3
Since we need to know what premises mean before we can appraise their truth, language is important for appraising arguments. Imagine someone giving this argument, which is deductively valid but has obscure premises:
· If there’s a cosmic force, then there’s a God.
· There’s a cosmic force.
· ∴ There’s a God.
What does the person mean by “cosmic force”? We can’t intelligently agree or disagree with premises if we don’t understand what they mean.
In this chapter, after looking at general uses of language, we’ll examine definitions and other ways to clarify meaning. Then we’ll talk about making distinctions and detecting unclarities. Finally, we’ll consider the distinction between analytic and synthetic statements, and the related distinction between knowledge based on reason and knowledge based on experience. The goal of this study of language is to enhance our ability to analyze and appraise arguments.
3.1 Uses of language
The four grammatical sentence types broadly reflect four uses of language:
· Declarative (make assertions): “Michigan beat Ohio State.”
· Interrogative (ask questions): “Did Michigan win?”
· Imperative (tell what to do): “Beat Ohio State.”
· Exclamatory (express feelings): “Hurrah for Michigan!”
Sentences can do various jobs at the same time. While making assertions, we can also ask questions, tell what to do, or express feelings:
· “I wonder whether Michigan won.” (This can ask a question.)
· “I want you to throw the ball.” (This can tell what to do.)
· “Michigan won!” (This can express feelings of joy.)
Arguments too can exemplify different uses of language. Suppose someone 0032 argues this way about the Cleveland river that used to catch on fire: “You can see that the Cuyahoga River is polluted from the fact that it can burn!” We could make this into an explicit argument:
· No pure water is burnable.
· Some Cuyahoga River water is burnable.
· ∴ Some Cuyahoga River water isn’t pure water.
One who argues this way also can (perhaps implicitly) be raising a question, directing people to do something, or expressing feelings:
· “What can we do to clean up this polluted river?”
· “Let’s all resolve to take action on this problem.”
· “How disgusting is this polluted river!”
Arguments have a wider human context and purpose. We should remember this when we study detached specimens of argumentation.
When we do logic, our focus narrows and we concentrate on assertions and reasoning. For this purpose, detached specimens are better. Expressing an argument in a clear, direct, emotionless way can make it easier to appraise the truth of the premises and the validity of the inference.
It’s good to avoid emotional language when we reason. Of course, there’s nothing wrong with feelings or emotional language. Reason and feeling are both important parts of life; but we often need to focus on one or the other for a given purpose. At times, expressing feelings is the important thing and argumentation only gets in the way. At other times, we need to reason things out in a cool-headed manner.
Emotional language can discourage clear reasoning. When reasoning about abortion, for example, it’s wise to avoid slanted phrases like “the atrocious crime of abortion” or “Neanderthals who oppose the rights of women.” Bertrand Russell gave this example of how we slant language: “I am firm; you are obstinate; he is pig-headed.” Slanted phrases can mislead us into thinking we’ve defended our view by an argument (premises and conclusion) when in fact we’ve only expressed feelings. Careful thinkers try to avoid emotional terms when constructing arguments.
In the rest of this chapter, we’ll explore aspects of the “making assertions” side of language that relate closely to analyzing and appraising arguments.
3.1a Exercise
For each word or phrase, say whether it has a positive, negative, or neutral emotional tone. Then find another word or phrase with more or less the same assertive meaning but a different emotional tone. 0033
old maid
This has a negative tone. A more neutral phrase is “elderly woman who has never married.”
(The term “old maid” suggests that a woman’s goal in life is to get married and that an older woman who has never married is unfortunate. I can’t think of a corresponding negative term for an older man who never married. A word or phrase sometimes suggests a whole attitude toward life, and often an unexamined attitude.)
1. a cop
2. filthy rich
3. heroic
4. an extremist
5. an elderly gentleman
6. a bastard
7. baloney
8. a backward country
9. authoritarian
10.a do-gooder
11.a hair-splitter
12.an egghead
13.a bizarre idea
14.a kid
15.booze
16.a gay
17.abnormal
18.bureaucracy
19.abandoning me
20.babbling
21.brazen
22.an old broad
23.old moneybags
24.a busybody
25.a bribe
26.old-fashioned
27.brave
28.garbage
29.a cagey person
30.a whore
3.2 Lexical definitions
We noted earlier that the phrase “cosmic force” in this example is obscure:
· If there’s a cosmic force, then there’s a God.
· There’s a cosmic force.
· ∴ There’s a God.
Unless the speaker tells us what is meant by “cosmic force,” we won’t be able to understand what’s said or tell whether it’s true. But how can the speaker explain what he or she means by “cosmic force”? Or, more generally, how can we explain the meaning of a word or phrase?
Definitions are an important way to explain meaning. A definition is a rule of paraphrase intended to explain meaning. More precisely, a definition of a word or phrase is a rule saying how to eliminate this word or phrase in a sentence and produce a second sentence that means the same thing, the purpose of this being to explain or clarify the meaning of the word or phrase.
