Moral Naturalism – The target of the Open Question Argument – seeks to identify moral properties with natural (worldly) properties. It falls under the Realist camp and is Cognitivist, so it says that our express independently existing moral facts objective of our opinions. Contrast this to the Expressivists we have just looked at who say that there are no moral facts, our moral talk is merely our approval or disapproval of some action.
You may remember that another big challenge to Moral Naturalism is Mackie’s Error Theory (outlined in various philosophy posts on this site – just knock it into the search box over there). Before moving on to outlining variations of the Moral Naturalist theory, I will first consider a few ways Moral Naturalism can respond to Error Theory.
Responding to Error Theory
Argument from Disagreement
Error Theory would say that the fact that rational people disagree about the nature of morality shows us that there can be no objective moral properties in the world to which we can identify moral properties with. People from different cultures can hold equally morally worthwhile and rational views and end up with different conclusions and recommended courses of action, or moral codes.
>We could say that moral disagreement is only apparent. This doesn’t seem to grasp the gravity of the disagreement however. Consider the Eskimo example (Dr. Pekka’s lecture slides):
Eskimos may think it right to kill off the old and sick, unlike us, but what if the alternative was that both the old and the young perish? Don’t we accept sacrificing some to save everyone?
We could answer this by saying that people have the same basic moral principles, but that they draw ultimately different conclusions from those principles. For example, killing off the old in Eskimo society is very similar to the ‘greatest good for the greatest number’ principle. If the Eskimos did not kill off the older citizens, everyone would die from lack of food. This could be said to be indicative of both survival instinct and a sociological moral code.
Furthermore, people’s own prejudices, beliefs, ideologies and other things like religion can have huge influences on our moral decision-making. If our moral choices are persuaded otherwise through these sort of external factors, it no surprise that ‘moral’ conflicts arise. Things like Just Wars in the name of religion and party uprisings like that of Nazi Germany are good examples of ‘morality’ gone haywire.
Disagreement also shows that morality may be circumstantially-based. Euthanasia may be wrong in this situation, but not in that situation, for example. Trying to assign strict moral codes like the Categorical Imperative and Utilitarianism will thus fail because they do not fully account for the individual circumstances of various situations.
Finally, just because we can’t all agree, it does not necessarily mean that there are no moral facts. Think about the times in history when many people mistakenly believed (and thus everyone did not agree) on the shape of the world. Pretty much all people now agree that it is round, but still not everyone. Does that change the fact that the world is round?
Argument from Supervenience
Supervenience means to ‘depend upon’ something. For example, if A supervenes on B, it means that, if there is B, there must be an A. Note however that, just because we have A, does not mean we have B. It only works one-way, unless we make sure to specify that both A and B supervene on one-another mutually.
The argument from Error Theory says that it seems strange how we can posit that moral properties depend on material (physical/natural) properties in the world. How can we explain the relationship between natural properties, which are discoverable by the sciences, and moral properties, which seem so different and unlike in nature to natural properties? Just as colour could be said to supervene (depend on) the underlying physical structure of an object, can moral properties really be said to supervene on the structure of our minds, or brains (or something else)?
The Naturalist can appeal to other natural but likewise mysterious supervening properties in order to fight back. The solubility of a sugar cube, for example, seems rather strange. We would say that the property ‘being soluble’ is inherent in the very chemical composition of the sugar cube, but find it quite tricky to actually explain how the property of ‘solubility’ actually depends on the chemical structure. That is, if the chemical structure of the sugar cube was changed, it may no longer be soluble (and probably no longer a sugar cube) and hence, the super cube would no longer have the property of solubility. But grasping exactly what the relationship is between these two things is difficult. How can we actually explain the link between solubility and chemical structure? Similarly, just because it is difficult to explain the link between moral properties and natural properties, does not mean that there is no link.
There is of course, much more to this debate than I have outlined here ^^.
Argument from Queerness
Mackie asks us to think about what moral facts really are. If they are to retain the sort of usability that we have long sought from them, he asserts that they would have to be a uniquely strange and ‘queer’ sort of property. Furthermore, he says that we would need a special faculty, not just simples ones like the 5 senses, to detect them. (We cannot see moral properties, smell, touch, taste or hear them, for example. So how do we ‘know’ them?)
