The Key to Understanding Others is to Be a Good Detective, Not a Code-Breaker

The Big Picture

The purpose of this site is to help you achieve your own goals by improving your communication skills. Better personal relationships. More success at work. Deeper connections with your parents, kids, spouses, friends, and lovers.

As in the previous post, I will give you tips, techniques, and strategies for handling all sorts of common communication tasks. But more than that I want to give you the big picture, to teach you how communication works in general. In the long run, you will rely less on tips and techniques. Instead, you will use your in depth understanding of communication and the social world to analyze and solve even the trickiest communication problems.

You need a new vocabulary for thinking about communication, the social world, and your place in it. Social interactions may seem endlessly variable, each one presenting its own unique challenges. But underneath that apparent variety, fundamental processes are always at work, structuring our interactions and allowing us to understand (or misunderstand) one another.

As I teach you how to comfort friends in need, how to be more persuasive, how to use just the right about of directness, or just the right amount of politeness, in any given situation, you will build up a a detailed map of the social world.

As you master the basics (e.g., cooperation, inference, directness and indirectness, face and politeness, identities and interactions, turn-taking, etc.), you will use these concepts to analyze and then optimize your approach any communication task.

The first thing to do to get to your final destination—mastery of the social world—is not to head off in the wrong direction. Unfortunately, when we learned about communication, many of us were taught a very simple and totally unhelpful model. So that's where we begin, by trying to get you to let go of any bad ideas you already have about how communication works. 

The Code Model (Sender-Message-Receiver)

One of the biggest barriers to understanding how communication works is the belief in a simple and incorrect model of language and communication. This model states that language is a code, and that communication is a process of a sender encoding an idea into language, transmitting across some (potentially noisy) medium, where the receiver then decodes the message and extracts the idea. When this process is complete, successful communication is said to have occurred.

This model, known variously as the Code Model or the Sender-Message-Receiver model, dominates popular consciousness and most non-scientific explanations of communication. A quick search on “models of communication” turns up dozens of images describing the Code Model, most of them derived from Shannon and Weaver’s original model, which was developed to explain technical processes in electronics and information theory, not to explain ordinary human interaction.

                                             Shannon and Weaver's "code model" of communication.

                                            Shannon and Weaver's "code model" of communication.

The sender-message-receiver Code Model is so deeply ingrained in our consciousness that it is hard to think any other way. The model seems to be consistent with some of our intuitions about language, and it certainly is consistent with the way we talk about talk (e.g., “I just need to put my thoughts into words.”) But here's the rub: almost no working communication scientists endorse the Code Model any more.

Real Codes vs. Human Language

To understand how human languages differ from codes, it is helpful to think about codes and then to examine how they may be similar to and different from human languages. The most obvious example of a true code is Morse Code.

Morse Code maps each letter of the English alphabet to a series of dots and dashes (or “dits” and “dahs”). For example, the letter /s/ in Morse code is dit dit dit (· · ·), and the letter /o/ is dah dah dah (­– – –). In a true code, the relationship between the symbol and the thing it represents is constant.  In Morse code, dit dit dit always symbolizes the letter /s/. It never means anything else, and no other series of dits and dahs means /s/. Regardless of the context, or the rate, or the volume at which the dits and dahs are transmitted, they always mean the same thing. Scream dit dit dit at the top of your lungs, and it means /s/. Whisper it in your sexiest bedroom voice, and it still means /s/. Say it sarcastically. It still means /s/. That’s a true code. Symbols have fixed meaning, and meaning doesn't depend on context.

Multiple Meanings

Here we have an obvious difference between true codes and human languages. At the simplest level, words have multiple meanings. The word “fine” means okay or all right, as in “I’m feeling fine.” It also means having delicate structure, like a “fine-toothed comb,” and it means a monetary penalty, as in “the fine for parking in the handicapped spot was $250.”

Meaning Depends on Context

More importantly, meaning in human languages are context dependent, and things like rate, volume, pitch, and intonation of speech can dramatically alter the meaning of the same word. Consider how the word “fine” changes its meaning when said matter of factly, as when you meet a colleague in the morning and they say “fine” when you ask how they are doing today, versus what it means when you ask a teenager to put away their phone, and they say, in an exasperated tone, “fine!”.

These examples should make us start to question whether and to what extent language is like a true code. Some will still insist it is a code, but it starts to look like a very peculiar code, one where the symbols in the code have multiple meanings and mean different things depending on the context.

We Mean More than We Say

Another defining characteristic of human communication is that we almost always mean more than we say. This is true even when we don't intend it to be, because, whether we like it or not, people will make inferences far beyond the 'literal meaning' of what we say.

Consider a letter of recommendation that says about the job applicant only that “She was always on time and had good personal hygiene.” Taken literally, there is nothing at all bad contained in this statement. But any competent adult communicator recognizes immediately that this is terrible letter of recommendation. Somehow what is not said says more than what is said. By failing to note any of the applicant’s other positive characteristics, we infer that perhaps there was nothing else good to say about her.

How do we know to make this inference? If language is a code, then all the meaning should be “in the words,” ready to be decoded. What would it mean to decode a sentence like the one above? The meaning does not appear to be in the words at all but rather in our knowledge of the world, including what is the appropriate amount of content to include in a recommendation letter. And how do we know or suspect that, in this particular context, we should not rely on the explicit, literal meaning of what is said but should instead rely on our knowledge of the world to make an inference beyond what is said to grasp what is really meant?

The answer lies not in language but in a set of shared assumptions we make about one another when we communicate. We assume that the other person is being cooperative, not in the ordinary sense of helping us unload the groceries, but in a more technical sense. We assume that they are reasonable people who say relevant things when they speak. So when something they say appears, on its surface, not to be relevant (or complete, or true, or straightforward), then we assume, since we know they are being cooperative, that finding their true meaning, the one that is relevant, must require us to make some inferences.