Suppose the person with the cosmic-force argument defines “cosmic force” as “force, in the sense used in physics, whose influence covers the entire universe.” This makes the first premise doubtful, since then it only means “If there’s a force [e.g., gravity], in the sense used in physics, whose influence covers the entire universe, then there’s a God.” Why think that this premise is true?
So a definition is a rule of paraphrase intended to explain meaning. Definitions may be lexical (explaining current usage) or stipulative (specifying your own usage). Here’s a correct lexical definition: 0034
· “Bachelor” means “unmarried man.”
This claims that we can interchange “bachelor” and “unmarried man” in a sentence; the resulting sentence will mean the same as the original, according to current usage. This leads to the interchange test for lexical definitions:
Interchange test: To test a lexical definition claiming that A means B, try switching A and B in a variety of sentences. If some resulting pair of sentences doesn’t mean the same thing, then the definition is incorrect.
According to our definition of “bachelor” as “unmarried man,” for example, these two sentences would mean the same thing:
· “Al is a bachelor”
· “Al is an unmarried man”
These do seem to mean the same thing. To refute the definition, we’d have to find two sentences that are alike, except that “bachelor” and “unmarried man” are interchanged, and that don’t mean the same thing.
Here’s an incorrect lexical definition:
· “Bachelor” means “happy man.”
This leads to incorrect paraphrases. If the definition were correct, then these two sentences would mean the same thing:
· “Al is a bachelor”
· “Al is a happy man”
But they don’t mean the same thing, since we could have one true but not the other. So the definition is wrong.
The interchange test is subject to at least two restrictions. First, definitions are often intended to cover just one sense of a word that has various meanings; we should then use the interchange test only on sentences using the intended sense. Thus it wouldn’t be a good objection to our definition of “bachelor” as “unmarried man” to claim that these two sentences don’t mean the same:
· “I have a bachelor of arts degree”
· “I have an unmarried man of arts degree”
The first sentence uses “bachelor” in a sense the definition doesn’t try to cover.
Second, we shouldn’t use the test on sentences where the word appears in quotes. Consider this pair of sentences: 0035
· “‘Bachelor’ has eight letters”
· “‘Unmarried man’ has eight letters”
The two don’t mean the same thing, since the first is true and the second false. But this doesn’t show that our definition is wrong.
Lexical definitions are important in philosophy. Many philosophers, from Socrates to the present, have sought correct lexical definitions for some of the central concepts of human existence. They’ve tried to define concepts such as knowledge, truth, virtue, goodness, and justice. Such definitions are important for understanding and applying the concepts. Defining “good” as “what society approves of” would lead us to base our ethical beliefs on what’s socially approved. We’d reject this method if we defined “good” as “what I like” or “what God desires,” or if we regarded “good” as indefinable.
We can evaluate philosophical lexical definitions using the interchange test; Socrates was adept at this. Consider cultural relativism’s definition of “good”:
· “X is good” means “X is approved by my society.”
To evaluate this, we’d try switching “good” and “approved by my society” in a sentence to get a second sentence. Here’s such a pair of sentences:
· “Slavery is good”
· “Slavery is approved by my society”
Then we’d see if the two sentences mean the same thing. Here they clearly don’t, since it’s consistent to affirm one but deny the other. Those who disagree with the norms of their society often say things like “Slavery is approved by my society, but it’s not good.” Given this, we can argue against cultural relativism’s definition of “good” as follows:
· If cultural relativism’s definition is correct, then these two sentences mean the same thing.
· They don’t mean the same thing.
· ∴ Cultural relativism’s definition isn’t correct.
To counter this, the cultural relativist would have to claim that the sentences do mean the same thing. But this claim is implausible.
Here are five rules for good lexical definitions:
1. 1. A good lexical definition is neither too broad nor too narrow.
Defining “bachelor” as “man” is too broad, since some men aren’t bachelors. And defining “bachelor” as “unmarried male astronaut” is too narrow, since some bachelors aren’t astronauts. 0036
· 2. A good lexical definition avoids circularity and poorly understood terms.
Defining “true” as “known to be true” is circular, since it defines “true” using “true.” And defining “good” as “having positive aretaic value” uses poorly understood terms, since “aretaic” is less clear than “good.”
· 3. A good lexical definition matches in vagueness the term defined.
Defining “bachelor” as “unmarried male over 18 years old” is overly precise. “Bachelor” is vague, since the exact age that the term begins to apply is unclear on semantic grounds; so “over 18” is too precise to define “bachelor.” “Man” or “adult” are better, since these match “bachelor” fairly well in vagueness.
· 4. A good lexical definition matches, as far as possible, the emotional tone (positive, negative, or neutral) of the term defined.
It won’t do to define “bachelor” as “fortunate man who never married” or “unfortunate man who never married.” These have positive and negative emotional slants; the original term “bachelor” is fairly neutral.
· 5. A good lexical definition includes only properties essential to the term.
