What the Receiver Rebuilds

Two registrars take over bed seven on different mornings. The handover note is the same on both days: keep an eye on him. The first registrar covered the ward last week. She knows that this patient’s observations drift before he deteriorates, that he under-reports pain, and that “keep an eye” from this colleague means check him within the hour. She reads four words and rebuilds a whole plan. The second registrar is new to the ward. She reads the same four words and rebuilds almost nothing: a patient, some mild vigilance, no particular deadline. Both read the sentence correctly. Neither misread it. They recovered different things.

The representation was identical. The meaning each person ended up with was not, and the difference came almost entirely from what each already held. This chapter is about that difference. Its claim is simple to state and awkward to absorb:

Receiving a representation is not the same as receiving its meaning. The receiver builds the meaning, and what they can build depends on what they bring.

representation  ≠  meaning
sent  ≠  understood

Everything the book later asks of a generated representation (a diagram, a brief, a timeline, an agent’s message) will be judged by this standard. The question is never only whether the output looks right. It is what a particular person can rebuild from it.

The pipe, and why it fails

A powerful picture of communication sits underneath most everyday talk about messages. A sender puts meaning into words, the words travel, the receiver takes the meaning out. Claude Shannon’s mathematical theory of communication (1948) gave transmission its engineering form: a message selected at one point is reproduced at another, with noise as the enemy. It is an extraordinary piece of engineering, and Shannon himself was careful to set meaning aside as irrelevant to the problem he was solving. A channel can transmit a signal perfectly without establishing that a receiver reconstructed what the sender meant.

Warren Weaver, introducing the book version of the theory in 1949, drew the line more explicitly. There are, he wrote, three levels of problem. How accurately can the symbols be transmitted? How precisely do the transmitted symbols convey the desired meaning? How effectively does the received meaning affect conduct in the desired way? Shannon’s theory answers the first. The second and third are where human communication actually succeeds or fails, and they are the levels this book lives on.

The trouble starts when the pipe is mistaken for a theory of understanding. If meaning travels inside words, then understanding is decoding, and a clearer message is simply one with less noise. But no sentence fixes its own meaning. Someone says “it’s cold in here” and might be reporting a temperature, asking for the window to be closed, or complaining about a colleague. The philosopher Paul Grice drew attention to this gap between what is said and what is meant, and to the way listeners cross it by reasoning about the speaker: why would they say this, now, in this way? Dan Sperber and Deirdre Wilson built on the idea in Relevance (1986). On their account a speaker does not hand over a thought. A speaker provides evidence of what they intend, and the listener infers the most relevant interpretation consistent with that evidence, given everything else they know and expect.

On this view, understanding is an act of inference by the receiver. That is what the registrar did with four words.

What a receiver brings

It helps to separate what the receiver contributes, because a system that wants to help communication will eventually have to reason about each part.

Prior knowledge. The first registrar knew the patient. The second knew medicine in general but not him. Knowledge decides which gaps in a representation can be filled and which remain gaps.

Shared context. Herbert Clark called the mutual knowledge two people rely on common ground, and showed how much of ordinary conversation is built on it: what has been said, what is visible, what each believes the other knows. “The usual place” works between friends and fails between strangers. A representation that is complete inside one common ground is cryptic outside it.

Purpose. People read for something: to decide, to check, to learn, to act. The same account of an outage is read differently by the engineer who must approve the fix and the student who wants to understand distributed systems, and what counts as having understood differs with it.

Condition. Time, attention and fatigue. A reader at the end of a night shift rebuilds less, from the same page, than the same reader rested.

None of these is a flaw in the receiver. They are the materials reconstruction is made from. Two receivers with different materials will build different things from the same representation, and both can be doing their job well.

A model, not a copy

What the receiver builds is not a transcript. Research on discourse comprehension has long distinguished the words of a text from the mental model a reader constructs of the situation it describes. Walter Kintsch’s construction–integration account (1988) describes comprehension as building a network from the text and the reader’s knowledge together, then settling on the interpretation that fits. Zwaan and Radvansky (1998) reviewed the evidence for situation models: readers track people, places, goals and causes, not sentences, and what they remember is largely that model. The practical consequence is that you can only learn what someone understood by asking what their model lets them do: answer a question, spot an error, apply the idea to a new case.

An experiment from the same tradition shows how much the receiver’s knowledge matters. McNamara, Kintsch and colleagues (1996) gave readers texts that were either highly coherent, with every connection spelled out, or less so. Readers who knew little about the topic understood more from the coherent text, as one would expect. Readers who knew a lot understood more from the less coherent one. The gaps forced them to do the integrating themselves, and that work built a better model. There is no best text in the abstract. There is a best text for this receiver, for this purpose.

The receiver is a poor judge of their own reconstruction, and the sender is a poor judge of the receiver

Two familiar errors make the whole process harder to see from the inside.

Receivers often feel they have understood when they have not. Rozenblit and Keil (2002) asked people to rate how well they understood everyday devices such as zippers and flush toilets, then asked them to explain how those devices work. Ratings fell sharply after the attempt: the feeling of understanding had been running ahead of the understanding itself. Fluent, well-formatted material makes the feeling stronger. This is why a representation that readers love is not thereby one that taught them anything.