And so we search for the inference that allows their statement to appear cooperative. The so-called cooperative principle is another bedrock idea in communication theory, and I will be writing an entire blog post soon to discuss it. For now, I am just introducing it to explain how we know when to make inferences beyond the information given.


Another challenge to the code model is that what we mean is often only indirectly related to what we say. I remember in college when my friend Kevin used to come into a room when we were eating pizza, and he would say “Is that pizza?”

Of course, the literal answer to the question was “Yes, Kevin, that’s pizza.”

But that answer would be heard as rude and uncooperative. Any competent communicator hears Kevin's question as a request for a piece of pizza. And it’s not just that one sentence, lots of others would have the same effect: “Did you guys get pizza?” “Are you going to eat that?” “That looks delicious.” “I’m starving.” “I love pepperoni.” All these would easily and obviously be heard as requests.

But how? Again, the meaning of the utterance does not seem to be in the words, but rather in our knowledge of the world. In addition to the meanings of words, we seem also to know complicated rules about the social world, such as “when a person says they are starving, and there is pizza on the table, then they are really requesting to have a piece of pizza."

In fact we do know such rules, but they are not specifically about pizza. Instead we know elaborate rules about how to do things with words like make requests, promises, commands, threats, compliments, and apologies. Each of these speech acts, as communication scholars call them, has a set of rules for their proper execution. We learn these rules. We then assume others know these rules, and we take advantage of this knowledge to communicate indirectly, and often strategically.

We will learn about indirectness and about speech acts in upcoming blog posts. For now, it is enough to just recognize that we can, and almost always do, mean more than we say, and that in part this ability is made possible by our knowledge of rules that govern the social world.

The challenge for the code model is that we talk like this all the time. We constantly say one thing and mean another. Literal communication, if it occurs at all, is much rarer than indirect communication, for reasons I will explain in subsequent posts (e.g., indirect allows us to use language strategically, to save face, to be polite, and to avoid accountability for what we mean). 

Perhaps we should begin to take more seriously the idea that it is not a code at all. We need a better metaphor.

General Principles Enable Inference, not Millions of Rules

It cannot be that we have memorized millions of rules, one for every imaginable social situation, rules that allow us to know how to make the right inference about whether to take someone literally or whether to make an inference about what they really mean.

Lucky for us, there are not millions of coding and decoding rules to remember. That code metaphor does not work to capture the complexity, power, and beauty of human language and communication.

Instead, it is more useful to think of communication as solving problems or mysteries. When you think of communication, don't think of yourself as Benedict Cumberbatch solving the Enigma encryption machine in the The Imitation Game. Instead, think of yourself as Benedict Cumberbatch solving mysteries as Sherlock Holmes.

The mystery is always the same: what are the other person's beliefs, emotions, attitudes, plans, goals, and intentions? For clues, you have the words people say, as well as their nonverbal behavior (e.g., posture, gesture, facial expression, personal space and distance, pitch, rate, and volume). You also have your common-sense knowledge of the world (e.g., what is pizza? how do requests work?), and you have your knowledge of the particular social context and the specific other people involved. 

This is the main take-home point: Communication is primarily about inference, not about decoding.

What Do You Mean When You Say "Inference"?

By inference, I mean reasoning from evidence to conclusion, the way Sherlock Holmes takes a clue and uses his powers of deduction to arrive at the solution to the mystery. The main difference between inference and decoding is that in decoding the information is assumed to be inside the message. Whereas, in inference the message is just a clue, and its meaning must be constructed using our powers of reasoning applied to this clue combined with of our other knowledge of the world. Because it relies on our unique powers of reasoning and our idiosyncratic knowledge of the world, inference is constructive. We don't extract the meaning of the message. We construct it. The meaning of a message is a kind of story we tell ourselves about how the world must be, and what the other person must be thinking, given that she just said what she said. But the story was not in the message. We made up the story based on the message.

The constructive nature of inference helps us understand why it is so difficult to clearly communicate our plans, goals, and intentions. When we talk, we are just offering clues. The people we are talking to are the detectives, using their powers of reasoning to try and solve the mystery of our intent based only on scanty clues we've provided them. You only need to watch a few episodes of Law & Order to realize that the same evidence can support different conclusions. So we shouldn't be surprised when people draw different conclusions than we might have intended. Their reasoning process may be different from ours, or their knowledge of the world may be different, or both. If either of these is true, then they are likely to arrive at different conclusions than we intended. 

Inference depends on general knowledge of the social world, especially knowing when to make inferences beyond the information given, and also on specific knowledge of a situation shared between you and the person you are communicating with. The more knowledge we share, the less we have to actually say anything. With the people we know best, we hardly need to speak at all. Just a look or a wink or nod can communicate an enormous amount of information, all via inference based on mutual knowledge. Like magic.

One implication of this way of thinking about communication is that mutual understanding depends critically on common knowledge. If you've ever been in a culture very different from your own, you'll  know how easy it is to be misunderstood. 

By thinking about communication as problem solving or as detective work, it also gives us a new way of thinking about the challenge of deciding what to say. The challenge is not to "put your ideas into words" but rather to figure out what to say or do, in this specific context, with this specific person, and these assumptions about the knowledge that we share, that will lead them to make the inferences I want them to make about my plans goals, and intentions.

So this is the first building block in our new understanding of how communication works. Communication is problem solving, detective work, supported by inferences based on knowledge shared by the communicators. The inferential nature of communication allows us to communicate indirectly, to mean much more than we say, and this ability enables us to use language strategically to manage relationships and achieve our goals. More about this soon.

Tell me about how you use indirectness.

If you like this, please share. To learn more about how communication works, subscribe to my mailing list.