Suppose all bachelors live on the planet earth. Even so, living on planet earth isn’t a property essential to the term “bachelor,” since we could imagine a bachelor who lives on the moon. So it’s wrong to include “living on planet earth” in the definition of “bachelor.”
3.2a Exercise: LogiCola Q
Give objections to these proposed lexical definitions.
“Game” means “anything that involves competition between two parties, and winning and losing.”
By this definition, solitaire isn’t a game, but a military battle is. This goes against the normal usage of the word “game.”
1. “Lie” means “false statement.”
2. “Adolescent” means “person between 9 and 19 years old.”
3. “God” means “object of ultimate concern.”
4. “Metaphysics” means “any sleep-inducing subject.”
5. “Good” means “of positive value.”
6. “Human being” means “featherless biped.”
7. “I know that P” means “I believe that P.”
8. “I know that P” means “I believe that P, and P is true.”
9. “Chair” means “what you sit on.” 0037
10.“True” means “believed.”
11.“True” means “proved to be true.”
12.“Valid argument” means “argument that persuades.”
13.“Murder” means “killing.”
14.“Morally wrong” means “against the law.”
15.“Philosopher” means “someone who has a degree in philosophy” and “philosophy” means “study of the great philosophers.”
3.2b Exercise
Cultural relativism (CR) claims that “good” (in its ordinary usage) means “socially approved” or “approved by the majority (of the society in question).” What does this definition entail about the statements below? If this definition were correct, then would each of the following be true (1), false (0), or undecided by such considerations (?)?
If torturing people for religious beliefs is socially approved in country X, then it’s good in country X.
1 (for “true”). On cultural relativism, the statement means this (and would be true): “If torturing people for religious beliefs is socially approved in country X, then it’s socially approved in country X.”
1. Conclusions about what is good are deducible from sociological data (based, for example, on opinion surveys) describing one’s society and what it approves.
2. If I say “Infanticide isn’t good” but an ancient Roman says “Infanticide is good,” then one or the other of us must be mistaken.
3. The norms set up by my society about what is good couldn’t be mistaken.
4. Judgments about what is good aren’t true or false.
5. It’s good to respect the values of other societies.
6. If our society were to favor intolerance, then intolerance would be good.
7. Representational democracy will work anywhere.
8. From an analysis of how people use the word “good,” it can be proved that whatever is socially approved must be good.
9. Different cultures accept different moral beliefs.
10.“The majority favors this” logically entails “This is good.”
11.If the majority favors war (sexual stereotypes, conservative politics, abortion, and so on), then this has to be good.
12.“Do good” means “Do what the majority favors.”
13.Doing something because it’s good isn’t the same as doing it because the majority favors it.
14.People who said “Racism is favored by the majority but it’s not good” were contradicting themselves.
15.Something that’s bad might nevertheless be socially approved (because society may be misinformed or irrational in its evaluations).
16.The majority knows what it favors.
17.If Nazism became widespread and genocide came to be what most people favored, then genocide would have to be good.
18.It’s not necessarily good for me to do what society favors. 0038
19.Suppose a survey showed that 90 percent of the population disapprove of people always following social approval. Then it follows that it’s bad to always follow social approval – in other words, it’s bad to always follow what is good.
20.Suppose your fellow Americans as a group and your fellow Anglicans as a group disapprove of racism, whereas your fellow workers and your social group (friends and relatives) approve of racism. Then racism is bad.
3.3 Stipulative definitions
A stipulative definition specifies how you’re going to use a term. Since your usage may be a new one, it’s unfair to criticize a stipulative definition for clashing with conventional usage. Stipulative definitions should be judged, not as correct or incorrect, but rather as useful or useless.
This book has many stipulative definitions. I continually define terms like “logic,” “argument,” “valid,” “wff,” and so forth. These definitions specify the meaning I’m going to use for the terms, which sometimes is close to their standard meaning. The definitions create a technical vocabulary.
A clarifying definition is one that stipulates a clearer meaning for a vague term. For example, a scientist might stipulate a technical sense of “pure water” in terms of bacteria level; this technical sense, while related to the normal one, is more scientifically precise. Likewise, courts might stipulate a more precise definition of “death” to resolve certain legal disputes; the definition might be chosen on moral and legal grounds to clarify the law.
Philosophers often use stipulative definitions. They might say: “Here I’ll use ‘rational’ to mean ‘always adopting the means believed necessary to achieve one’s goals.’” This signals that the author will use “rational” to abbreviate a certain longer phrase; it doesn’t claim that this exactly matches the term’s ordinary meaning. Others may use “rational” in different senses, such as “logically consistent,” “emotionless,” or “forming beliefs solely by the methods of science.” These thinkers needn’t be disagreeing; they may just specify their technical vocabulary differently. We could use subscripts for different senses; “rational1” might mean “logically consistent,” and “rational2” might mean “emotionless.” Don’t be misled into thinking that, because being rational in one sense is desirable, therefore being rational in another sense must also be desirable.
