Table of Contents
- Guidelines for a Theory of Ethics
- Primitive Ethical Systems
- Systems of Notable Ethical Philosophers
- Other Ethical Systems
- A Summary Evaluations
- End Notes
Guidelines for a Theory of Ethics
As described in my essay on naturalism in ethics, there are numerous theories of ethics that differ in significant ways and lead to different conclusions. Because of the is-ought problem, no system of ethics can be derived from the state of world as it actually is. There is also significant disagreement in metaethics, the branch of philosophy which asks what ethics is about and what it is supposed to achieve. If there is no way that we can agree on how to arrive at a universal ethics, can we at least agree on some minimum features that we would require any workable system of ethics to possess? It would be just as foolhardy to suspect that we could derive these features as it would be to expect an analytic system of ethics. Therefore, any set of features is in itself bound to be controversial. However, it is my hope that they would be less controversial than ethics itself. I present a possible list of these features below.
Before presenting them, however, I should mention there is at least one moral requirement that I consider necessary for any viable theory. Since this is a moral requirement, I suspect it may be more controversial than the others, but I believe that it is so important that if a system does not meet this requirement, it is not a viable system at all; this requirement is, so to speak, a “showstopper”. Specifically, any theory of ethics must respect the rights of individuals, in that it should not permit excessive harm to come to individuals as a means to an end. In other words, I believe that no purely consequentialist theory is adequate; there must be at least one deontological constraint in the theory. In essence, I believe that a mandatory axiom of any ethical system is that the ends do not justify the means. I will call theories that meet this criteria “protective”.
To further evaluate the many possible ethical systems that are protective, I propose the following additional seven criteria that I would want to see satisfied by any proposed system of ethics. I will call these the “ethical quality criteria”:
- Complete – The theory should provide a method for any rational agent to answer any ethical question regarding a specific situation or act, in particular:
- The theory should indicate whether any act is ethically permissible and/or or ethically optimal, as follows:
- An act is “ethically permissible” if and only if it is not ethically prohibited.
- An act is “ethically optimal” if and only if it is the best possible act from among all of the acts available to an agent in a particular situation.
- The theory should provide this type of answer for any ethical question: its scope must be complete. In particular, it must provide answers for questions involving all types of agents, including animals, plants, and the whole of nature.
- Note that this is a much weaker requirement than completeness in the sense of mathematical logic. Therefore, it is possible for a theory of ethics to be complete in this sense and still be consistent.
- The theory should not provide any contradictory answers. In particular, it should not lead to any moral dilemmas, such that in a particular situation it prohibits all of the available actions, as well as the option of doing nothing.
- Any two rational agents should obtain the same answers to any given ethical question.
- Any rational agent should arrive at the same answer to the same question under identical circumstances at different times.
- Rigorous – Ideally, the theory should be completely formal and rigorous. In particular:
- It should be expressed in the formal syntax of logic. By logic I mean first-order logic (predicate calculus), second-order logic (or set theory), and if necessary, modal logic.
- Its primitive terms should be clearly identified; i.e. primitive terms should be clearly delineated from defined terms.
- Any other terms should be defined explicitly from these primitives.
- Any principle statements not derivable from definitions alone should be listed as axioms in the formal syntax.
- All conclusions from the theory should be derivable as theorems from the definitions and/or axioms using the predicate calculus.
- The theory should not have any arbitrarily-defined thresholds. This is a big problem with many theories which state that a certain type of behavior may be allowed up to certain point. I want to prohibit these since there is no logical way to select the threshold values.
- The process to answer ethical questions should not depend on the intuitions or whims of any individual rational agent. Agents can be required to interpret empirical data, and to interpret possibly ambiguous maxims and definitions, but the logical step from these to an ethical decision should not involve any judgements. Note that, due to the inability to derive collective good and normativity from natural facts in a clear way, intuitions must be used to devise a theory of ethics. But the ethical system itself should not depend on intuitions.
- It should be clear what the primitive terms of the theory are, and they should all be unambiguous. Any reasonable person, given an object, ought to be able to indicated whether or not the object has a primitive attribute or relation, and all persons should be consistent in their assessments.
