Thursday, August 23, 2018

Moral Ought Facts are Non-Natural


In my third Maverick Christian Vlog episode I refer to a scholarly paper which is called A Folk Semantics Argument for Moral Non-Naturalism. In this blog entry I’ll provide some of the technical background so that those of us who aren’t analytic philosophers can better understand it.

Why is it important that morality is non-natural? One reason is that it reveals that there is more to reality beyond the natural, physical world. Another reason is that morality being non-natural makes it so that atheism doesn’t fit in very well with the existence of morality, especially objective morality for reasons I explain in my third vlog episode. In contrast, the existence of an objective and non-natural morality makes perfect sense in a theistic worldview.

Next I’ll explain some philosophy lingo before explaining the math used in the paper.

Philosophical Terminology

Moral semantics is about how to define moral terms. In philosophy, the word “folk” refers to colloquial stuff that laypersons use; e.g. “folk psychology” is (an albeit derogatory) term for beliefs about the human mind that ordinary people accept. In the paper, “folk semantics” with respect to morality refers to what most ordinary people mean when they use terms like “morally wrong.”

A stipulative definition assigns a meaning to a particular word or phrase to be used in a given context (as a philosophy paper). For example, in a philosophy paper one might give a stipulative definition of “fully justified” by saying, “I will say that a belief is fully justified to denote the belief being justified to the point where one can rationally say one knows it to be true.” Stipulative definitions are often used for conveniently assigning a label to some concept and won’t necessarily match the lexical (“dictionary”) definition.

A hypothetical imperative takes the form of something like, “If you want to do X, you should do Y” and describes what to do as a matter of practical necessity to accomplish some goal. For example, “If you want to do well in school, you ought to study” meaning something like, “As a matter necessity, you need to study to do well in school.” The sort of ought used in hypothetical imperatives is called a hypothetical ought.

A category mistake (or category error) is attributing a characteristic to something that it can’t possibly have because it’s not of the right category; e.g. saying that the number six has mass or volume, when the category of abstract objects is such that they can’t have mass or volume.

Set Theory

Some Basics

Sets are collections of stuff where order and duplicates are irrelevant. For example, the followings sets are all identical.

{1, 2, 3, 4}
{1, 2, 2, 3, 4}
{4, 3, 2, 1}

There’s the empty set, sometimes symbolized as {} which is a set that has no members.

To illustrate some set operations, suppose our “universe” consists entirely of natural numbers 1 through 9. Now let A, B, and C be the following:
A = {1, 5, 9}
B = {1, 5, 7, 8}
C = {2, 3}

(element of)
1 ∈ AFor any set S, x ∈ S means that x is an element of S.

(not an element of)
1 ∉ CFor any set S, x ∉ S means that x is not an element of S.

A ∩ B = {1, 5}Given sets S and T, S ∩ T contains all the elements x such that x ∈ S and x ∈ T.

A ∪ B = {1, 5, 7, 8, 9}Given sets S and T, S ∪ T contains all the elements x such that x ∈ S or x ∈ T.

{1, 5} ⊆ BGiven sets S and T, S is a subset of T if and only if each member of S is also a member of T.

(not a subset)
{2, 9} ⊄ BGiven sets S and T, S is not a subset of T if and only if it is not the case that S ⊆ T.

The set “All x such that x > 3” can be symbolized like this:

{ x | x > 3 }

The set “All x ∈ A such that x > 3” can be symbolized as:

{ x ∈ A | x > 3 }

That set described above would be {5, 9}.


Unlike sets were order and duplicates don’t matter, they do matter in tuples. The following are all different from each other:
(1, 2, 3, 4)
(1, 2, 2, 3, 4)
(4, 3, 2, 1)
Those who have taken algebra might remember the tuple known as the ordered pair:
(2, 3)
(11, -3)
Relations are sets of tuples, with a binary relation being a set of ordered pairs. For example, suppose we have this set:
{Diana, Steve, Barbara}
The relation “taller-than” could consist of this set of ordered pairs, where e.g. Diana is taller than Steve.
{(Diana, Steve), (Steve, Barbara), (Diana, Barbara)}
If we symbolize our taller relation as T then we could say that (Diana, Steve) ∈ T.

Relations between different sets are also possible. Suppose we have these two sets:
L = {Reed, Scott, Clark}
F = {Sue, Jean, Lois}
And the “is-husband-of” relation is a relation from set L to set F; e.g. Reed is the husband of Sue:
H = {(Reed, Sue), (Scott, Jean), (Cark, Lois)}
An inverse of a binary relation R goes like this:
R-1 = {(y, x) | (x, y) ∈ R}
For example, the inverse of the “is-husband-of” relation would be the “is-wife-of” and be this:
H-1 = {(Sue, Reed), (Jean, Scott), (Lois, Clark)}
A relation from set A to set B is a function if each member of A is paired off with exactly one member of B. The “input” part of a function is the domain (set A) and the “output” part is called the range (set B). For instance, the “is-husband-of” relation is a function because each member L is paired off with exactly one member of F, with L being the domain and F being the range, whereas an “is-husband-of” relation would not be a function if there were polygamous marriages.