Stipulative definitions, while they needn’t accord with current usage, should:
· use clear terms that the parties involved will understand,
· avoid circularity,
· let us paraphrase out the defined term,
· accord with how the person giving it will use the term, and
· aid our understanding and discussion of the subject matter.
A stipulative definition is a device for abbreviating language. Our Chapter 1 0039 starts with a stipulative definition: “Logic is the analysis and appraisal of arguments.” This definition lets us use the one word “logic” in place of the six words “the analysis and appraisal of arguments.” Without the definition, our explanations would be wordier and harder to grasp; so the definition is useful.
Stipulative definitions should promote understanding. It’s seldom useful to stipulate that a well-established term will be used in a radical new sense (for example, that “biology” will be used to mean “the study of earthquakes”); this would create confusion. And it’s seldom useful to multiply stipulative definitions for terms that we’ll seldom use. But when we keep repeating a cumbersome phrase over and over, a stipulative definition can be helpful. Suppose your essay keeps repeating the phrase “action that satisfies criteria 1, 2, and 3 of the previous section”; your essay may be easier to follow if some short term were stipulated to mean the same as this longer phrase.
Some of our definitions seem to violate the “avoid circularity” norm. Section 6.1 defines “wffs” as sequences constructable using these rules:
1. Any capital letter is a wff.
2. The result of prefixing any wff with “∼” is a wff.
3. The result of joining any two wffs by “•” or “∨” or “⊃” or “≡” and enclosing the result in parentheses is a wff.
Clauses 2 and 3 define “wff” in terms of “wff.” And the definition doesn’t seem to let us paraphrase out the term “wff”; we don’t seem able to take a sentence using “wff” and say the same thing without “wff.”
Actually, our definition is perfectly fine. We can rephrase it in the following way to avoid circularity and show how to paraphrase out the term “wff”:
“Wff” means “member of every set S of strings that satisfies these conditions: (1) Every capital letter is a member of set S; (2) the result of prefixing any member of set S with ‘∼’ is a member of set S; and (3) the result of joining any two members of set S by ‘•’ or ‘∨’ or ‘⊃’ or ‘≡’ and enclosing the result in parentheses is a member of set S.”
Our definition of “wff” is a recursive definition – one that first specifies some things that the term applies to and then specifies that if the term applies to certain things, then it also applies to certain other things. Here’s a recursive definition of “ancestor of mine” – followed by an equivalent non-recursive definition:
1. My father and mother are ancestors of mine.
2. Any father or mother of an ancestor of mine is an ancestor of mine.
“Ancestor of mine” means “member of every set S that satisfies these conditions: (1) my father and mother are members of S; and (2) every father or mother of a member of S is a member of S.” 0040
3.4 Explaining meaning
If we avoid circular definitions, we can’t define all our terms; instead, we must leave some terms undefined. But how can we explain such undefined terms? One way is by examples.
To teach “red” to someone who understands no language that we speak, we could point to red objects and say “Red!” We’d want to point to different kinds of red object; if we pointed only to red shirts, the person might think that “red” meant “shirt.” If the person understands “not,” we also could point to non-red objects and say “Not red!” The person, unless color-blind, soon will catch our meaning and be able to point to red objects and say “Red!” This is a basic, primitive way to teach language. It explains a word, not by using other words, but by relating a word to concrete experiences.
We sometimes point to examples through words. We might explain “plaid” to a child by saying “It’s a color pattern like that of your brother’s shirt.” We might explain “love” through examples: “Love is getting up to cook a sick person’s breakfast instead of staying in bed, encouraging someone instead of complaining, and listening to other people instead of telling them how great you are.” It’s often helpful to combine a definition with examples, so the two reinforce each other; so Chapter 1 defined “argument” and then gave examples.
In abstract discussions, people sometimes use words so differently that they communicate poorly and almost seem to speak different languages. Asking for definitions may then lead to the frustration of having one term you don’t understand being defined using other terms you don’t understand. It may be more helpful to ask for examples: “Give me examples of an analytic statement (or of a deconstruction).” Asking for examples can bring a bewilderingly abstract discussion back down to earth and mutual understanding.
Logical positivists and pragmatists gave other ways to clarify statements. Positivists proposed that we explain a statement’s meaning by specifying which experiences would show the statement to be true or to be false. Such operational definitions connect meaning to an experimental test:
· To say that rock A is “harder than” rock B means that A would scratch B but B wouldn’t scratch A.
· To say that this string is “1 meter long” means that, if you stretch it over the standard meter stick, then the ends of both will coincide.
· To say that this person “has an IQ of 100” means that the person would get an average score on a standard IQ test.
Such definitions are important in science.
Logical positivists like A. J. Ayer appealed to the verifiability criterion of meaning as the cornerstone of their philosophy. We can formulate their principle (to be applied only to statements not true-by-definition, see §3.6) as follows: 0041
Logical positivism (LP)
To help us find a statement’s meaning, ask “How could the truth or falsity of the statement in principle be discovered by conceivable observable tests?”
