Posted by: dacalu | 18 June 2012

What “Should” Should Mean

My friend Matthew and I have been arguing about moral realism (the idea that values exist independent of the people holding them and that they may be discerned).  After writing this long response to his last email, I thought I would share it with you.  It seems to be a recurring theme, and the older I get, the more important I feel it is.  (Thanks, Matthew, as always for a brilliant provocation.)

Should/Ought has three meanings:

1) Statistical Expectation:  “A fair coin flipped 1000 times should come up heads roughly 500 times.”

In this case, “should” means “I expect” without any sense of proper outcome, only predicted outcome.  The manner of prediction turns out to be surprisingly complicated philosophically precisely because it can presume you unquestionably know what outcome to expect.  That doesn’t relate to my central argument, but if you’re interested, check out the footnote (***).

1A) Strategic Expectation: If you want to have a 50% chance of heads, you should get a fair coin.

It presumes a preference (wanting a 50% chance) and states the methodology which will probably (or in some cases definitely) provide that outcome.  Note this is a conditional statement which presumes a preference.

2) Universal Preference:  “One should not kill innocents.”

The universal “should” presumes some universal preference for a given outcome.  It’s an if/then statement with both if and then.  It’s a statement of strategic expectation where the initial preference statement has been assumed, but not stated.  Thus, the statement above really means:

“If one wants to be right with God, then one will not kill innocents.”
“If one wants to avoid Hell, then one will not kill innocents.”
“If one wants to manifest good karma, then one will not kill innocents.”

It’s always important to identify the “if” for yourself, even when you don’t say it out loud.  I want to point out the common error of using an initial preference that a listener does not agree with.  Unless you’re talking to yourself, you need to assess the preferences of the listener before using the word “should” in this sense.

2A)  Common Preference:  “You should stop for the red light.”

We live within communities of common preference.  We fall into the habit of using should so easily because most of the time we can assume that those around us have the same priorities.  The statement above usually means:

“If you wants to obey the law (and all Americans want to obey the law, snort), then you will stop for the red light.”

Note the rhetorical purpose of the statement is almost always to convince someone to act in accord with a preference you assume they already have.  You assume they have it because all people of group X have it.

3)  Personal Desire:  “You should go out with me.”

Starting in the Enlightenment, Westerners began to be skeptical of anything that could not be assessed empirically.  Wittgenstein and others argued that “You should do X” has no logical content beyond “I don’t want you to do X.”  (Mind you Wittgenstein later changed his mind.)  A strand of thinkers became particularly concerned with asserting that they did not have the same preferences as the religious should-ers.  Once again, unless you are talking to yourself, this must be assessed logically as:

“I want you to go out with me.”
“If you want to please me, then you will go out with me.”
(or possibly “If you want to agree with me, then you will go out with me.”)

The trouble comes when you realize how weak such an argument tends to be rhetorically.  I really want to encourage people to figure out what the “if” part of the statement contains.  Often we use meaning 3, but want the moral weight of meaning 2 attached to the logical necessity of meaning 1.

So, when speaking philosophically, I try to use “expect” for meaning 1 and “desire” for meaning 3, reserving “should” and “ought for meaning 2.  When dealing with moral reasoning and the classical Is/Ought gap (a la Kant), we mean universal or at least common preference.  Rhetorically, meaning 2 has more weight the more common the preference is.  A statement of universal preference (“God wills it” or “That’s just the way it is”) commands the most respect, but also requires the most justification.

Almost always the point in contention has to do with meaning 2, and yet so often we get into debates where one person is using a meaning 1 argument (cap and trade should reduce CO2 emissions) against a meaning 2 objection (we should care more about humans than the “environment”).  That debate will never be resolved because the two people are actually talking about different things.  In this case, and many others in the religion / science dialogue, the meaning 1 argument is a slam dunk, but the people making the claim have radically difference meaning 2 expectations (i.e., no, you can’t care for humans without caring for the environment OR we have to care about polar bears and countless other species as well as humans).

Even harder to resolve, some debates involve one person using meaning 2 (you should not kill) which the other person insists on interpreting as meaning 3 (you don’t want me to kill).  I understand the latter position, even sympathize with the desire for individual freedom and moral humility; however one issue stops me from taking their position seriously.  Anti-should-ers almost invariably think I should (meaning 2) agree with them.

I don’t see what the point of discussion is when no common preferences are assumed.  Why would I be talking with you unless we had a common preference for communication?

