Gewirth's argument for human rights

Alan Gewirth's argument for human rights is made from the perspective (or from 'within the standpoint') of any agent A. The argument says the agent must accept in turn:

(1) I am an agent
(2) I have purposes
(3) In order to achieve my purposes in general I require freedom and well-being (or whatever the necessary conditions of achieving my purposes in general are)
(4) I have rights to freedom and well-being (= Others ought not to interfere with my freedom and well-being)
(5) 'I am an agent' is a sufficient reason for 'I have rights to freedom and well-being'
(6) Every agent has rights to freedom and well-being (= Others ought not to interfere with its freedom and well-being)
(7) I ought not to interfere with the freedom and well-being of any other agent

The 'ought' in (4) is an 'indirectly prudential ought'. A does not say that others ought not to interfere for the sake of achieving their own purposes (as in a normal prudential ought), but that they ought not to interfere for the sake of A achieving A's purposes. By contrast the 'ought' of (6) is a moral ought. So Gewirth in effect analyses moral oughts as 'generalised indirectly prudential oughts'.

The key questions are: what does (4) mean, and how does the argument from (1) to (4) work? In my 2008 critique of Gewirth's argument ('Protagonist and Subject in Gewirth's Argument for Human Rights') I concluded that A's accepting (4) must be interpreted as A's demanding, on the grounds that it needs freedom and well-being in order to act for any of its purposes, that others not interfere with its freedom and well-being. I then argued that on this interpretation (6) does not follow from (5). For (6) would only follow from (5) if the argument from (1) to (4) could be generalised to an argument showing that A had to conclude that any agent  B had rights to freedom and well-being. But such a generalised version of the argument would be invalid.

The argument from (1) to (4) is essentially that since A cares about achieving its own purposes it is rationally compelled to demand that others not interfere with the freedom and well-being it requires in order to achieve those purposes. So the argument is essentially prudential: it goes from A's purposes to what A must do to achieve them. Therefore its validity depends on the person whose purposes, freedom and well-being are at stake (the 'subject' of the argument) being the very same person as  person for whom the argument is supposed to be rationally compelling (the 'protagonist' of the argument). A must be both protagonist and subject of the argument for it to work.

By contrast a generalised version of the argument would have to go like this. For any agent B, A must accept:

(9) B is an agent
(10) B has purposes
(11) In order to achieve those purposes B requires freedom and well-being
(12) B has rights to freedom and well-being (= Others ought not to interfere with B's freedom and well-being)

In this argument while the protagonist is still A the subject is B. This argument fails because A does not necessarily care about B achieving its purposes and so is not rationally compelled to demand that others (including A) not interfere with the freedom and well-being B requires in order to achieve them, and that is what A's accepting (12) would amount to.

Of course we can generalise (1) to (4) to a valid argument showing that any agent B has to accept that B has rights to freedom and well-being. Here protagonist and subject would again be the same person. But that is not the kind of generalisation needed.  (In Gewirth's terminology, it is 'judgmental' universalisation, whereas what is needed is 'possessive' universalisation.)

In short, the argument from (1) to (4) is 'identity-dependent'; its validity depends on the fact that its protagonist is the same person as its subject. Therefore it cannot be generalised to an argument with A as protagonist and other agents as subjects. Therefore (5) cannot be generalised to (6). We can concede that in virtue of the argument from (1) to (4) A must accept (5), as long as it means:

(5a) 'I am an agent' is a sufficient reason for me to have to demand that others not interfere with my freedom and well-being

But A can accept this without having to accept (6), i.e. without having to demand that others not interfere with the freedom and well-being of any agent. For the argument on which (5a) is grounded is identity-dependent: it only works for the case in which its subject is A, the protagonist of the argument.

Beyleveld's and Bos's 2009 response ('The Foundational Role of the Principle of Instrumental Reason in Gewirth's Argument for the Principle of Generic Consistency') is that if the argument from (1) to (4) was essentially prudential then it would indeed be impossible to get from (5) to (6). But the argument is not essentially prudential. Rather it is 'dialectical'. For A is rationally compelled to accept (4) by virtue of accepting nothing more than (1), i.e. on pain of contradicting that A is an agent, and not as a means of achieving its ends. There is a prudential element to the argument from (1) to (4) but this is subsidiary. Accordingly the move from (5) to (6) is after all valid.

(NB Beyleveld and Bos's use of 'dialectical' seems to differ from Gewirth's original use. Gewirth called an argument dialectical if it proceeded from 'within the standpoint' of a particular agent, i.e. if it relied on that particular agent's assumptions and operated via inferences which that particular agent was rationally committed to drawing. In this broad sense of 'dialectical' the argument from (1) to (4) as I have interpreted it is dialectical.)

My response is that I do not see how the mere acknowledgement 'I am an agent' can rationally compel A to demand that others not interfere with its freedom and well-being via some interpretation of the argument from (1) to (4) without the argument relying on the assumption that the protagonist of the argument cares about purposes of its subject, and therefore without the argument remaining identity-dependent. Yet as long as the argument from (1) to (4) is identity-dependent then the move from (5) to (6) is illegitimate. The question is then whether Beyleveld and Bos's own restatement of the argument from (1) to (4) provides an interpretation of it that is not identity-dependent. If it does then we are in new territory. But I am not persuaded that it does.