If there’s no way to test a statement, then it has no meaning (it makes no assertion that could be true or false). If tests are given, they specify the meaning.
There are problems with taking LP to be literally true. LP says any untestable statement is without meaning. But LP itself is untestable. Hence LP is without meaning on its own terms; it’s self-refuting. For this reason and others, few hold this view anymore, even though it was popular decades ago.
Still, the LP way to clarify statements can sometimes be useful. Consider this claim of Thales, the ancient Greek alleged to be the first philosopher: “Water is the primal stuff of reality.” The meaning here is unclear. We might ask Thales for a definition of “primal stuff”; this would clarify the claim. Or we might follow LP and ask, “How could we test whether your claim is correct?” Suppose Thales says the following, thus giving an operational definition:
Try giving living things no water. If they die, then this proves my claim. If they live, then this refutes my claim.
We’d then understand Thales to be claiming that water is needed for life. Or suppose Thales replies this way:
Let scientists work on the task of transforming each kind of matter (gold, rock, air, and so on) into water, and water back into each kind of matter. If they eventually succeed, then that proves my claim.
Again, this would help us understand the claim. But suppose Thales says “No conceivable experimental test could show my claim to be true or show it to be false.” The positivists would immediately conclude that Thales’s claim is meaningless – that it makes no factual assertion that could be true or false. We non-positivists needn’t draw this conclusion so quickly; but we may remain suspicious of Thales’s claim and wonder what he’s getting at.
LP demands that a statement in principle be able to be tested. Consider “There are mountains on the other side of the moon.” When the positivists wrote, rockets were less advanced and the statement couldn’t be tested. But that didn’t matter to its meaningfulness, since we could describe what a test would be like. That this claim was testable in principle was enough to make it meaningful.
LP hides an ambiguity when it speaks of “conceivable observable tests.” Observable by whom? Is it enough that one person can make the observation? Or does it have to be publicly observable? Is a statement about my present feelings meaningful if I alone can observe whether it’s true? Historically, most positivists demanded that a statement be publicly verifiable. But the weaker version 0042 of the theory that allows verification by one person seems better. After all, a statement about my present feelings makes sense, but only I can verify it.
William James suggested a related way to clarify statements. His “Pragmatism” essay suggests that we determine the meaning, or “cash value,” of a statement by relating it to practical consequences. James’s view is broader and more tolerant than that of the positivists. We can formulate his pragmatism principle as follows (again, it’s to be applied only to statements not true-by-definition):
Pragmatism (PR)
To help us find a statement’s meaning, ask “What conceivable practical differences to someone could the truth or falsity of the statement make?” Here “practical differences to someone” covers what experiences one would have or what choices one ought to make.
If the truth or falsity of a statement could make no practical difference to anyone, then it has no meaning (it makes no assertion that could be true or false). If practical differences are given, they specify the meaning.
I’m inclined to think that something close to PR is literally true. But here I’ll just stress that PR can be useful in clarifying meaning.
PR often applies much like the weaker version of LP that allows verification by one person. LP focuses on what we could experience if the statement were true or false, while PR includes such experiences under practical differences.
PR also includes under “practical differences” what choices one ought to make. This makes PR broader than LP, since what makes a difference to choices needn’t be testable by observation. Hedonism claims “Only pleasure is worth striving for.” LP asks “How could the truth or falsity of hedonism in principle be discovered by conceivable observable tests?” Perhaps it can’t; then LP would see hedonism as cognitively meaningless. PR asks “What conceivable practical differences to someone could the truth or falsity of hedonism make?” Here, “practical differences” include what choices one ought to make. The truth of hedonism could make many differences about choices; if hedonism is true, for example, then we should pursue knowledge not for its own sake but only insofar as it promotes pleasure. Ethical claims like hedonism are meaningless on LP but meaningful on the more tolerant PR.
In addition, PR isn’t self-refuting. LP says “Any untestable statement is without meaning.” But LP itself is untestable, and so is meaningless on its own terms. But PR says “Any statement whose truth or falsity could make no conceivable practical difference is without meaning.” PR makes a practical difference to our choices about beliefs; presumably we shouldn’t believe statements that fail the PR test. And so PR can be meaningful on its own terms.
So we can explain words by definitions, examples, verification conditions, and practical differences. Another way to convey meaning is by contextual use: we 0043 use a word in such a way that its meaning can be gathered from surrounding “clues.” Suppose a person getting in a car says “I’m getting in my C”; we can surmise that C means “car.” We all learned language mostly by picking up meaning from contextual use.
Some thinkers want us to pick up their technical terms in this same way. We are given no definitions of key terms, no examples to clarify their use, and no explanations in terms of verification conditions or practical differences. We are just told to dive in and catch the lingo by getting used to it. We should be suspicious of this. We may catch the lingo, but it may turn out to be empty and without meaning. That’s why the positivists and pragmatists emphasized finding the “cash value” of ideas in terms of verification conditions or practical differences. We must be on guard against empty jargon.
