fond of dogs. The name ‘Aristotle’ designates Aristotle in a non-actual situation just as it does in the actual situation, whereas the description ‘the last great philosopher of antiquity’ designates whoever is the last great philosopher of antiquity in that situation, whether Aristotle or not. So the name is not synonymous with the description. Similarly, it is not synonymous with any other description, or cluster of descriptions, that is a candidate to give its meaning. (Note that in assessing rigidity we are evaluating the truth and reference of our expressions with the meanings that they actually have as a result of our usage, but we consider them as used to characterize hypothetical situations. Of course, any expression could have a different meaning as a result of different usage in a non-actual situation – language is ‘arbitrary’ – but that is beside the point) (see Proper names §§2–3).

Statements (3) and (4) are not modal statements (although we have been evaluating them as characterizations of non-actual situations). Other versions of the rigidity argument concern modal statements. For example, whereas ‘Hesperus is necessarily Hesperus’ is true, ‘Hesperus is necessarily the star that rises in the evening’ is not: had the solar system been differently arranged, Hesperus might not have been visible in the evening but it still would have been Hesperus. This sort of difference between descriptions and names in modal statements had been emphasized earlier by other philosophers, particularly Ruth Barcan Marcus (1961).

Some philosophers, notably Michael Dummett (1973), resisted Kripke’s two arguments by focusing on modal statements. These philosophers exploited the well-known ambiguities of scope in these statements to undermine the apparent difference between names and descriptions. Whatever the truth of this matter, the apparent difference in nonmodal statements remains.

These two arguments challenge the description theory as a theory of the meaning of a name, a meaning that determines the name’s reference. This is how the theory is naturally understood. However, as Kripke points out, the theory could be understood as simply a theory of reference: the reference of a name is fixed by a description, but the name is not synonymous with that description. This weaker theory is impervious to the two arguments. Of course, the weak theory has a defect: because it is no longer a theory of meaning, it no longer solves the problems that troubled the Millian theory. Indeed, the relation between meaning and reference becomes a pressing issue on the basis of this theory.

There is another way of saving the description theory from the two arguments while avoiding this defect. Instead of weakening the original theory into a mere theory of reference, we revise it along the following lines: a name is synonymous with a ‘rigidified’ description. Our language already seems to have descriptions that contain ‘rigidity operators’; for example, the italicized part of ‘the person who, in the actual world, is the last great philosopher of antiquity’ seems to make this description designate Aristotle in every possible world. If descriptions of this sort are indeed rigid, the revised theory claims that a name is synonymous with such a description. If such descriptions are not rigid, the revised theory can claim that the name itself supplies the rigidity operator and so would be synonymous with an ordinary nonrigid description governed by that operator.

All of these description theories – original, weak and revised – have the consequence that the users of a name associate with it a description that identifies its bearer. The third argument against description theories, the argument from ignorance and error, challenges this. So, if the argument is good, it counts against all description theories.

The argument shows that people who seem perfectly able to designate with a name are very often too ignorant to supply an identifying description. Thus some may fail with the name ‘Cicero’ because they associate with it only the description ‘a famous Roman orator’, which applies to many people. Others may fail because they associate ‘the man who denounced Catiline’ with ‘Cicero’ and are unable to supply an appropriate description for ‘Catiline’: the description that they associate with ‘Catiline’ is ‘the man denounced by Cicero’, which takes us in a circle and leaves both names without reference, according to the description theory.

The argument shows also that people often associate with a name a description that identifies something other than the name’s bearer: they are simply wrong about the bearer. Thus some associate ‘the inventor of the atomic bomb’ with ‘Einstein’ and some associate ‘the first person to realize that the world was round’ with ‘Columbus’. Almost everyone who has heard of Peano associates ‘the discoverer of Peano’s axioms’ with ‘Peano’, but the axioms were actually discovered by Dedekind. Despite such errors, people succeed in designating Einstein, Columbus and Peano by their names.

The description theory can be improved by allowing people to ‘borrow’ their reference from others. So the description Martha associates with ‘Einstein’ might be ‘the person Joe referred to yesterday as “Einstein”’. Provided Joe can supply an appropriate description – either one that describes Einstein directly or one that borrows reference from someone who can supply an appropriate description – Martha will succeed in designating Einstein. There is a danger of a circle, of course. Apart from that, there are problems of ignorance and error once more. Perhaps Martha cannot remember the reference lender; or she can remember the lender by his name, ‘Joe’, but cannot supply the identifying description that the theory requires; or the lender is identified but he cannot identify Einstein, perhaps identifying something else instead. The description theory still seems to place too great an epistemic burden on speakers.

An argument from ignorance and error can also be brought against another, more general, theory that some – for example, Dummett – have taken from Frege. This is the theory that to understand a name a person must be able to ‘identify’ its bearer. This ability is usually evidenced by providing a description, but it may be evidenced by ‘recognizing’ the bearer. The epistemic burden that this more general theory places on speakers still seems too great.

Various moves have been made to save the description theory in the face of these difficulties. Most popular, perhaps, have been theories that the reference of, for example, ‘Einstein’ is determined by a description along the lines of ‘the person referred to by (called, named and so on) “Einstein”’, for this description does identify Einstein and speakers surely associate it with the name. However, such theories still risk circularity.