The Naturalist will want to argue that actually, moral facts are natural facts, just like those natural facts that are in the world that we observe and detect with the senses. Also, we can know these moral facts through the usual methods, such as by using science. Clearly this seems like a dangerous strategy for the Naturalist to take, so how do they explain themselves in saying such seemingly implausible things?
First, recall the distinction between Synthetic and Analytic Moral Naturalists (Dr. Pekka’s lecture slides):
Synthetic moral naturalists believe that moral terms cannot be defined in non-moral natural terms, but deny (contra OQA) that this semantic thesis implies the ontological thesis that moral properties are non-natural; they can still be natural.
Non-reductive naturalists believe that moral properties fall into the category of natural properties in their own right and are not reducible to other natural properties.
Reductive naturalists believe that moral properties are reducible to other natural properties (e.g., some combinations of physical, psychological, and social properties).
Analytic moral naturalists believe that moral terms can, after all, be defined in non-moral natural terms. The few contemporary analytic naturalists have more sophisticated views of what analysis, sameness of sense, etc., are than OQA assumes.
Remember the Open Question Argument? And the Hesperus-Phosphorus/Water-H2O objections put forward by Kripke? Basically, in the OQA, we are making a big mistake in trying to identify two properties only semantically (such as saying water is H2O). What we need to consider is that Sense and Reference can mean different things; that we can have different senses that points to the same reference (the object itself). So analyticity is not the same as necessity; just because two properties are necessarily the same thing in real life, it does not follow that analytically (semantically) they have the same meaning.
Hence, ‘good’ may actually be synonymous with ‘pleasure’, but we just have not found the reference to prove this yet. This could be reductive realism, like Railton, Boyd would argue for.
Likewise, consider a non-reductive stance like that of Sturgeon, Brink. ‘Constitution is not the same as identity’. An example might illustrate this best:
Even if a pigeon is just a certain arrangement of particles, is could still remain a pigeon if we rearranged some of the particles. For example, taking away its tail feathers probably would stop us calling it a pigeon. A lame pigeon perhaps, but still a pigeon.
Similarly for morality: morals can be multiply realisable. Just because one certain arrangement of events is classed as ‘good’, it does not mean that another, quite different arrangement of events cannot also be classed as ‘good’.
But that’s to skip ahead of ourselves. Let’s take a look at the theory of Railton:
Railton’s Reductive Naturalism
Note that Railton’s brand of Realism can be distinguished from Cornell Realism. To solve the problem of reducing moral properties to natural properties, Railton thinks that we should provide ‘reforming definitions‘, rather than try and provide explanations that capture both the actual nature of some object and the meaning of our ordinary day-to-day language. This aims to respond to the OQA by saying that it’s okay that our definitions don’t give proper synonymy; what is important is that we get the facts about what moral properties are correct.
There are two stages to Railton’s argument (from Dr. Pekka’s notes):
1) First, give an account of the non-moral value of a person’s good: what it is for something to be good for a person, in their self-interest, in the interest of their well-being.
2) Second, extend this to an account of moral rightness.
Then, we follow the Inference to the Best Explanation to arrive at an explanation for ‘good’:
1) First, posit (provisionally) the existence of objective facts about what is good for a person.
2) Then, show how the existence of such facts would better explain certain relevant phenomena than any alternative.
(a) How satisfying an individual finds his or her life and choices
(b) Evolution of a person’s desires and behaviour, and thereby, how learning about one’s good takes place.
(c) Certain “truisms” that surround our talk about what’s good for a person, such as that individuals tend to know their good better than third parties, and tend to get better at knowing their good over time.
Okay. Imagine yourself. You have desires – things you want to have and do. Some of these you may ‘know’ are good for you, but to some of them you are completely oblivious. You, try as you may, do not always know what is good for you.