3.4a Exercise
Would each claim be meaningful or meaningless on LP? (Take LP to require that a statement be publicly testable.) Would each be meaningful or meaningless on PR?
Unless we have strong reasons to the contrary, we ought to believe what sense experience seems to reveal.
This is meaningless on LP, since claims about what one ought to do aren’t publicly testable. It’s meaningful on PR, since its truth could make a difference about what choices we ought to make about beliefs.
1. It’s cold outside.
2. That clock is fast.
3. There are five-foot-long blue ants in my bedroom.
4. Nothing is real.
5. Form is metaphysically prior to matter.
6. At noon all lengths, distances, and velocities in the universe will double.
7. I’m wearing an invisible hat that can’t be felt or perceived in any way.
8. Regina has a pain in her little toe but shows no signs of this and will deny it if you ask her.
9. Other humans have no thoughts or feelings but only act as if they do.
10.Manuel will continue to have conscious experiences after his physical death.
11.Angels exist (that is, there are thinking creatures who have never had spatial dimensions or weights).
12.God exists (that is, there’s a very intelligent, powerful, and good personal creator of the universe).
13.One ought to be logically consistent.
14.Any statement whose truth or falsity could make no conceivable practical difference is meaningless. (PR)
15.Any statement that isn’t observationally testable is meaningless. (LP) 0044
3.5 Making distinctions
Philosophers faced with difficult questions often make distinctions:
“If your question means … [giving a clear phrasing], then my answer is …. But if you’re really asking …, then my answer is ….”
The ability to formulate various possible meanings of a question is a valuable skill. Many of the questions that confront us are vague or confused; we often have to clarify a question before we can answer it intelligently. Getting clear on a question can be half the battle.
Consider this question (in which I underlined the tricky word “indubitable”):
· “Are some beliefs indubitable?”
What does “indubitable” here mean? Does it mean not actually doubted? Or psychologically impossible to doubt? Or irrational to doubt? And what is it to doubt? Is it to refrain from believing? Or is it to have some suspicion about the belief (although we might still believe it)? And indubitable by whom? By everyone (even crazy people)? By all rational persons? By at least some individuals? By me? Our little question hides a sea of ambiguities. Here are three of the many things that our little question could be asking:
· Are there some beliefs that no one has ever refused to believe? (To answer this, we’d need to know whether people in insane asylums sometimes refuse to believe that they exist or that “2 = 2.”)
· Are there some beliefs that no rational person has suspicions about? (To answer this, we’d first have to decide what we mean by “rational.”)
· Are there some beliefs that some specific individuals are psychologically unable to have any doubts about? (Perhaps many are unable to have any doubts about what their name is or where they live.)
It’s risky to answer questions that we don’t understand.
Unnoticed ambiguities can block communication. Often people are unclear about what they’re asking, or take another’s question in an unintended sense. This is more likely if the discussion goes abstractly, without examples.
3.5a Exercise
Each of the following questions is obscure or ambiguous as it stands. Distinguish at least three interesting senses of each question. Formulate each sense simply, clearly, and briefly – and without using the underlined words. 0045
Can one prove that there are external objects?
· Can we deduce, from premises expressing immediate experience (like “I seem to see a blue shape”), that there are external objects?
· Can anyone give an argument that will convince (all or most) skeptics that there are external objects?
· Can anyone give a good deductive or inductive argument, from premises expressing their immediate experience in addition to true principles of evidence, to conclude that it’s reasonable to believe that there are external objects? (These “principles of evidence” may include things like “Unless we have strong reasons to the contrary, it’s reasonable to believe what sense experience seems to reveal.”)
1. Is ethics a science?
2. Is this monkey a rational animal?
3. Is this belief part of common sense?
4. Are material objects objective?
5. Are values relative (or absolute)?
6. Are scientific generalizations ever certain?
7. Was the action of that monkey a free act?
8. Is truth changeless?
9. How are moral beliefs explainable?
10.Is that judgment based on reason?
11.Is a fetus a human being (or human person)?
12.Are values objective?
13.What is the nature of man?
14.Can I ever know what someone else feels?
15.Do you have a soul?
16.Is the world illogical?
3.6 Analytic and synthetic
Immanuel Kant long ago introduced two related distinctions that have become influential. He divided statements, on the basis of their meaning, into analytic and synthetic statements. He divided knowledge, on the basis of how it’s known, into a priori and a posteriori knowledge. We’ll consider these distinctions in this section and the next.1
1 I’ll sketch a standard approach to these Kantian distinctions. Willard Quine, in his Philosophy of Logic, 2nd ed. (Cambridge, Mass.: Harvard University Press, 1986), criticizes this approach.