Suppose relations S and T are the following:
S = {(1, 2), (10, 11)}
T = {(2, 3), (11, 12)}
A composition of two relations S and T can be symbolized as T ∘ S, and when the relations are binary, the set of ordered pairs in such a composition goes like this:
{(x, z) | (x, y) ∈ S and (y, z) ∈ T}
In our example, T ∘ S would be the following:
{(1, 3), (10, 12)}
Suppose relation V is the following:
V = {(1, 2), (1, 3), (1, 9), (2, 3), (2, 4)}
Because the relation is binary, V(x, ⋅) is { y | (x, y) ∈ V }

V(1, ⋅) = {2, 3, 9}
V(2, ⋅) = {3, 4}

Formal Logic

Deductive Arguments

A deductive argument tries to show that it’s logically impossible (i.e. self-contradictory, like a married bachelor) for the argument to have true premises and a false conclusion, and thus that the conclusion follows from the premises by the rules of logic. If it’s logically impossible for an argument to have true premises and a false conclusion the argument is deductively valid or valid. An example of a deductively valid argument:
  1. If it is raining, then my car is wet.
  2. It is raining.
  3. Therefore, my car is wet.
The above example uses a famous rule of logic called modus ponens which has this structure:
  1. If P, then Q
  2. P
  3. Therefore, Q.
Another famous rule of logic is called modus tollens where “not-Q” means “Q is false.”
  1. If P, then Q
  2. Not-Q
  3. Therefore, not-P.
An argument is deductively invalid or invalid if it is not deductively valid. An example of an invalid argument:
  1. If it is raining, then my car is wet.
  2. My car is wet.
  3. Therefore, it is raining.
In logic lingo, a deductively valid argument with all its premises being true is called a sound argument. And since a valid argument having true premises guarantees the truth of its conclusion, a sound deductive argument has a true conclusion.

Basic Symbols and Rules of Inference

Here’s a summary of how the connectives in propositional work where p and q represent propositions (claims that are either true or false):

Type of
When it’s true/false
Conjunctionp and qp ∧ qTrue if both are true; otherwise false
Disjunctionp or qp ∨ qFalse if both are false; otherwise true
ConditionalIf p, then qp → qFalse if p is true and q is false; otherwise true
NegationNot-p¬pTrue if p is false; false if p is true

As suggested in the above table, the symbols →, ¬, ∨, and ∧ are called connectives. It’s a somewhat misleading name since ¬ doesn’t connect propositions even though the other three connectives do. Still, it’s a popular label a lot of logic textbooks use. While the terminology varies among writers, I’ll call a single letter a simple statement and one more or more simple statements with one or more connectives is called a compound statement. For example, “¬P” and “A ∧ B” are compound statements.

The type of conditional (pq) being used here is called a material conditional. A material conditional is equivalent to “It is not the case that the antecedent (p) is true and the consequent (q) is false,” such that the only way for a material conditional to be false is for it to have a true antecedent with a false consequent. A material conditional might seem like a pretty weak claim (in the sense that it doesn’t claim very much), since the antecedent and consequent don’t even have to be related to each other for a material conditional to be true. Thus, “If there is a married bachelor, then Minnesota is awesome” constitutes a true material conditional since it is not the case that we have a true antecedent (there is a married bachelor) with a false consequent (Minnesota is awesome). But it turns out that a material conditional is enough for modus ponens and modus tollens to be valid rules of inference, since in a true material conditional if the antecedent is true, then the consequent is true as well.

Speaking of which, here are those rules of inference I’ve already mentioned in symbolic form:

modus ponens
In English In Symbolic Logic
If p then q

Therefore, q
p → q

∴ q
modus tollens
In English In Symbolic Logic
If p then q

Therefore, not-p
p → q

∴ ¬p

In the convention I’m using, the lower case letters p, q, r,...z are placeholders for both simple and compound statements. Thus, below is a valid instance of modus tollens.
  1. (A ∧ B) → C
  2. ¬C

  1. ¬(A ∧ B) 1, 2, modus tollens
It’s worth noting that the order of the premises doesn’t matter when using rules of inference. So below is also a valid use of modus tollens.
  1. ¬C
  2. (A ∧ B) → C

  1. ¬(A ∧ B) 1, 2, modus tollens
Some rules of inference can be used in more than one way. Examples include disjunctive syllogism and simplification.

