Follow up to: The Meaning of Morality
Continuing my quest started in “The Meaning of Morality”, I seek to understand what is going on in the frequent use of moral language and the establishing of moral fact — if such a thing exists.
Morality specifically, and normativity most generally, seems to hinge on a lot of words, but most frequently words like “good” and “ought”. What do these words mean? Why is something morally good, or just plain good? And what ought we do? Is there a difference between what we ought to do and what we morally ought to do? If so, which ought we actually do?
Thus the problem I seek to solve is to make sense of normativity, and provide some answer to the nagging question of what I ought to do. In order to start, I seek to find the meaning of “good” and “ought”, and their related linguistic cousins. Bear with me, and it should all make sense.
The Linguistics of Value
To gain further insight on the meaning of the word “good”, I turn us to Judith Jarvis Thompson’s book “Normativity”. Here, she argues for a rethinking of the word good — that things aren’t good period — but rather good for something specific.
For example, we might again consider a good hammer. Here, “good hammer” means the hammer is good at fulfilling the purpose of the hammer, or rather is effective at pushing nails into other surfaces. A good tennis player is good at playing tennis, or likely to win tennis matches. A good baseball is one that can be used to play such a game, a good police officer is one that can effectively catch criminals, a good senator is one that well represents his or her constituents, etc.
In all these cases, “good” is an adjective that describes the noun as meeting a certain evaluative standard by which that noun is judged. Whenever we call something “good”, we have an evaluative standard in mind, or we wouldn’t know whether or not that something was good after all. Likewise, there can be many different evaluative standards for different objects — what makes a good tennis player is entirely different from what makes a good tennis ball, which is again entirely different from what makes a good hammer or good police officer.
Not only does this principle underly “bad” and “good”, but also what is meant by “better” and “worse”. These too take standards — a cup can be better at holding liquid and worse as a paperweight than another cup all at the same time. Likewise, if a cup is better than another cup without further elaboration or context, we can assume an implied standard for comparison — the purpose for which the cup was designed.
So far, all these different concepts of goodness/badness/beterness/worseness have been distinctly non-moral, or have nothing to do with morality. But we can easily account for moral goodness as well using this same analysis — consider moral goodness as yet another standard by which we can judge someone or something.
A good police officer might be good at catching criminals, but a morally good police officer is something entirely different, because the standard we’re looking at changes. Now, we’re thinking in context of a moral standard — refraining from killing innocent bystanders, immunity to corruption, etc. A morally good police officer might not even be that good at catching criminals. Likewise, people and things can be compared based on well they adhere to these moral standards, being morally better and morally worse.
Thus, we could propose that value terms like “good”/”bad” and “better/worse” are comparative adjectives, similar to “shinier”/”duller” or “bigger”/”smaller” — just as “shinier”/”duller” compares objects based on their luster and “bigger”/”smaller” compares objects on their size, “good”/”bad” and “better/worse” compares objects based on how well they meet their purpose or the contextually implied standard.
The Linguistics of Ought
Now we take a small break to go onto the other side of things, to oughtness, or normativity. What is meant by the word “ought”?
Let’s look at a couple different sentences that make use of the word “ought”:
: Young children ought to get at least eight hours of sleep.
: All people ought to refrain from murder.
: The “Bishop” chess piece ought be moved diagonally.
: The mailman ought to be here any moment now.
Multiple Senses of Ought
The use of the word “ought” in  through  is uncontroversial. However, each of these three uses is different semantically:  gives practical advice to young children (or their caregivers),  states a moral command,  states the rules of the game Chess, and  gives an assessment of what is expected of the future. It’s clear that neither , ,or  are intended as a practical suggestion; neither , , or  are intended as moral commands; neither , , or  are intended as stating the rules of a convention; and neither , , or  give a simple assessment of what is likely to occur.
So what gives? How is the same word used in three different senses? Stephen Finlay presents a theory in his paper “Of Oughts and Ends” that , , and  are all subtypes of  — or rather, that all statements about ought relate to an assessment of probability, with just a little modification:
[1*]: In order that young children stay healthy, young children ought to get at least eight hours of sleep.
[2*]: In order that people comply with moral standards, all people ought to refrain from murder.
[3*]: In order that one plays Chess, the “Bishop” chess piece ought be moved diagonally.
: The mailman ought to be here any moment now.