Senders, for their part, tend to overestimate how much of their intent has crossed. Keysar and Henly (2002) found that speakers believed their ambiguous statements had been understood more often than they were, because the speaker’s own knowledge of what they meant fills the ambiguity automatically. A well-known demonstration, from a doctoral study by Elizabeth Newton (1990), asked people to tap out the rhythm of a famous song for a listener to identify. Tappers estimated that about half of listeners would identify the tune. In Newton’s dissertation, listeners actually identified two of 150. For the tapper the song plays in their head. For the listener there is only knocking. The tapper has access to a richer experienced whole than the tapping carries. Speakers are in a similar position: they may begin with an image, a relation, a direction or a partly formed thought, not a finished inner sentence, and the words they choose carry only some of it. The gap between what the sender has and what the representation carries is part of what the receiver has to rebuild across; no general theory of wordless thought is needed to see that.

Put the two together and a plain lesson emerges for anyone designing communication, human or machine. Neither the sender’s confidence nor the receiver’s sense of having understood is evidence that the meaning arrived. The only evidence is what the receiver can then do.

A target for everything that follows

This gives the book a measurement target it can carry from here to the last chapter. Whenever a source is turned into a representation, four things can be asked of what the receiver rebuilds from it:

SOURCE → REPRESENTATION → RECEIVER'S RECONSTRUCTION
                          ├─ what survived?
                          ├─ what changed?
                          ├─ what was added?
                          └─ what became easier?

The first three are the losses and distortions already named. The fourth is what makes the audit two-sided. A representation can improve communication: by making a relationship visible that prose buried, by cutting the work the receiver has to do, by putting the decisive distinction first, or by choosing a form that suits the task. That is gain, and it means an improvement in the communicative outcome. It does not mean new information.

GAIN  ≠  ADDITION

A diagram that invents a link the source never made has not gained anything; it has added something unsupported. A diagram that lets a reader see, in ten seconds, a cycle that took three paragraphs to describe has gained.

Notice what this does to the idea of a good representation. It cannot be defined by the representation alone, because the same one can be excellent for one receiver and useless for another. It has to be defined by the relationship between the representation, a receiver and a purpose. That is exactly why a system that sits in the channel and chooses representations will sooner or later have to form some picture of the receiver: of what they already know, what they are trying to do, and what state they are in. The later parts of this book are largely about how far that picture may go, and who controls it.

What can and cannot be claimed

A note on limits, because the territory invites overreach. Reconstruction is not directly observable. What can be observed are its products: answers, decisions, errors, time taken, and what a person can do with the material a day later. The book’s vocabulary of source, representation and reconstruction is a working model for engineering and for experiment. It is not a theory of how minds encode meaning, and nothing here claims to have discovered one. The research cited in this chapter comes from psycholinguistics, philosophy of language and the psychology of comprehension, and each field’s findings are about the settings it studied. Where the book uses them it uses them as grounds for caution, not as proof of a mechanism.

Where this leaves us

A representation does not contain its meaning. It is evidence that a receiver uses, with their own knowledge, context, purpose and condition, to build a model. What counts as a good representation therefore depends on the receiver as much as on the words, and neither side is a reliable judge of how well the building went. For any transformation, the questions to ask are what survived, what changed, what was added and what became easier, as seen from what the receiver can do afterwards.

If the receiver rebuilds from whatever channels are available, then the channels matter. A timetable, a subway map and a spoken warning give a receiver very different things to build from, and the next question is what different kinds of representation make possible.

References

• Shannon, C.E. (1948). A Mathematical Theory of Communication. Bell System Technical Journal, 27, 379–423, 623–656.
• Shannon, C.E. & Weaver, W. (1949). The Mathematical Theory of Communication. Urbana: University of Illinois Press. (Weaver’s introductory essay distinguishes the technical, semantic and effectiveness problems.)
• Grice, H.P. (1975). Logic and Conversation. In P. Cole & J. Morgan (eds.), Syntax and Semantics 3: Speech Acts. New York: Academic Press, pp. 41–58.
• Sperber, D. & Wilson, D. (1986). Relevance: Communication and Cognition. Oxford: Blackwell. (2nd ed. 1995.)
• Clark, H.H. (1996). Using Language. Cambridge: Cambridge University Press.
• Kintsch, W. (1988). The role of knowledge in discourse comprehension: A construction-integration model. Psychological Review, 95(2), 163–182.
• Zwaan, R.A. & Radvansky, G.A. (1998). Situation models in language comprehension and memory. Psychological Bulletin, 123(2), 162–185.
• McNamara, D.S., Kintsch, E., Songer, N.B. & Kintsch, W. (1996). Are good texts always better? Interactions of text coherence, background knowledge, and levels of understanding in learning from text. Cognition and Instruction, 14(1), 1–43.
• Rozenblit, L. & Keil, F. (2002). The misunderstood limits of folk science: an illusion of explanatory depth. Cognitive Science, 26(5), 521–562.
• Keysar, B. & Henly, A.S. (2002). Speakers’ overestimation of their effectiveness. Psychological Science, 13(3), 207–212.
• Newton, E.L. (1990). The Rocky Road from Actions to Intentions. Doctoral dissertation, Stanford University. (Widely cited under the title “Overconfidence in the communication of intent: Heard and unheard melodies”; the secondary-report figure of 3 in 120 is not used here.)