NOTE-361: Jokes and the Logic of the Cognitive Unconscious

Contribution

The memo extends Freud’s censor theory of jokes from emotional taboos to intellectual failures. It puts nonsense, paradox and wordplay under one mechanism with sexual and aggressive humour. That mechanism is a learned store of “negative meta-knowledge”, censors that suppress known-unproductive states of mind. The memo grounds the store in an argument from the frame memo: commonsense logic cannot be made safe, and general methods cannot be patched for particular failures.

Key insight

Minds get smarter as much by learning what not to think as by learning what to think. A general method that fails in a particular case cannot be fixed without breaking it elsewhere. All one can do is remember to suppress it there. Humour is the social and internal channel for these subtractive lessons. It explains why jokes are about the prohibited and the absurd, and why they wear out.

Assumptions

• Two brains. An “A-brain” faces the world. A “B-brain” takes its inputs from A’s interior and can affect A, recognising “metapsychological” conditions such as meandering, circling or confusion without knowing what A’s states mean.
• Frames. Perception and understanding use frames with terminals and weak defaults, after the frame memo. Commonsense inference is shifting between frames that share terminals (§III).
• Society of Mind. The mind is “composed of many smaller minds” (Note 1). Partial mental states follow K-lines (Note 2).
• Censors learn, at different rates. Censors for different subjects have different time-constants (§V, “Time constants”).
• Method. The author would “experiment, not on people … but on computers to see if a proposed theory-fragment can be made part of a system that shows mind-like activity” (Note 16). No such experiment is reported.

Key results

These are conjectures, argued by example.

• Five-part thesis (pp. 2–3).
  1. Commonsense logic is unreliable and unrepairable.
  2. Errors must be anticipated and prevented by censors, hence humour’s concern with the prohibited.
  3. Analogy and metaphor must be policed, hence its concern with nonsense.
  4. Faulty reasoning is as “naughty” to the cognitive unconscious as Freudian wishes.
  5. Humour evolved socially, as a disarming way to correct others.
• Problems of commonsense reasoning (§I).
  • The liar and the barber show that no logic barring self-reference is adequate, since “without the power to think about Thinking”, no mind is powerful enough.
  • The response is to find “islands of power” and mark their unsafe boundaries. Paradigms like the Barber, the Tortoise and the Liar are road-signs to “stop — and laugh”.
• Censorship (§II).
  • A type (i) censor stops a thought once it has started. A type (ii) censor prevents it, by learning its precursors.
  • Precursor-sensitive censors act quickly enough to leave no phenomenology.
  • “This ‘negative meta-knowledge’ … may be a large portion of all we know.”
• Meaning and metaphor (§III).
  • The frame theory is restated, including the invisible chair under a sitting person.
  • Logic is cast as frame-shifting, and the “game” problem as a system of prototype frames.
  • Particular analogies are recorded additively, as difference-network links after Winston. General analogy methods are corrected subtractively, by censors: “positive general principles need always to be supplemented by negative, anecdotal censors.”
  • “Unexpected frame-substitution” is the element common to humour.
• Traumatic cognitive experiences (§IV).
  • The nearness paradox (“Ten is almost a Hundred”), Zeno, and chicken-and-egg act as intellectual traumas that build censors and cognitive phobias.
  • “Mystical experience” is a last-resort short-circuit that creates the illusion a problem is settled.
• Humour and evolution (§V).
  • Laughter disrupts reasoning and focuses attention on the absurd thought, so that censors can learn it, “much as ‘pleasure’ is often supposed to mediate ordinary learning”.
  • A social origin is offered for its mixed conciliatory and aggressive display, later internalised.
  • “Humor” names a web, not a thing: “less a Thing-Part of the mind, and more a Thing-Theory in the mind”.
  • The persistence of sexual humour is put down to slow censors, with attachment and mourning given as examples of long time-constants.
• Notes.
  • Note 8: ethnic humour as a side-effect of the need for stereotyped person-frames.
  • Note 12: pleasure as a “constructed illusion” of coherence, with a conjectured central brain unit.
  • Note 17: an “acquisition envelope” hypothesis, in which puberty shuts off phonetic learning so that children learn the parents’ language, not the reverse.