Kant gave two definitions of “analytic statement”:
1. An analytic statement is one whose subject contains its predicate.
2. An analytic statement is one that’s self-contradictory to deny. 0046
Consider these examples (and take “bachelor” to mean “unmarried man”):
· (a) “All bachelors are unmarried.”
· (b) “If it’s raining, then it’s raining.”
Both examples are analytic by definition 2, since both are self-contradictory to deny. But only (a) is analytic by definition 1. In (a), the subject “bachelor” (“unmarried man”) contains the predicate “unmarried”; but in (b), the subject “it” doesn’t contain the predicate.
We’ll adopt definition 2; so we define an analytic statement as one that’s self-contradictory to deny. Logically necessary truth is another term for the same idea; such truths are based on logic, the meaning of concepts, or necessary connections between properties. Here are some further analytic statements:
· “2 = 2”
· “1 > 0”
· “All frogs are frogs.”
· “If everything is green, then this is green.”
· “If there’s rain, then there’s precipitation.”
· “If this is green, then this is colored.”
By contrast, a synthetic statement is one that’s neither analytic nor self-contradictory; contingent is another term for the same idea. Statements divide into analytic, synthetic, and self-contradictory; here’s an example of each:1
1 Modal logic (Chapters 10 and 11) symbolizes “A is analytic (necessary)” as “☐A,” “A is synthetic (contingent)” as “(◇A • ◇∼A),” and “A is self-contradictory” as “∼◇A.”
· Analytic: “All bachelors are unmarried.”
· Synthetic: “Daniel is a bachelor.”
· Self-contradictory: “Daniel is a married bachelor.”
While there are three kinds of statement, there are only two kinds of truth: analytic and synthetic. Self-contradictory statements are necessarily false.
3.6a Exercise
Say whether each of these is analytic or synthetic. Take terms in their most natural senses. Some examples are controversial.
All triangles are triangles.
This is analytic. It would be self-contradictory to deny it and say “Some triangles aren’t triangles.”
1. All triangles have three angles.
2. 2 + 2 = 4.
3. Combining two drops of mercury with two other drops results in one big drop.
4. There are ants that have established a system of slavery.
5. Either some ants are parasitic or else none are.
6. No three-year-old is an adult. 0047
7. No three-year-old understands symbolic logic.
8. Water boils at 90°C on that 10,000-foot mountain.
9. Water boils at 100°C at sea level.
10.No uncle who has never married is an only child.
11.All swans are white.
12.Every material body is spatially located and has spatial dimensions.
13.Every material body has weight.
14.The sum of the angles of a Euclidian triangle equals 180°.
15.If all Parisians are French and all French are European, then all Parisians are European.
16.Every event has a cause.
17.Every effect has a cause.
18.We ought to treat a person not simply as a means but always as an end in itself.
19.One ought to be logically consistent.
20.God exists.
21.Given that we’ve observed that the sun rose every day in the past, it’s reasonable for us to believe that the sun will rise tomorrow.
22.Unless we have strong reasons to the contrary, we ought to believe what sense experience seems to reveal.
23.Everything red is colored.
24.Nothing red is blue (at the same time and in the same part and respect).
25.Every synthetic statement that’s known to be true is known on the basis of sense experience. (There’s no synthetic a priori knowledge.)
3.7 A priori and a posteriori
Philosophers traditionally distinguish two kinds of knowledge. A posteriori (empirical) knowledge is based on sense experience. A priori (rational) knowledge is based on reason, not sense experience. Here’s an example of each:
· A posteriori: “Some bachelors are happy.”
· A priori: “All bachelors are unmarried.”
While we know both to be true, how we know them differs. We know the first statement from our experience of bachelors; we’ve met many bachelors and recall that some have been happy. If we had to justify the truth of this statement to others, we’d appeal to experiential data about bachelors. In contrast, we know the second statement by grasping what it means and seeing that it must be true. If we had to justify the truth of this statement, we wouldn’t have to gather experiential data about bachelors.
Most knowledge is a posteriori – based on sense experience. “Sense experience” here covers the five “outer senses” (sight, hearing, smell, taste, and touch). It also covers “inner sense” (the awareness of our own thoughts and feelings) and any other experiential access to the truth that we might have (perhaps even 0048 mystical experience or extrasensory perception).
Logical and mathematical knowledge is generally a priori. To test the validity of an argument, we don’t go out and do experiments. Instead, we just think and reason; sometimes we write things out to help our thinking. The validity tests in this book are rational (a priori) methods. “Reason” in a narrow sense (in which it contrasts with “experience”) deals with what we can know a priori.
A priori knowledge requires some experience. We can’t know that all bachelors are unmarried unless we’ve learned the concepts involved; this requires experience of language and of (married and unmarried) humans. And knowing that all bachelors are unmarried requires the experience of thinking. So a priori knowledge depends somewhat on experience (and thus isn’t just something that we’re born with). But it still makes sense to call such knowledge a priori. Suppose we’ve gained the concepts using experience. Then to justify the claim that all bachelors are unmarried, we don’t have to appeal to any further experience, other than thinking. In particular, we don’t have to investigate bachelors to see whether they’re all unmarried.1
1 David Hume, who thought that all concepts come from experience, defended a priori knowledge. By comparing two empirical concepts, we can sometimes recognize that the empirical conditions that would verify one (“bachelor”) would also verify the other (“unmarried”); so by reflecting on our concepts, we can see that all bachelors must be unmarried.