Now imagine a perfect, objective version of yourself, knows is omniscient and knows everything is good for you (in a non-moral sense) as a matter of fact. This is a reforming definition of ‘good for a person’; the objective version of you would know exactly what the real, completely human and fallible you, wants. As Railton says:
X is non-morally good for A if A would want her actual self to want X, were she fully informed and rational (i.e., were she A+), and contemplating the position of her actual self as someone about to assume it. (FE 190-2)
This seems logical. There do seem to be ‘matter of fact’ things that we as human beings need, but that we may be unaware of. How satisfying we find our lives will depend on how much our choices and desires correspond with what is actually, matter-of-fact good for us. Furthermore, we learn about what is good for us through trial and error. We learn that eating a balanced diet makes us feel good, whereas eating too much sugar or fat will ultimately make us weary and unhealthy. This is not an intuitive fact; it is learned. Also, what makes one person feel good (non-morally) may not be the same for another person. Consider a healthy person and a person with diabetes – the person with diabetes may not need the exact same diet as the healthy person in order to feel good.
So to use this logic in finding moral rightness, just apply the same strategy:
Consider moral rightness as being the assessment of what is best for all people considered equally from an impartial standpoint. In Railton’s own words:
“[T]he notion of social rationality by considering what would be rationally approved of were the interests of all potentially affected individuals counted equally under the circumstances of full and vivid information. Because of the assumption of full and vivid information, the interests in question will be objective interests.
…The discontent produced by departures from social rationality may produce feedback that, at a social level, promotes the development of norms that better approximate social rationality.”
So, what we see Railton doing is advocating an account of ‘rightness’ or ‘goodness’ that says that what is morally good is considering things from an impartial and ‘social’ point of view. In Dr. Pekka’s words then:
X is morally right iff (if and only if) X maximally promotes the intrinsic non-moral good of individuals, as considered from the social point of view.
Railton also says that, where two (or more) groups disagree about what they desire or want, there will still be an objective fact that is best for both groups. It is just that finding out exactly what this is difficult. So social unrest, for example, can be accounted for when members of a particular group have had their needs ignored or overlooked. Their dissatisfaction and exerting pressure (on the government – etc) is testimony to the objective fact that they have been morally wronged. Consider the protests of slaves or Civil Wars as an example. The solution, to meet the needs of everybody, is difficult, but the uprising demonstrates that a breach of moral fact has been made.
Further to this, we can postulate that certain social forms go out of existence when they do not adequately meet the needs of a given society. This could be thought of in terms of evolution. Consider capital punishment as an example. Not all societies have outlawed capital punishment (but the usage has been significantly reduced over generations), suggesting that it is not yet known whether or not keeping it would promote the needs of human beings best. Perhaps it will forever remain undecided, because it has both positive and negative affects on society.
Hence, we can suppose that moral progress not be constant and flawless, but that it will eventually triumph, else we as human beings would likely perish through corrupt societies. Railton’s theory can also account for disagreement. People act selfishly, for sure, but they also recognise the need for cooperation and that their own desires can infringe on the desires of others. So mutual tolerance seems attractive. However, there may be situations where no single set of circumstances suits everybody (consider religious ideals and ideologies that commit certain people to a particular way of life). Nevertheless, one could argue that there will still be a fundamental moral fact that is common to all human beings underneath the surface disagreement. For example, even though different religions disagree about God, existence of God and exactly which God to follow, one may infer that everybody has a fundamental interest in matters of life and death and real and unreal entities.
We begin the next section by considering science. More exactly, how do we come to ‘know’ things in science? Many, such as Boyd, argue that science can be considered as Coherentist. To put it simply, consider the example of fixing a boat, one plank at a time:
Neurath’s ship metaphor: just like you can fix a ship at sea one plank at a time, relying on other planks to keep it floating, we can fix our belief system one belief at a time, provisionally relying on other beliefs that may themselves come into question.
So, although our scientific theories may not be completely verifiable by experience, they mutually support one-another. That is, in testing a new hypothesis, one must rely on other supporting hypothesis or auxiliary assumptions, such as unobservable objects or unproven theories. The realm of quantum physics seems most apt to raise as an example here – much experimental hypothesis, but little actually proven. The background theory which must be fixed in order to test new hypotheses can itself be called into question. (Evolution, for example, is still just a theory, though it has much weight and support to it). So, one might argue, the inferential gap between ‘is’ and ‘ought’ in ethics could be likened to this.
