What you will get from this article

The answer first. The previous article rejected the claim that language determines thought. So what is thought made of? Pinker’s answer: people think not in English, Chinese, or Apache but in a “language of thought” (Mentalese). To know a language is to know how to translate Mentalese into strings of words and back.

People do not think in English or Chinese or Apache; they think in a language of thought.

People who think without words (pp. 67–70)

First, cases of thinking without language.

A languageless adult. In Susan Schaller’s account, Ildefonso, a twenty-seven-year-old illegal immigrant from a small Mexican village, had no sign language, writing, lip-reading, or speech. Yet he was visibly intelligent; with Schaller’s teaching he learned to add on paper in three minutes and grasped the logic of base-ten numbers. Taught the sign for “cat,” he had an epiphany like Helen Keller’s that things have names, and demanded signs for everything he knew. Soon he could tell, in pantomime, how as a child he had begged his poor parents to send him to school (pp. 67–68).

Pinker notes that for ethical reasons the mental life of the languageless cannot be probed (when found, the priority is to teach them language), and gives three other kinds of beings who think without language: babies, monkeys, and adults who say their best thinking is done without words.

Babies (pp. 68–69). The psychologist Karen Wynn showed that five-month-old babies can do simple arithmetic. Babies get bored with a familiar scene and look away; change the scene, and if they notice the difference they regain interest. By this method even five-day-olds have been shown to be sensitive to number.

The babies must have been tracking the number of dolls behind the screen, updating it as dolls were added or subtracted, says Pinker.

Monkeys (pp. 69–70). Vervet monkeys in Kenya live in stable groups where extended families form alliances. When one juvenile wrestled another to the ground screaming, twenty minutes later the victim’s sister approached the perpetrator’s sister and bit her on the tail without provocation. This retaliation requires solving an analogy problem using relations like “sister of.” In a playback experiment, females looked at the mother of an infant whose scream was played, showing they recognized not only the infant but recalled who its mother was.

Creative adults (pp. 70–72). Many creative people say that at their most inspired they think not in words but in mental images: Coleridge’s “Kubla Khan,” Faraday, Maxwell, Tesla, Kekulé (the benzene ring), Watson and Crick, Einstein imagining riding a beam of light.

The mental-rotation experiment (pp. 71–73)

Is the introspection of thinking in images accurate? The psychologists Shepard and Cooper’s experiment checked (pp. 71–73).

If subjects had been handling verbal descriptions (“an upright spine with one horizontal segment extending rightwards from the top”), the results would have been very different, says Pinker. With verbal descriptions, the upside-down (180 degree) version would be easiest, since one simply swaps top and bottom, left and right. In fact it was slowest (pp. 72–73). Visual thinking uses not language but a mental graphics system with operations that rotate, scan, zoom, pan, and displace.

How can “symbols in the head” work? (pp. 73–78)

What does it mean that images, numbers, kinship relations, or logic can be represented in the brain without being couched in words? In the first half of the twentieth century philosophers thought that to reify thoughts as things in the head was a logical error: a picture in the head would need a little man (homunculus) to look at it, and he would need smaller pictures and a smaller man.

What made the idea scientifically respectable was Alan Turing. Pinker explains reasoning like this.

Let a machine carry out “Socrates is a man; all men are mortal; therefore Socrates is mortal.”

  1. Representation. Write ink-mark patterns on a page: “Socrates isa man” and “Every man ismortal.” “Socrates” is the symbol for the concept of Socrates, “isa” for being an instance of, “man” for the concept of man. The shapes and positions of the marks are conventions (a star or a smiley would do, as long as used consistently) (p. 74).
  2. Processor. Not a little man but a stupid gadget with a fixed number of reflexes. For example, on spotting the shape “isa” it copies the marks to its left, then to its right; it scans for “Every” and checks the following “man” for a match; if matched, it copies the “ismortal” that follows (pp. 75–76).
  3. The page now shows “Socrates ismortal.”