- The axioms of the theory should also be clearly delineated and unambiguous. It should also be clear how to use the theory to derive answers to ethical questions.
- The theory should not make specific identifications of any type of agent, group, or individual; e.g. it should apply equally to all species, all races, and both sexes.
- It should not contain an implicit but significant bias against any particular species, race, sex, group, or individual.
Regarding the requirement that a theory be non-arbitrary, there is another sense in which a theory can be arbitrary in its choice of primitives, axioms and definitions. Strictly speaking, these are always arbitrarily chosen. However, in a scientific theory the test of their truth is empirical. Due to the fact that collective ethical statements cannot be derived in a singular manner from naturalistic facts, every theory of ethics will be, in this sense, arbitrary. The criteria listed above represent a method for reducing this arbitrariness.
I will state at this point that I have not been able to develop a system of ethics that completely satisfies all eight of the criteria above, but I have developed a theory that satisfies every criteria except the requirements to be unambiguous and unbiased. My theory has some ambiguity because it assumes the ability to measure well-being in an agent, as well as interpersonal (and inter-species!) comparisons of well-being. It it also not completely unbiased, since it makes a distinction between three different types of agents. I refer the reader to the summary of my theory for future reference1.
Below I will review some significant historical and modern systems of ethics in order to evaluate their degree of compliance with the ethical quality criteria. It goes with out saying that these will be my subjective evaluations, and I invite scholars who are specialists with respect to each of these systems and philosophers to provide their own evaluations with respect to these criteria.
Primitive Ethical Systems
Let us start first with the Hobbesian “state of nature”2. One of the primary assumptions regarding the state of nature is that there is no moral integrity possessed by any of the agents in this state, and that furthermore, any cooperation between agents is extremely limited, perhaps to their immediate family. Note that this is a more primitive state than pure anarchy, since many anarchists and political libertarians such a Robert Nozick3 claim that in a state of pure anarchy, groups of agents would naturally form cooperative coalitions for mutual benefit. Clearly the state of nature is not protective; indeed this was Hobbes’s biggest motivation for wanting to leave that state. For this reason, it is not a viable ethical system. However, with respect to the other ethical quality criteria, it does better than one might expect, with its only other serious problem being its complete arbitrariness of results: any agent in the state of nature will take whatever action it deems appropriate to its interests.
One step above the state of nature is to devise an ethics based purely on individual intuitions. (I will refer to this as intuitionism, which is an abuse of terminology in that I mean an ethical system rather than a metaethical stance.) One might think that intuitionism is protective, in that the intuitions of some agents may lead them to realize that they should not sacrifice individuals for the common good. However, history shows that this is not the case, especially among those with the power to force these sacrifices. Intuitionism by this definition also fares rather poorly against the ethical quality criteria, most notably being inconsistent as well as arbitrary.
A third prevalent primitive ethics is theological voluntarism. This is, of course, a very broad, ambiguous term, so I will by way of example perform my analysis specifically on Christianity. I will assume that God does not provide direct ethical guidance (one cannot ask God what to do), therefore one needs to infer correct ethical choices from the Bible. Theological voluntarism in this case is notoriously non-protective. Individuals may be sacrificed at God’s whim to God’s ends. It also fares extremely poorly against the ethical quality criteria, having serious problems with every category except universality. In particular, it is incomplete, in that there are many modern ethical questions which are not anticipated in the bible, it contains many inconsistencies4, contains many arbitrary quantities5, and is clearly biased6. It is a sad irony that theological voluntarism is probably the system of ethics adopted by the majority of people on earth.
Systems of Notable Ethical Philosophers
Next let us consider two philosophers whose systems are primarily focused on the individual: Aristotle and Nietzsche, starting with the former. Aristotle’s system is not explicitly protective, but since it is not consequentialist in any way, it contains no driving force to exploit individuals as a means to an end. Therefore, I list it among the protective systems. With regard to the other ethical quality criteria, Aristotle’s ethics does extremely well in my opinion. Its only real weakness is with regard to ambiguity, in that it is not clear how to derive answers to moral questions in all cases from his principles; i.e., it is not clear how to apply Aristotle’s virtue ethics to difficult moral problems7.
