All so-called duties to oneself, in so far as they deserve the name of duties, are indirect, and as such presuppose duties to other persons. With this finding we are safeguarded against the error of fallaciously extending the scope of our duties. But an opposite error is possible, that of fallaciously restricting the scope of our duties. If we designate a being in relation to whom we have duties as an "object of duties," we may say that only other persons can be objects of duties. In addition, we assert that all other persons, in so far as we affect them by our actions, are objects of duties for us. For every person, being a subject of interests, has rights, i.e., has a claim to respect of his interests under the law of equality of persons.
Failure to understand this leads to the above-mentioned false restriction of the scope of duties—a dangerous error far more prevalent in ethics than the opposite error of fallaciously extending the scope of duties. If we assume the existence of more duties than there actually are, we at least do not directly violate any duty; but if we assume that there are fewer duties than we actually have, we are directly led to such a violation. The proposition asserting that all persons can be objects of duties is therefore of greater practical significance than the proposition asserting the nonexistence of duties to oneself.
To recognize the full import of the former proposition, we must sharply distinguish between the concepts "subject of duties" and "subject of rights": for we cannot rule out a priori the possibility that some subjects of rights are not subjects of duties. Under the moral law, all beings who have interests are subjects of rights, while all those who in addition to having interests, are capable of grasping the demands of duty, are subjects of duties. Only rational beings are capable of such an understanding. Accordingly, we may classify all duties remaining after exclusion of duties to oneself into duties in relation to rational and to nonrational beings. If we designate as an animal a being that is a subject of rights, but is by its nature incapable of attaining rational self-determination, and as a man a being that is a subject of rights and at the same time potentially endowed with reason, we may state briefly that duty is always either to an animal or to a man. It is my contention that we have duties to animals, and that these duties are direct, i.e., that they are not derived from duties to men, or rational beings.
(Leonard Nelson, System of Ethics, trans. Norbert Guterman [New Haven: Yale University Press, 1956], 136-7 [italics in original] [first published in German in 1932])
Note from KBJ: Nelson's "subject of duties" is what philosophers now call a moral agent. The defining characteristic of moral agency is autonomy ("rational self-determination"). His "subject of rights" is what philosophers now call a moral patient. The defining characteristic of moral patiency is the possession of interests. To see the relation between the concepts, draw two partially overlapping circles. Let the left circle represent the class of moral agents and the right circle the class of moral patients. This creates four mutually exclusive and logically exhaustive categories: (1) things that are moral agents but not moral patients; (2) things that are both moral agents and moral patients; (3) things that are moral patients but not moral agents; and (4) things that are neither moral agents nor moral patients. Category 1 is empty. Normally functioning adult human beings are in category 2. Nonhuman animals (even apes) are in category 3. Plants, rocks, chairs, and other things are in category 4. Nelson says that we have duties to both animals and humans. Our duties to animals are grounded in their possession of interests. Our duties to humans are grounded either in their possession of interests or in their possession of autonomy. Already, by 1927, Nelson had laid the conceptual groundwork for two types of rights: interest-rights (which humans share with animals) and autonomy-rights (which only humans have).