Claims

id	claim	strength	support
C1	Commonsense reasoning cannot be made reliable; minds must instead learn where it fails	moderate	argued from the paradoxes and the logistic critique of the frame memo; not shown impossible
C2	Errors of general methods can only be corrected subtractively, by censors, not by modifying the method	moderate as a design argument	”it hardly ever pays to alter a general mechanism to correct a particular bug” (§III); the child-and-toy example
C3	Censors that act on precursors work too fast to be noticed, which is why they are unconscious	weak	inference from the type (ii) design; no evidence
C4	The common element across forms of humour is unexpected frame-substitution	weak	examples (cake, mayonnaise, bird-hunting dog, puns); Note 16 admits that “good humor” with good frame-shifts strains the positive/negative distinction
C5	Laughter’s function is to disrupt the absurd thought and hold it in focus for censor-learning	weak	conjecture (§V); no data on laughter
C6	The affect–intellect distinction does more harm than good in psychology	assertion	argued by parallel between cognitive and emotional “traumas”
C7	Old jokes are less funny because a censor has been built; sexual jokes persist because their censors learn slowly	weak	conjecture; the slow-censor premise is introduced to save the first claim from the counterexample
C8	Puberty genetically shuts off phonetic learning so that children, not parents, adopt the common language (Note 17)	weak	conjecture from the commonplace observation of adult accents

Concepts

• Censor. A learned suppressor of a mental activity. Type (i) stops it; type (ii) recognises its precursors and prevents it.
• A-brain and B-brain. A two-level mind in which B perceives and acts on A’s internal states, as A does on the world.
• Negative meta-knowledge. Knowledge about patterns of thought and inference that have been found defective or harmful.
• Islands of power. Domains within which naive reasoning stays safe.
• Frame-substitution. Re-viewing all the elements of a scene through a different frame.
• Cognitive unconscious. The unconscious working of censors on intellectual, not emotional, matters.
• Acquisition envelope. A genetically controlled window on how something is learned (Note 17).

Connections

• Frames. The memo extends the frame memo (LIT-410). §III’s account of frames, defaults and frame-shifts is the base of the humour theory. It cites the frame memo’s discussion of logic’s limits (Note 4), and its “invisible chair” default is the frame memo’s §1.11 example.
• K-lines and Plain Talk. It cites K-lines (LIT-402) for partial mental states (Note 2) and for memories not all “made of the same stuff”. It cites Plain Talk (LIT-399) for ambiguous thoughts and for scarce “central” mechanisms (Notes 3, 7).
• Freud. Jokes and Their Relation to the Unconscious (1905) and “Humour” (1927), both quoted at length. Note 11 argues Freud spoiled the mayonnaise joke by moving the absurdity.
• Other named sources. Doyle’s truth maintenance and deliberation thesis; Kornfeld; Korzybski; Wittgenstein on “game”; Winston’s difference networks; Papert, Sussman and Goldstein on bugs; Lewis Thomas’s “scrambler in the brain”; Lorenz and Tinbergen on displacement.

Bearing on the record

• Somatic markers. The somatic-marker hypothesis (LIT-384) and the gambling-task study (LIT-383) describe learned, nonconscious signals that prune options before reasons are available. Censors are a cognitive analogue that works on lines of thought, not on options for action. The two together make the record’s case against a clean affect–intellect split from two directions, neural and computational. The pairing is mine. Neither work cites the other, and the memo offers no neural mechanism.
• Logic. The memo treats paradox as a developmental event and argues that no logic forbidding self-reference is adequate for psychology. That is a claim about minds, not about formal systems, and it does not bear on the logic papers held here.
• THEORY. No THEORY in the record should cite it as support.
• Boundary. No machine-learning practice instruction; no anthology topic holds it.

Limitations

• No evidence. There are no experiments on humour, laughter or learning, and no program.
• The unifying claim (C4) bends. Note 16 concedes that adult “good humour” uses good frame-shifts as well as bad ones, so the positive/negative distinction “seems to fall apart”. The memo saves it by appeal to adult development, not by evidence.
• Ad hoc time-constants. The “slow censor” premise is introduced to absorb the counterexample of sexual humour (C7).
• Untested side theories. The notes add hypotheses (ethnic humour, pleasure, phonetic acquisition) that are independent of the main argument and equally untested.

Open questions

• Can a system with a B-brain that learns precursor censors be built, and does it show anything like humour’s wear-out with repetition? This is the test Note 16 itself proposes.
• Does the funniness of a joke track whether its frame-shift exposes a “bug”, as C4 and C5 would predict, more than it tracks surprise in general?

Corrections

• none to a seeded skim (there was no seed).
• Title. The brief’s title is wrong. The memo and DSpace read “Jokes and the Logic of the Cognitive Unconscious”. “Jokes and their Relation to the … Unconscious” is Freud’s 1905 title, which this memo quotes. Memo number (603) and year (1980) are right; the cover month is November 1980.
• The published chapter’s date. Crossref’s record for the chapter (DOI 10.1007/978-94-010-9188-6_10) gives 1980. Its record for the book (DOI 10.1007/978-94-010-9188-6) gives 1984, as do the catalogues. The chapter appeared in 1984; 1980 is the memo’s year.