Nietzche’s system is clearly not protective, as it offers the weak no protection from the powerful. His system does moderately well against the ethical quality criteria, although not as well as Aristotle’s. Nietzche’s main weakness is his theory’s aristocratic bias, which is implicit if not explicit (consider the parable of the lamb and birds of prey8), and certainly very significant.
Two additional philosophers who I would like to consider here are Hobbes and Kant. The essence of Hobbes’ system I consider to be political voluntarism deriving from the absolute power of the sovereign9. Hobbes’s system is clearly not protective, as the sovereign can ask individuals for sacrifice as he sees fit. The power of the sovereign also leads to the other main weakness of his system, which is the arbitrariness of the sovereign’s decisions.
Kant’s is a difficult system to evaluate because he focused so much of his effort on explaining why we ought to act morally rather than what constitutes correct moral action. The essence of his prescriptive theory I take to be derived from the categorical imperative10, with the assumption that as rational beings in any circumstance we must all derive the same conclusion if we reason correctly11. I will also assume that Kant is protective, as implied by his equal moral worth for all individuals12. Despite being protective, I find Kant’s theory to be extremely problematic when weighed against the ethical quality criteria. Indeed, because there is no guarantee that the universal laws willed by different agents will be the same (practical reason not withstanding), Kant’s theory is potentially inconsistent. Reliance on the human will is necessarily arbitrary. His system is ambiguous in that it is not clear how to apply the categorical imperative to particular ethical questions. Furthermore, it is not universal in that clearly it only applies to rational humans, and for the same reason it is biased, as animals clearly do not have “equal worth” to humans.
Finally, let us consider Rawls’s political philosophy in an ethical context. Much of Rawls’s work focused on the process for forming a just society. I will restrict my discussion to the characteristics of a just society, which Rawls’s concluded should be based on his two principles of justice13 involving liberty, opportunity, and distribution of resources. The latter is based on a variation of the maximin principle, in which a deviation from resource egalitarianism is justified to the extent that it increases the well-being of the least well-off. In his initial considerations of the principles of justice, Rawls had considered the difference principle14, in which these deviations are justified to the extent that they raise the well-being of all. It should be noted that the maximin principle will, assuming different well-being vs. resource curves15, actually deviate significantly from the resource egalitarianism that I believe Rawls thought a just society should approach, since the well-being of the least-well-off can always be raised by increasing their share of the resources. This would not be the case if he had stuck with the difference principle, which would have deviated moderately from resource egalitarianism.
One of the key features of the maximin principle is that it is implicitly protective, as it restricts the extent to which the well-being of the least well off can be allowed to deviate from the norm. With regard to the other ethical quality criteria, Rawls does well, suffering only from the fact that his system is clearly not universal: it only applies to communities of human beings. Indeed, the maximin principle would be impossible to extend to animals, in that presumably the vast majority of animals could never have well-being equal to that of a typical human. This is one area where the difference principle would not have helped, in that it would also be foolish for animals to share resources equally with humans.
Other Ethical Systems
Finally, let us consider four additional ethical systems that are relatively modern and promoted by various philosophers. The first to consider will be political libertarianism, by which I mean egoism with a constraint against harm and fraud. I will consider this constraint sufficient to rank libertarianism as a protective system, although clearly some individuals can be dominated by more capable persons16. I also believe that libertarianism suffers from ambiguity and bias. With a strict constraint against violence and fraud, libertarianism is not ambiguous. However, it is not clear that the constraint against violence is enough. Many claim that there should be a broader constraint against any kind of harm that is not simply an offense17. Separating harm from offense has proven historically problematic, yet traditional libertarianism has not included a prohibition against offenses. Libertarianism is also biased in that the constraint against violence applies only to humans. Indeed a pillar of modern political libertarianism is the right to bear arms, not the least reason being to hunt animals.