Consider physical facts that have been discovered by science or psychological facts that have been discovered by psychology. We have to rely on various scientific and psychological theories in order to derive these ‘facts’ about the world (which could later be disproven). In the same way then, why not say that moral facts must rely on moral theory. I.e. the moral facts are dependent on the theory of ethics. If they are dependent on a theory, such as Railton’s, then they may be discoverable a posteriori (empirically).
Consider the example from Dr. Pekka’s lecture slides:
1) Surprising thesis: Hitler was a morally admirable person.
2) Modest principle: No morally admirable person would instigate and oversee the death of millions of persons.
3) A testable consequence: Hitler didn’t do so.
4) Fact: But Hitler did do so.
Here, we can come to the conclusion empirically that Hitler did in fact oversee the deaths of millions of people, discrediting the testimony of other people who say he did not. So we can arrive at the conclusion empirically.
But it’s not all plain sailing. Consider some criticisms of the Science-Ethics debate again from Dr. Pekka:
>> Scientific objectivity is possible because of the fundamental role of
observations in science. What plays, in moral reasoning, the role played by observation in science?
>> Moral intuitions don’t seem to play this role. They are diverse between people and seem to reflect the particular moral outlooks they already accept, or their culture or upbringing.
>> Forms of factual inquiry seem to be getting closer to truth in relative independence from cultural distortions. If moral reasoning is about objective moral facts, then what explains our lack of progress in ethics and the persistent cultural variability in moral beliefs?
>> Ethics contains hard cases about which no agreement can be
achieved. Science can solve its hard cases. Doesn’t this show an
important difference in kind between ethics and science?
How to go about defending the analogy between Ethics and Science? Boyd goes along the old lines of saying that Moral Realism holds the best explanation that accounts for our picture of morality. He goes into something known as ‘Homeostatic Consequentialism‘.
Here’s another chunk of Dr. Pekka’s explanations about what Boyd says:
>> Actions, policies, character traits, etc., are morally good to the extent to which they tend to foster the realization of goods that satisfy important human needs or to develop and sustain the homeostatic mechanisms which unify them. (FE 175)
>> These goods include love and friendship, engaging in co-operative efforts, physical recreation, intellectual appreciation and expression, autonomy and liberty, happiness.
>> These things are homeostatically clustered: they co-occur contingently in nature because the presence of each tends to favour that of the others. They causally interrelate in ways that foster and sustain each others’ presence both directly and via certain psychological and social mechanisms.
>> This causal structure of properties is what moral goodness is.
It may fail “open question” tests, but that’s OK.
This seems to fit with the aforementioned ‘evolutionary theory’ for moral goodness. It is no accident that we recognise and respect the needs of fellow human beings to maintain a stable and harmonious environment – “good things attract other good things”, as my Philosophy partner Jack comments. Just like in real life where natural properties cluster together, the same could be said of moral properties. For example, a animal that has the property of being very fast, will usually also have the property of being agile. Likewise for moral properties, a person who is helpful will also usually be trustworthy. So our moral judgements are relatively reliable indicators of necessary human goods and thus indicative of the truth of moral goodness; our moral ‘intuitions’ or judgements can be said to be steeped in reliable tradition that promotes a harmonious society. Without ‘clustering’ or homeostasis of our moral properties, our society would not have evolved in the way it has.
What about moral reasoning? In science, observation is largely how we determine results and prove or disprove hypotheses. But in morality?
Boyd says that, “if goodness is a homeostatically clustered causal structure, then observations should play the same role in the study of goodness as they play in the study of all the other natural properties.” (Dr. Pekka)
That is, HC is considered to be true, then the ways in which our society has developed over time moving towards a higher standard (i.e. THE goodness) allows us to call different things good at different points in time. In other words, while we may have called capital punishment ‘good’ at one time, the fact that it is now considered bad in many countries does not mean that we were wrong about what ‘goodness’ is, only that we were (and still are) in the process of moving towards a more harmonious and synergistic society. We have observed change. In the light of new evidence about how societies best function, we have discovered (a posteriori) more about the truth of goodness.