Neither gadget nor page is itself intelligent; they are paper, ink, cutouts, photocells. What is smart is the correspondence between the logician’s rule and the way the device scans, moves, and prints. Blindly following physics, the device responds to the shape of the marks without understanding and mimics the logical rule (p. 76).

Pinker says the same principle holds in the brain (p. 77). Groups of neurons might represent individuals (Socrates, Aristotle), relations (is a, is like), and kinds (men, dogs), and a network produces one group’s firing pattern from another’s; it could be done in silicon too. This is the physical symbol system hypothesis, the “computational” theory of mind (pp. 77–78). Pinker calls it as fundamental to cognitive science as the cell doctrine is to biology and plate tectonics to geology (p. 78).

But there are ground rules. No little man inside, no peeking. Representations are arrangements of symbols and the processor is a device with a fixed set of reflexes. The theorist is forbidden to “read” the symbols and “make sense” of them, like some deus ex machina (p. 78).

Why English is a poor medium of thought (pp. 78–82)

Now Whorf’s question can be posed precisely. Do internal representations in an English speaker’s mind look like English, or do they use some other code? Pinker’s answer: English is hopelessly unsuited as our internal medium (p. 78). There are five problems.

ProblemContentExample
1. AmbiguityOne word or sentence corresponds to several thoughtsReal headlines: Child’s Stool Great for Use in Garden (stool as seat or as feces), Stud Tires Out, Iraqi Head Seeks Arms
2. Lack of logical explicitnessEnglish sentences do not carry the information inference needs“Ralph is an elephant. Elephants live in Africa. Elephants have tusks.” → “Ralph lives in Africa; Ralph has tusks.” A human knows the tusks are Ralph’s own, not those of other elephants; a mark-copying gadget does not
3. Co-referenceThe same person is referred to with different words“the tall blond man with one black shoe” → “the man” → “him.” Three expressions, one person. The brain must treat them as the same, but English does not
4. DeixisMeaning is not fixed without contextkilled a policeman vs. killed the policeman (the definite article presumes a policeman already mentioned); here, there, I, you, now, then likewise
5. SynonymyThe same content in different word ordersSam sprayed paint onto the wall / Sam sprayed the wall with paint / Paint was sprayed onto the wall by Sam / The wall was sprayed with paint by Sam. Four arrangements of one event; a mechanical device would not know they are the same

For the deixis point Pinker uses an old joke: “I didn’t sleep with my wife before we were married, did you?” “I don’t know. What was her maiden name?” (p. 80). It shows that without context, meaning is not fixed.

On the fifth, Pinker says the event could be represented without relying on English words, something like ⟨(Sam spray paint) cause (paint go to (on wall))⟩, which is not far from one leading proposal about what Mentalese looks like (p. 81).

Language and thought have different purposes (p. 81)

Another reason is difference of purpose.

Pinker’s conclusion (pp. 81–82)

Compared with any given language, Mentalese is richer in some ways and simpler in others.

Knowing a language, then, is knowing how to translate mentalese into strings of words and vice versa.

People without a language, babies, and many nonhuman animals presumably have Mentalese. If babies did not have a Mentalese to translate to and from English, Pinker says, it would not be clear how learning English could take place or even what it would mean.

The 2050 prediction

Pinker ends the chapter with predictions for Newspeak in 2050.

  1. Since mental life goes on independently of particular languages, concepts of freedom and equality will be thinkable even if nameless.
  2. Since there are far more concepts than words, and listeners always charitably fill in, existing words will gain new senses and perhaps even regain their original ones.
  3. Since children will not reproduce input but create a complex grammar that goes beyond it, they would creolize Newspeak into a natural language, possibly in a single generation. The twenty-first-century toddler may be Winston Smith’s revenge.

(This applies the creolization argument of article 295 to Newspeak.)

Cautions

Next