Disjunctive Syllogism
In English In Symbolic Logic
p or q

Therefore, q
p ∨ q

∴ q
p or q

Therefore, p
p ∨ q

∴ p
In English In Symbolic Logic
p and q

Therefore, p
p ∧ q

∴ p
p and q

Therefore, q
p ∧ q

∴ q

Before moving forward, I’ll introduce a quick example of how to use some rules of inference. Suppose we wanted to get C from premises 1 and 2 below:
  1. A ∨ (B ∧ C)
  2. ¬A

  1. B ∧ C 1, 2, disjunctive syllogism
  2. C 3, simplification
Not too hard, right? After learning the above rules of inference, you might even have mentally “seen” that C followed from premises 1 and 2 above. Hopefully you are familiar enough with the symbols by now for me to remove the training wheels of english translation. Some more rules of inference:


∴ p ∧ q
hypothetical syllogism
p → q
q → r

∴ p → r


In propositional logic, two statements are logically equivalent whenever the connectives make it so that they’re always the same truth-value (i.e. both true or both false). Some rules of propositional logic are themselves equivalences, such as these:

equivalencename of equivalence
p ⇔ ¬¬pdouble negation
p → q ⇔ ¬q → ¬ptransposition (also called contraposition)
¬(p ∧ q) ⇔ ¬p ∨ ¬qDe Morgan’s laws
¬(p ∨ q) ⇔ ¬p ∧ ¬q

Equivalence rules can be used to replace stuff “inline” whenever their equivalence appears. As an example of how to use some equivalences, suppose we want to prove ¬C ∨ ¬D from premises 1 and 2 below:
  1. A
  2. (C ∧ D) → ¬A

  1. ¬¬A → ¬(C ∧ D) 2, transposition
  2. A → ¬(C ∧ D) 3, double negation
  3. ¬(C ∧ D) 1, 4 modus ponens
  4. ¬C ∨ ¬D 5, De Morgan’s laws

Conditional Proofs

The conditional is symbolized as p → q where p is called the antecedent and q is called the consequent. The conditional proof aims to prove that a conditional is true, with the antecedent of the conditional being the conditional proof assumption which is often used to help show that if the antecedent is true then the consequent is true also. The structure of a conditional proof takes the following form below:

conditional proof
a) p conditional proof assumption
c) p → q a-b, conditional proof

For example, suppose we want to prove A → (B ∧ C) from premises 1 and 2 below:
  1. A → B
  2. A → C

  1. A conditional proof assumption
    1. B 1, 3, modus ponens
    2. C 2, 3, modus ponens
    3. B ∧ C 4, 5, conjunction
  1. A → (B ∧ C) 3-6, conditional proof
Notice that the validity of a conditional proof does not rely on the conditional proof assumption actually being true; rather it relies on the fact that if it is true then it properly leads to the consequent. Nothing in the proof above, for example, relies an A actually being true.

Predicate Logic

To give an example of predicate logic, consider the following symbolization key:

B(x) = x is a Bachelor.
U(x) = x is Unmarried.

The letters B and M in these examples are predicates which say something about the element they are predicating. Sometimes parentheses aren’t used; e.g. Bx being used to mean “x is a bachelor.” The symbol; ∀ means “For All” or “For Any” such that the following basically means “All bachelors are unmarried:”

universal quantification
In English In Symbolic Logic
For any x: [if x is B, then x is U] ∀x[B(x) → U(x)]

The domain of discourse is the set of things we’re talking about when we make statements like ∀x[B(x) → U(x)], such that the “∀x” means “For any x in the domain of discourse (i.e. set of things we’re talking about here).” We can let an individual lowercase letter signify a specific element in our domain of discourse; e.g. c can signify a guy named “Charles” and we can let B(c) to signify c is B (i.e. Charles is a bachelor).

A rule of predicate logic called Universal Instantiation allows us to instantiate a universal quantification (a ∀x[...] statement) for a specific individual, like so:
  1. ∀x[B(x) → U(x)]
  2. B(c)

  1. B(c) → U(c) 1, universal instantiation
  2. U(c) 2, 3, modus ponens
There’s a somewhat complicated rule called universal generalization (also called universal introduction) to get a universal quantification statement. Roughly, the idea is that if a statement contains some variable that is a placeholder for anything in the domain of discourse, we can generalize this to get “For any x, such-and-such holds true.” The universal generalization rule is fairly complicated (you can only use it under certain specified conditions) but the gist of universal generalization should be enough to follow along this proof.
  1. ∀x[A(x) → B(x)]
  2. ∀x[B(x) → C(x)]

  1. A(t) → B(t) 1, universal instantiation
  2. B(t) → C(t) 2, universal instantiation
  3. A(t) → C(t) 3, 4, hypothetical syllogism
  4. ∀x[A(x) → C(x)] 5, universal generalization
And that should be all the technical stuff you need to know to read A Folk Semantics Argument for Moral Non-Naturalism. You still might not find it an easy read if you’re not used to analytic philosophy, but at least you have the background knowledge even if applying it is a bit tricky.