Next, let us consider egalitarianism. Egalitarianism is a consequentialist theory that states that our actions ought to maximize equality among agents in some way. There are two possible forms of egalitarianism:
- Resource Egalitarianism (RE): All Agents should receive the same quantity of resources.
- Well-Being Egalitarianism (WBE): All Agents should experience the same level of well-being.
Both are ridiculous in their pure form, in that a disproportionate amount of resources would be spent on animals and lesser living things. Therefore, I will assume that egalitarianism applies only to human beings, and that the theory says nothing about how to treat other types of agents. Even this may not be enough for WBE, as it may be impossible to raise the well-being of severely handicapped people to the level of the norm. RE may suffer from the additional problem that well-being is not strictly a function of resources.
Like Rawls’s system, both forms of egalitarianism are implicitly protective, as the resources or well-being of no individual may be lowered beyond the current minimum (although RE might not be protective in that it would allow non-resource-based harms). With respect to the ethical quality criteria, like Rawls, both versions of egalitarianism must be considered non-universal in that the principle cannot be extended to animals. For this reason one might also say that the theories are biased, in that they leave open the possibility of excessive harm to animals.
The last two systems to consider are utilitarianism and communitarianism. Both systems I will treat as purely consequentialist. Although Mill, as the most famous proponent of utilitarianism, is also known for his constraint against harm (as cited above), I will not consider this constraint in my evaluation of pure act utilitarianism. By communitarianism, I will mean a purely act-consequentialist theory that uses some measure of the collective well-being for a set of agents other than total well-being. This may not be the conception of this approach that is held by others, but communitarianism is still in the process of being fully developed. Therefore, I will take this approach for lack of a more precise conception of the theory. Also, I find a consequentialist communitarianism to be interesting in its own right.
By utilitarianism, I will mean act total utilitarianism. I choose total rather than average utility because of the propensity for the latter to lead to the conclusion that it would be best to slaughter all agents except a very small number (for synergistic effects; i.e. rather than one) of the most efficient at converting resources to well-being18. Before analyzing act utilitarianism with respect to protectivity and the ethical quality criteria, let us go on a small tangent and consider some common criticisms of the theory using a quantitative model. It would appear reasonable to assume that, aside from drops in well-being done by violence or personal betrayal, we can model well-being for an agent as a function of the agent’s rate of consumption of resources R. (For the rest of this discussion I will use the phrase “Net Positive Well-Being”, or “NPWB” rather than simple well-being. This concept takes into account discounted future well-being similar to Net Present Value in economics. For details see my essay on naturalism in ethics.)
It would seem that an agent requires some minimal level of resources to survive; i.e., at some finite positive resource rate the NPWB of the agent would drop to 0. It would also seem that the law of diminishing returns would apply, such that the second derivative of NPWB with respect to R is always less than zero. Let us, for the sake of exploration, assume an idealized case in which the agent’s NPWB function can be represented by NPWB = ln(R). Note that this function satisfies both of the requirements above, in that ln(1) = 0, and d2NPWB/dR 2 = -1/R 2 .
One criticism of utilitarianism that has been raised by Derek Parfit and John Rawls is that utilitarianism leads to an infinite population 19 . However, this does not make sense in the face of finite resources and the type of NPWB curve described above. In fact, if we assume for simplicity that all agents have roughly the same NPWB vs. R curve, then it can be shown that total NPWB is maximized when all agents have the same rate of resource consumption, and that rate is an optimal rate found at the point where NPWB(R)/R = dNPWB(R)/dR20. Thus, in this simplistic case, utilitarianism leads to egalitarianism of both resources and NPWB. Furthermore, it leads to an optimal population level. Even if we adopted the more realistic model in which different agents have different NPWB vs. R curves, as long as they both exhibited the two features described above, total utilitarianism would still lead to some optimal finite population, rather than an infinite population.