Naturalism and Motivation
Let’s have another look at that problem of motivation that I was considering in an the Moral Judgement Internalism vs. Moral Judgement Externalism article
Naturalists like Boyd and Railton can accept that judgements about a person’s good are necessarily action-guiding:
>> What is good for a person “must have a connection with what he would find in some degree compelling or attractive, at least if he were rationally aware.” Anything else would “be an intolerably alienated conception of someone’s good.” (Railton, “Facts and Values,” 47)
However, as I discussed, MJI is more suited to the Expressivist account of morality and MJE more suited to the Naturalist account of morality. For Railton and Boyd to say what they do supposes that demands they take an MJI view of morality. But perhaps not. They could say that promoting the good of persons in society is among one’s personal aims and is also necessary for promoting one’s personal aims. Hence, morality would be contingent upon the larger aims for society, and be an MJE theory.
Furthermore, Railton and Boyd argue that psychologically ‘normal’ people care about morality and goodness. So they are motivated (contingently) to do what is right. Boyd goes further and argues that psychologically abnormal people suffer from a cognitive defect that inhibits their judgement in assessing human goods and harms. This is because, he argues, the cognitive capacity relies on sympathy – being able to imagine ourselves in the shoes of others. Without this capacity, we would not be able to consider moral goods and harms impartially and hence not be motivated in the right way to promote society and the needs of others in addition to ourselves.
Harman’s Challenge to Cornell Realism
Gilbert Harman issues a challenge to the Cornell Realists (the proponents of Cornell Realism), who can be considered as Non-reductivists, different from Railton whom we have just looked at. (From here I will be working closely from ‘An Introduction to Contemporary Metaethics’ by Alexander Miller).
Cornell Realism hold that moral properties are supervenient on (or are multiply realised by) non-oral properties. In contrast, reductivists like Railton argue that moral terms are analysable in terms of non-moral terms and as such, claim that moral properties are identical to non-moral properties as a matter of synthetic fact. Both views can be considered forms of Naturalism.
Non-reductivists (the Cornell Realists) argue that any one moral property (goodness, for example) can be realised in an indefinite number of non-moral properties. But they claim that for all good actions, there is no one non-moral property that all situations have in common to which goodness can be reduced. The Reductivists like Railton, in contrast, argue that there is a common natural (non-moral) property to all good actions.
Moving onto Harman then, he argues against the Cornell Realist claim that we are justified in postulating the existence or moral properties because they are required in our best overall explanation and picture of the world around us, are wrong. Harman says that moral facts can never have this sort of explanatory role and so says that moral facts and properties cannot be reduced to non-moral facts and properties.
I have left the task of Harman’s Challenge and Sturgeon’s reply to my fellow student Jack to muddle through. What follows is his explanations of both positions, with minor corrections and additions from me. Thanks Jack!
Harman describes a scenario in which we come across some young hooligans setting fire to a cat. Our reaction to the scenario would be that what we are witnessing is wrong. Railton would say that the wrongness of the cat being harmed is the best explanation for our judgement that the action IS wrong. But Harman says “Whoa! No way, man!”. He attributes our diagnosis of “wrongness” to our own internal processes: our aversion to seeing animals hurt, or the fact that we were raised to believe that such things are wrong.
For Harman, moral observations don’t provide evidence of moral facts, and moral facts can’t cause anything. Harman then describes a scientific instance of observation: a scientist sees a vapour trail in a cloud chamber. He knows that protons leave a vapour trail and so comes to the conclusion that the vapour trail he sees denotes the presence of a proton. For Harman, there is no equivalence between this and purported instances of judgements based on moral observations. He says that physical facts earn their “explanatory keep” in a way that moral facts do not.
(My addition:) For example, in the cat scenario, Harman holds that our explanation of why the hooligans are wrong does not need to refer to moral properties to explain why the action is wrong. All we need to do is to explain the non-moral facts about the situation; the pouring of petrol on the cat, causing the cat extreme pain (etc) and non-moral facts about ourselves (being brought up to believe mindless violence is wrong). So we suffer no explanatory loss in describing the situation purely in terms of non-moral properties, and thus, we do not need to posit the existence of moral properties (in holding that the inference to the best explanation assumes that the simplest explanation is the best one).