Interpreting terms like “should,” “ought,” “moral,” and “ethical” with meaning 3 seems both pointless and counter-productive.  It’s perfectly reasonable to believe that no common moral truths exist.  It is not reasonable, though to argue about them if they don’t.  You basically have to drop out of the rational discussion just as “physicists” who deny a common definition of mass can’t really participate.  One can put a giant red flag on the (ontological) question of whether common preferences exist but having done that, one must act as though they did when one argues [if one wants a stronger rhetorical position than “please please me”].  It’s enough to appeal to common preference, without a universal preference, but it must go beyond personal desire.

In summary, I don’t feel compelled to be attentive and comprehensible (much less reasonable) about this unless we agree to a common preference for listening and clear-speaking.  It has to be more than a personal desire – it has to be a priority of the community before actual communication can occur.

To be honest, I don’t completely trust my moral intuition.  Just as I do not trust myself to do biology or physics by myself*, I do not trust myself to do morality by myself – nor am I particularly impressed by others who try.  I think everyone should (meaning 2) take a hard look at their preferences and desires.  If we want to do this well, we should (meaning 1) do it in groups and use a common language (meaning 2**).


*I use “by myself” to mean without consulting the expertise of others in writing, terminology, and methodology.  I do not insist that, having been educated by the tradition, one could not – in principle – do science alone.  I still think science in community is better science, necessarily.  To be explicit, it might on the short term produce better results, but over the long term, communal reasoning works better.

** I am as much a Bayesian with regards to morals as I am with regard to statistics.  I think there are real, objective, external morals and I also think humans cannot have complete certainty about them, only high confidence, so I am (ontologically) a fan of universal preferences, but (methodologically) a user of common preferences.  I defend the high confidence I have in a given proposition by claiming I have evidence supporting it’s absolute truth.

***It turns out to be far more complicated than this, as two competing philosophies exist.  I go into detail for those who want evolutionary biology and other empirical approaches to bridge the gap between expectation and preference (below).  If you’re not interested in philosophy of probability, skip down to “1A).”

Frequentists think that probability is, strictly speaking, only the frequency of an event occurring within a historical sequence OR the ideal number of times a thing will happen given an eternal perspective.  Frequentists want probability to be objective, but events don’t seem to occur ideally, so there’s always a slight disconnect between the desire for a perfect perspective and the accidents of history – what we call “contingency.”  We still say that the expected number of heads is 500 despite the fact that the ideal probability of a perfectly fair coin landing heads exactly 500 times out of 1000 is only 2.5%.  That should be 0.5^1000 x (1000! / (500! x 500!)).  We call 500 the expected value, because the actual value “should” be higher as often as lower.

Bayesians think probability reflects nothing more nor less than the subjective confidence of the speaker.  They believe one can meaningfully constrain that confidence by acknowledging prior assumptions and updating them with the frequency of certain observations.  Say I start very confident in the existence of unicorns.  I have an initial “prior” of 99%.  If I “update” my prior every year and never see a unicorn in 37 years, the prior steadily decreases.  Bayesians don’t really do 100% or 0% because you never have absolute knowledge, only relative confidence.  So, back of the envelope, my final prior should be less than 1 in 10^74 (that’s 0.99 x 0.01^37).  Bayesian subjectivity does not mean arbitrary confidence.  It can be very well informed by the data but it always maintains a recognition that your prior effects the final probability, if only minimally.

Mind you, Bayesians tend to be realists.  Most think there is a real (frequentist) distribution of events and probabilities in the world.  The subjective nature of probability comes from human inability to have perfect access to that distribution.



  1. Thanks, Lucas, for this thorough response to my “provocation.” From my current understanding of your position, I don’t find anything overtly disagreeable. Here is my perspective on the points you raised above:

    Strictly speaking, Lucas and I have separate ontological positions. Lucas has high confidence that universal preferences exist and that people can discern them with incomplete certainty. Currently, I am ontologically agnostic on the question of whether universal preferences or objective moral truths exist; however, epistemologically I am highly skeptical that anyone can discern or demonstrate them.

    Furthermore, the existence of universal preferences or values, for me, seems counter-intuitive. The universe or nature, as a whole, appears to lack preferences. People don’t usually say that matter, space, and energy want X. Preferences, as far as I can tell, appear to be unique to living beings, particularly those with sentience. Living beings are a part of nature, but not the whole, so even if a certain preference were common to all living beings (e.g. survival), it would not be universal in the sense that I think Lucas is describing.

    Lucas is astute in pointing out that people often want the authority of meaning 2 universal preferences. One approach is to appeal to an omnipotent deity who “commands” what moral truths should be. From my perspective, however, this argument is a house of cards, since I do not think that there is any good reason to believe that an omnipotent deity exists. Furthermore, I think that logical problems like the Euthyphro dilemma (does god love what is good, because it is good, or is what is good good, because god loves it?) create many problems for a theistic objective morality. I’m less skeptical of the idea that objective moral truths can be intrinsic (as Lucas put it, “that’s just the way it is”), but, again, I don’t think that anyone can prove them, and usually these arguments seem to be nothing more than meaning 3 personal desires seeking the authority of meaning 2 universal preferences.