Here are some further examples of statements known a priori:
· “2 = 2”
· “1 > 0”
· “All frogs are frogs.”
· “If everything is green, then this is green.”
· “If there’s rain, then there’s precipitation.”
· “If this is green, then this is colored.”
We also gave these as examples of analytic statements.
So far, we’ve used only analytic statements as examples of a priori knowledge and only synthetic statements as examples of a posteriori knowledge. Some philosophers think there’s only one distinction, but drawn in two ways:
· a priori knowledge = analytic knowledge
· a posteriori knowledge = synthetic knowledge 0049
Is this view true? If it’s true at all, it’s not true just because of how we defined the terms. By our definitions, the basis for the analytic / synthetic distinction differs from the basis for the a priori / a posteriori distinction. A statement is analytic or synthetic depending on whether its denial is self-contradictory; but knowledge is a posteriori or a priori depending on whether it rests on sense experience. Our definitions leave it open whether the two distinctions coincide.
These two combinations are very common:
· analytic a priori knowledge
· synthetic a posteriori knowledge
Most of our knowledge in math and logic is analytic a priori. Most of our everyday and scientific knowledge about the world is synthetic a posteriori. These next two combinations are more controversial:
· analytic a posteriori knowledge
· synthetic a priori knowledge
Can we know any analytic statements a posteriori? It seems that we can. “π is a little over 3” is presumably an analytic truth that can be known either by a priori calculations (the more precise way to compute π) – or by measuring circles empirically (as the ancient Egyptians did). And “It’s raining or not raining” is an analytic truth that can be known either a priori (and justified by truth tables, see §6.6) – or by deducing it from the empirical statement “It’s raining.” But perhaps any analytic statement that’s known a posteriori also could be known a priori.
The biggest issue is this: “Do we have any synthetic a priori knowledge?” This asks whether there’s any statement A such that:
· A is synthetic (not self-contradictory either to affirm or to deny),
· we know A to be true, and
· our knowledge of A is based on reason (and not sense experience)?
In one sense of the term, an empiricist is one who rejects such knowledge – and who thus limits what we can know by pure reason to analytic statements. By contrast, a rationalist is one who accepts such knowledge – and who thus gives a greater scope to what we can know by pure reason.1
1 More broadly, empiricists are those who emphasize a posteriori knowledge, while rationalists are those who emphasize a priori knowledge.
Empiricists deny the possibility of synthetic a priori knowledge for two main reasons. First, it’s difficult to understand how there could be such knowledge. Analytic a priori knowledge is fairly easy to grasp. Suppose a statement is true simply because of the meaning and logical relations of the concepts involved; then we can know it in an a priori fashion by reflecting on these concepts and logical relations. But suppose a statement could logically be either true or false. How could we then possibly know by pure thinking which it is?
Second, those who accept synthetic a priori truths differ on what these truths are. They just follow their prejudices and call them “deliverances of reason.”
Rationalists accept synthetic a priori knowledge for two main reasons. First, the opposite view (at least if it’s claimed to be known) seems self-refuting. Consider empiricists who claim to know “There’s no synthetic a priori knowledge.” Now this claim is synthetic (it’s not true by how we defined the terms “synthetic” and “a priori,” and it’s not self-contradictory to deny). And it would have to be known a priori (since we can’t justify it by sense experience). So the empiricist’s claim would have to be synthetic a priori knowledge, which it rejects. 0050
Second, we seem to have synthetic a priori knowledge of ideas like this:
If you believe you see an object to be red and have no special reason to doubt your perception [e.g., the lighting is strange or you’re on mind-altering drugs], then it’s reasonable for you to believe that you see an actual red object.
This claim is synthetic; it’s not true because of how we’ve defined terms, and skeptics can deny it without self-contradiction. It’s presumably known to be true; if we didn’t know such truths, then we couldn’t justify any empirical beliefs. And it’s known a priori; empirical knowledge depends on it instead of it depending on empirical knowledge. So we have synthetic a priori knowledge of this claim. So there’s synthetic a priori knowledge.
The dispute over synthetic a priori knowledge influences how we do philosophy. Can basic ethical principles be known a priori? Empiricists say no; so then we know basic ethical principles either empirically or not at all. But rationalists can (and often do) think that we know basic ethical truths a priori, from reason alone (through either intuition or some rational consistency test).
3.7a Exercise
Suppose we knew each of these to be true. Would our knowledge likely be a priori or a posteriori? Take terms in their most natural senses. Some examples are controversial.
All triangles are triangles.
This would be known a priori.
Use the examples from §3.6a.