The Probabilistic Ought
In , the “ought” statement follows probabilistically — it makes a statement that the mailman is likely to arrive soon. However, with the “in order that” modifier, [1*] – [3*] now follow probabilistically as well, despite also maintaining their various unique semantic senses. Indeed, getting eight hours of sleep probably makes a young child healthier than he or she would be otherwise, refraining from murder is the only way to comply with moral standards, and moving the chess piece diagonally is the only way to play Chess (since any other movement would be a different game).
We also can use other words — like “can”/”can’t”, “could”/”couldn’t”, “might”/”might not”, “may”/”may not”, , “should”/”shouldn’t”, “will”/”won’t”, , and “must”/”mustn’t” to modulate the certainty of the probabilistic relationship. Just as saying the mailman should be here soon seems to indicate more doubt, and saying the mailman may be here soon indicates yet more doubt, saying the mailman must be here soon indicates more certainty, bordering on absolute certainty.
The levels of probability work out semantically somewhat like this, although this is a bit loose:
- The mailman can be here any moment now = it is possible that the mailman arrives soon
- The mailman could be here any moment now = it is possible that the mailman arrives soon
- The mailman might be here any moment now = the mailman is marginally likely to arrive soon
- The mailman may be here any moment now = the mailman is somewhat likely to arrive soon
- The mailman should be here any moment now = the mailman is more likely than not to arrive soon
- The mailman ought be here any moment now = the mailman is more likely than not to arrive soon
- The mailman will be here any moment now = the mailman is very likely to arrive soon
- The mailman must be here any moment now = the mailman is almost certainly going to arrive soon
Based on the probabilistic understanding of these relationships, we can rephrase  –  as follows:
[1**]: In order that young children stay healthy, young children should get at least eight hours of sleep.
[2**]: In order that people comply with moral standards, all people must refrain from murder.
[3**]: In order that one plays Chess, the “Bishop” chess piece must be moved diagonally.
: The mailman ought to be here any moment now.
This new understanding of “ought” gives us a unified definition of “ought” for all its semantic purposes: [A] ought be the case if [A] is likely to occur. “In order that [something], it ought be the case that [A] then means “[something] is more likely if [A] is the case”. Note here the level of likelihood (somewhat likely, more likely than not, or almost certainly) is not a part of the proposed definition because that level of detail is unnecessary.
Given this definition, we can then rephrase  –  finally as such:
[1***]: Young children are more likely to be healthy if they get at least eight hours of sleep.
[2***]: Moral standards will be broken if anyone murders someone else.
[3***]: Chess cannot be played unless the “Bishop” chess piece is moved diagonally.
[4***]: It is likely that the mailman will arrive.
If this view is correct, [1***] maintains the same meaning as it did back in . Same goes with the other three statements.
But how do we know that “ought” really is about probabilistic relationships about specific standards? For this, I suggest three more sentences:
: It ought to have rained today, but it did not.
: The Senator thought that in order to get re-elected, she ought to have held a fundraised in her home state, but she lost anyway.
: The clock ought to chime at each hour.
Here,  shows that something that was likely to happen actually did not happen — the conditions made it likely to rain, yet surprisingly it didn’t rain! Any robust definition of “ought” would have to account for the clear probabilistic meaning of .
 steps up  by linking it to a standard — that which would get a Senator re-elected. In this sentence, “holding a fundraised in her home state” is what would likely meet this standard — an action linked to a standard in a probabilistic fashion. Yet, it turns out that this relationship didn’t actually hold in this instance.
 is like , except the standard is contextually implied — the assumption that the clock in question is functioning and the fact that all functioning clocks chime at each hour add up that the clock will likely chime at each hour.
In , , and  there is no moral question, rule, or practical advice. “Ought” is used exclusively in its clear probabilistic sense.
I have now proposed a definition of “good” (and its cousins “bad”, “better”, and “worse”) as an adjective that descibres how well things measure up to a standard.
I also have proposed a definition of “ought” (and its cousins “can”, “could”, “might”, “may”, “should”, “will”, “must” and their related negatves) as a modal auxilary verb that expresses a likelihood of something being the case, including the likelihood of something meeting a standard. Both of these linguistic views also neatly account for distinctly moral goodness and moral commands as another standard to compare or express the likelihood of meeting.
This view of “good” and “ought” is called end-relational theory, because it proposes that “good” and “ought” both relate things to ends, or standards of comparison. In the next essay, we’ll look more in depth at how this theory can capture categorically normative language.
Followed up in: Categorical Ought as Rhetorical Ought
Author’s Note: I rewrote a portion of this essay on July 2 to make more sense with my current positions on meta-ethics.
I now blog at EverydayUtilitarian.com. I hope you'll join me at my new blog! This page has been left as an archive.