Here’s Jack again to outline Sturgeon for us:
Sturgeon replies to Harman on behalf of the Realists arguing that moral facts do explain our beliefs. He uses a counterfactual argument (counterfactual arguments show that an assumption is irrelevant to the explanation of a fact, if that fact would still obtain even if our assumption were false). For example:
>> Hitler did the terrible things he did because he was morally depraved.
>> If Hitler were NOT morally depraved, would he have still done the things he did?
Well, the bad traits that Hitler had constituted his depravity, his racism, his megalomania, his cruelty, etc. So could Hitler have committed the atrocities that he did without those traits? No. So it seems that Hitler’s moral depravity IS the explanation for what happened.
In the case of the cat, Sturgeon would have us imagine a scenario in which hurting the cat was NOT wrong and to decide whether or not we STILL thought it was wrong. For this, we would have to envision a situation in which there was no cruelty or pain, so we substitute a toy cat. Do we still observe wrongness? No. So this indicates that the wrongness of what we saw happening to the cat WAS the best explanation for our judging there to be wrongness.
Harman’s counter reply is couched in terms of epiphenomenology. The theory of epiphenomenology holds that all mental states are caused by underlying brain states. Harman would say that Hitler, for example, had non-moral traits in his brain (such as racism, megalomania, cruelty – etc.) that caused his actions, and made it the case that his actions were evil. Hitler’s ‘being’ evil explains nothing.
A Realist response to this could make use of the concept of the Program Explanation, developed by Jackson and Pettit. This is a way of looking at cause and effect and identifying first and second order causes of events. For example, if you have boiling water in a glass container, the container will break. What causes this? The obvious answer to this is “the high temperature of the water”. This is the Program Explanation.
A more accurate reply could be that the glass breaks when a water molecule collides with a glass molecule. That is called a Process Explanation. In this scenario, the molecule clash is obviously the direct cause of the breakage, whereas the high temperature is not actually responsible. Consider again the cat scenario: You come across some hooligans torturing a cat. Your response to this is that it’s wrong. The PROCESS explanation, for your response is that the pouring of gasoline on the cat and setting it alight is what caused you to reach your moral judgement. This does not involve moral observation but rather practical observation and this practical observation causes you to form a moral judgement. This is in line with Harman’s thinking.
But, program explanations are first order descriptions that entail second order descriptions. The high temperatures (program) lead to molecule clashes (process). So we can say things like:
>>If the molecule that caused the breakage hadn’t caused it, another one would have.
>> People rose up against apartheid because of its injustice.
The latter is a program explanation that identifies the moral property INJUSTICE as the best explanation for an event. A process explanation for the same event would cite people in Soweto being beaten and blacks being denied education, leading to the overthrow of apartheid. This would lead to an Anti-Realist interpretation attributing the downfall of apartheid to these specific non-moral things. But, if the beatings had occurred elsewhere in a place other than Soweto, and DIFFERENT instances of injustice had led to the revolt, the PROGRAM explanation would still hold, leaving the moral property INJUSTICE as the cause of an event.
In the cat case:
>> The program explanation of the cat case involves judging that the wrongness of the hooligan’s action is the best explanation for your judgement. The process explanation would be that the pouring of the gasoline and setting the cat on fire caused your reaction. But if the hooligans had done something else, say electrocuting the cat, the process explanation would be different. But the program explanation would remain the same. This leads to the conclusion that program explanations are more reliable and better predictors than process explanations.
So where does that leave us? My feeling is that Sturgeon has adequately rebutted Harman’s carping and quibbles. The use of counterfactuals as well as the way the Program Explanation gets at what is REALLY behind our judgements (discernment of wrongness, goodness, cruelty or whatever) makes it seem probable that we DO react to moral properties, and the fact that we react to them is a good indicator that they are there.
‘An Introduction to Contemporary Metaethics’ by Alexander Miller (2003)
The lecture notes of Dr. Pekka Vayrynen at the University of Leeds