However, it is fairly easy to see that in the more realistic model, not only would utilitarianism not lead to egalitarianism, but assuming for the moment that the population was fixed, it would allocate more resources to the agents with the more efficient NPWB vs. R curve. Consider two agents a1 and a2 with the following NPWB curves: NPWB1 = ln(R1) and NPWB2 = 2*ln(2*R2), and a finite total R = 1.5e. In this case the optimal resources levels are R1 = e/2 and R2 = e, with NPWB of 0.3 and 3.4 respectively. Thus the least efficient agent a1 would get fewer resources than a2, and by view of it’s lower efficiency, experience less than 1/10 the NPWB of a2.
The situation gets even worse if we allow for the pure utilitarian fact that populations can be manipulated (by any means!). Consider a population of agents like a1 and a2, ignoring for the moment any benefits of diversity. Then utilitarianism would say to exterminate all of the a1’s and adjust the population of a2’s to the optimal level described above. The consequences of this conclusion are disturbing. It says that utilitarianism implies speciesism, and recommends a type of eugenics in which the population is bread to maximize NPWB efficiency. Furthermore, if there is a species that is more efficient than humans, then we would be morally bound to exterminate all species except this one (including as many humans as possible!), and divert all resources to this species with its population at the optimal level21.
All of these issues with utilitarianism derive from the fact that it is not protective; strictly speaking, utilitarianism does not explicitly prohibit any means to maximizing NPWB. However, with respect to the ethical quality critiria, it does extraordinarily well. Indeed, it at least partially satisfies each of them.
Finally, let us consider communitiarianism. Strictly speaking, communitarianism as an ethical system is in its infancy, and the details of most proposed systems still need to be worked out. However, I would like to discuss a specific form a communitarianism, one that is strictly consequentialist, and that differs from utilitarianism only in it’s measure of collective NPWB. Thus, while utilitarianism says that the NPWB of a set of agents NPWBset = ∑ NPWB(ai) for all agents ai in the set, communitarianism would calculate NPWBset by some other means. In particular, I would envision some sort of sociological objective criteria in which a number of factors would combine to produce a measure of collective well-being, in the same way that scientific objective criteria would be used to measure well-being of an individual.
Such a measure could eliminate many of the problems with utilitarianism described above. For example, a measure of collective well-being that valued diversity would eliminate the need to divert all resources to a single species and/or to “improve” that species through breeding or eugenics. However, in its purely consequentialist form, communitarianism could not be guaranteed to be protective. I have therefore assume it to be not protective, until such time as the theory, and in particular, the measure of collective well-being, has been more fully developed. Because of this lack of development, I also consider this form of communitarianism to be ambiguous, the only one of the ethical quality criteria from which it suffers.
A Summary Evaluation
The following tables present a summary of how each system discussed above fares against the ethical quality criteria. For each criterion and system, I have indicate either that “Yes”, the system meets the criterion, “No” the system does not meet the criterion, or that the system achieves a “Partial” satisfaction of the criterion. Systems are presented in roughly best to worst order in terms of how they fared against the criteria. Because of the importance of protectiveness, I have divided them into two tables, the first containing only protective theories, and the second containing those that, in their pure form, are not protective (and therefore not reasonable). I have included the evaluation in the first table my proposed system of ethics, so that its relative merit can be compared against historical systems.
|State of Nature||Yes||Partial||Partial||No||Partial||Partial|
As can be seen, my proposed system fares best against these criteria (since it was designed to do so), with Aristotle’s close behind. Aside from their lack of protection for individuals, Utilitarianism and Communitarianism also fared well. Conceivably any of these systems could be modified to improve their standing with respect to these criteria.
- See my essay on A Theory of Ethics. This version is not fully rigorous, but contains a link to the fully rigorous and formal exposition of the theory.
- Hobbes, Thomas, Leviathan, Chapter 13.
- Nozick, Robert, Anarchy, State, and Utopia, Basic Books, New York (1974), Part I.
- See, for instance Merritt, Jim, “A List of Biblical Contradictions”, The Secular Web, 1992, URL=<http://www.infidels.org/library/modern/jim_meritt/bible-contradictions.html>.
- For example, the maximum term for a man sold into slavery was six years (see for example Exodus 21:7).
- E.g., Genesis 1:28 gives humans dominion over nature.
- I will use the term “ethical” when discussing situations involving any type of value, whereas I will use “moral” to discuss situations involving ethical obligations of one rational agent to another. Thus if an act is moral it is also ethical, but there can be questions of ethics that do not fall under morality. This follows the usage adopted by Stephen Darwall in Philosophical Ethics, Westview Press, Boulder, CO (1998). See in particular, Part Four: Philosophical Ethics Without Morality (pg. 175).
- Nietzsche, Friedrich, The Genealogy of Morals, First Essay, Section 13.
- Hobbes, op. cit., Chapter 18.
- Kant, Immanuel, Groundwork of the Metaphysics of Morals, Chapter 1, pg. 402.
- Ibid, Chapter 2, pg. 412.
- Ibid, Chapter 2, pg. 433. This is Kant’s famously statement that individuals should treat each other “never merely as a means, but always at the same time as an end in himself.”
- Rawls, John, A Theory of Justice, Revised Edition, Harvard University Press (1999), pg. 266.
- Ibid, pg. 65.
- This is a very reasonable assumption. For example, people differ widely in their levels of health.
- Of course, this was the principal claim of Marx. For a modern example, see Toyama, Kentaro, “Income Inequality Around the World Is a Failure of Capitalism”, The Atlantic, May 13, 2011, URL=<http://www.theatlantic.com/business/archive/2011/05/income-inequality-around-the-world-is-a-failure-of-capitalism/238837/>.
- E.g., Mill, John Stuart, On Liberty, Oxford University Press, pg. 14.
- See Parfit, Derek, Reasons and Persons, Oxford University Press, Oxford (1984), pg. 368 and also pg. 420.
- See Ibid, pg. 388 and Rawls, op. cit., pg. 153.
- It can be demonstrated that in a homogenous population with the characteristics that I have described, total NPWB will be maximized via egalitarianism with a finite population. I will first shown that, given my assumptions, utilitarianism leads to egalitarianism. In particular, assume that there is a fixed population of N Agents all with the same NPWB vs. R curve. Since this curve is the same for all Agents, Resource Egalitarianism in this case will be equivalent to Well-Being Egalitarianism. Furthermore, assume that this curve is such that for all R, d2NPWB/dR2 < 0. Then it can be shown that any deviation from egalitarianism will lower total NPWB. Let ε be an amount of the resource stream that is taken from one Agent and given to another. Since d2NPWB/dR2 < 0, dNWPB(R+ε) < -dNPWB(R-ε), so dNPWB Total = dNWPB(R+ε) + dNPWB(R-ε) < 0. In other words, shifting resources from egalitarianism will always lower total NPWB. Next, I will illustrate how total NPWB is maximized under these conditions at a finite population. Let N be the number of individuals (the independent variable in this case), and let RT be the total resource rate such that each individual receives resources at the rate of R = RT/N. For the purposes of this illustration, let NPWB = ln(R) for each Agent. Note that this function has the two properties that 1) R > 0 when NPWB(R) = 0, and 2) d2NPWB/dR2 < 0 for all R. Then total NPWB, or NPWBT = N*ln(RT/N). dNWBT/dN = ln(RT/N) – 1, which equals 0 when N = RT/e. Since d2NPWB/dR2 < 0, this point will be a maximum. Thus, total NPWB is maximized by a population of N agents, where N = RT/e (for this NPWB vs. resource curve). I conjecture that for any such curve with the two characteristics mentioned above, this maximum will always occur where NPWB(R)/R = dNPWB(R)/dR. This is true for NPWB = ln(R), since at N = RT/e (and thus R = RT/N = e), NPWB(R)/R = 1/e. Likewise, dNPWB(R)/dR = 1/R = 1/e. Thus, indeed NPWB/R = dNPWB/dR in this case.
- This is similar to, but an even more extreme conclusion than that of Nozick’s “Utility Monsters”; see Nozick, op. cit. pg. 41.