    As I said previously, Lucas and I clearly differ in our ontological positions. However, my goal in posting this comment is not to refute or debate Lucas’ ontological position, since I believe that methodologically we can agree on some common preferences, even if we come from different perspectives.

    I think that Lucas was very fair and reasonable in acknowledging that the meaning 2A common preferences can be a persuasive and authoritative use of “should.” Lucas is correct that there are different communities of common preference (e.g. religious, social, etc.). However, I believe that at least one preference is common to all human beings: happiness. Like Aristotle, I believe that happiness is the chief goal (preference) from which all other goals derive. Since I believe that the preference for happiness is common to all human beings, it is about the closest one can come to a “universal” preference.

    I also think that Lucas was very astute in pointing out that much of the confusion about “should,” comes from the fact that we often forget or assume the “if” conditional. When I make moral arguments, if you were to boil out the details, my most basic formula is: “If you want happiness (and all humans want happiness), you should not do X.” I do believe that this is a logically consistent position. However, I do not think, by any means, that this argument is conclusive. It raises many other questions: What constitutes happiness? How do we achieve happiness? What if some things make you happy, but other things make me happy?

    I think that these are important questions and more than I can answer here. However, I do believe that it establishes a logical foundation, from which a Utilitarian system of morality can be derived. Within this system, one can authoritatively say that one “should” not do X, because it violates the common preference for happiness.

    Often when I argue with Christian apologists about issues of morality, they assert that an atheist (really an ontological physicalist) cannot have an objective basis for making moral claims. I think that this criticism is due to a misunderstanding about the “if” conditional. For example, when I state, “We should not discriminate against homosexuals,” the apologist assumes that I am making the claim based on a universal preference: “if you want to do what is objectively right, you should not discriminate against homosexuals.” They then assert that there cannot be objective morals without a god. As I discussed above, I believe that there are many problems with the latter statement; however, more importantly, they have confused the fact that I was actually making the claim on the basis of a common preference: “If you want happiness (and all humans want happiness), you should not discriminate against homosexuals.”

    Obviously, more can be said on these issues and I don’t pretend in the slightest to have offered a complete moral theory above. However, I think that we can agree that a person can use the word “should” and make an authoritative claim, without appealing to a universal preference or objective morality. As Lucas said, “It’s enough to appeal to common preference, without a universal preference, but it must go beyond personal desire.” I think that atheists can quite logically make appeals to common preference, and thus, apologists who try to shut down their moral arguments, by assuming that atheists are making appeals to universal preference, are committing a straw man.

    Nevertheless, there remains much work to be done on fleshing out what our common preferences are. I am still developing my moral philosophy, so I don’t claim to have an extensive answer at this time. However, I do think that this discussion has been very valuable, for me at least, in distinguishing between the meaning 2 universal preferences and the 2A common preferences. I thank you Lucas for this insight! I think we can agree that you and I have a common preference to better understand these issues together, even if we come from different ontological perspectives.


    • Thanks, Matt.
      I think that calling happiness the common human preference is tautological. It doesn’t add anything. Why should there be a common preference at all? How do you know what it is? How do you know it’s common? I feel like those are presumptions that need to be defended as much as calling God’s will the common human preference (as Aquinas does).


      • Hey Lucas,

        I think that asking “why” about a common preference could have two meanings: 1) Why “should” there be a common preference? 2) Why “is” there a common preference?

        I arrived to this conclusion inductively. From what I can tell of other people’s behavior (both individually, socially, and historically), everyone appears to be striving for what I would call happiness. I already said in the post that I have more work ahead of me to flesh out a definition for happiness.

        However, here are my answers to the two alternate readings of the question: 1) The common preference of happiness simply appears to be intrinsic to humans, so there is no “should,” it just “is.” 2) I’m not entirely sure how this common preference came to be, but I would stake confidence in an evolutionary natural cause. How do we bridge the “Is/should” gap? I would say that happiness is by nature a common preference, so it jus “is,” but the things that achieve happiness are strategic choices, so they belong to the meaning 1 “should.”

        The existence of a deity is not arrived at inductively, so I wouldn’t compare the two. One presumes a deity. I arrived at my conclusion that humans commonly want happiness a posteriori by observing human behavior.

        I am confident that no deity exists. I am likewise confident that there is a solid basis for morality derived from the common human preference of happiness, which I observe inductively and infer is derived from nature. Fleshing out a solid secular moral theory to put the pieces together is the task in front of me.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: