Ladder: In 1939, linguist S. I. Hayakawa published Language in Action, later revised as Language in Thought and Action, introducing a model that remains useful in communication, rhetoric, and organizational management.
The model describes a straightforward reality: words do not all sit at the same distance from observable things. Some words, such as "Bessie" or "this cow right here in front of me," are grounded in the specific. Others, such as "livestock," "assets," or "wealth," are progressively more general, each level adding conceptual distance from anything you can point at.
Hayakawa called this hierarchy the ladder of abstraction.[1] His point was that the level you choose is not neutral: different levels serve different purposes, carry different risks, and produce different kinds of understanding.
Moving up without grounding produces statements that sound significant but give no traction. Staying down without interpretation produces detail with no organising pattern. The practical skill is moving between levels according to what the audience needs, not descending whenever possible.
Alfred Korzybski made a related point in Science and Sanity (1933) with the phrase most associated with him, that the map is not the territory: descriptions and models are not identical to the realities they represent.[2]
The Ladder Itself
Hayakawa illustrated the levels with a cow named Bessie. The table below is adapted from that example rather than reproduced verbatim from a specific edition.
| Level | Example | What it does |
|---|---|---|
| Most concrete | "Bessie right now, in this field" | The actual thing at a specific moment |
| Very concrete | "Bessie" | This particular cow as an individual |
| Concrete | "The Brown family's cow" | One of potentially several animals owned |
| Mid-level | "Cow" | A class of similar animals |
| More abstract | "Livestock" | A broader category with economic uses |
| Abstract | "Farm assets" | Includes equipment, land, buildings |
| Very abstract | "Assets" | Any property with financial value |
| Most abstract | "Wealth" | A broad and contestable category |
Each step upward involves selection and generalisation: treating some features as relevant, ignoring others, and grouping things that share what remains. "Cow" groups Bessie with millions of other animals and discards the differences between them. "Livestock" discards more still.
Abstraction is not necessarily imprecise. A well-defined abstraction can express category-level relationships precisely, which is what makes scientific and legal categories work. But abstract terms can also become vague when their boundaries are unclear or contested: "vehicle" and "mammal" have reasonably firm edges, while "wealth," "professionalism," "fairness," and "innovation" are defined differently across contexts and disciplines.
What You Gain and Lose on Each Rung
| Moving upward | Moving downward |
|---|---|
| Finds categories and patterns | Finds instances and observable evidence |
| Supports transfer to new cases | Supports verification and action |
| Enables comparison and theory | Enables checking whether a claim is true |
| Risks vagueness and drift | Risks fragmented, patternless detail |
"Organizations need clear feedback loops" is a high-level statement. It does not tell you what a feedback loop looks like in a hospital versus a software company, but it says something true about both. "This department's customer satisfaction score fell from 82 percent to 71 percent between Q1 and Q2" is a low-level statement: checkable, actionable, and silent about what it means.
High abstraction is not a defect. It is essential for scientific theory, law, strategy, categorisation, comparison, and planning. The question is fit between level and purpose, not a general preference for the concrete.
Two Failure Modes
Abstraction without grounding. Statements that are technically meaningful but give no traction on decisions. Strategy documents written entirely as "leverage core competencies to create sustainable value through customer-centricity" are the organizational version of never coming down the ladder. Recipients cannot act, because acting requires knowing what to do in a particular situation.
The subtler damage is that ungrounded terms accumulate private meanings. "Customer focus" means one thing to support, another to product, another to sales. Those divergent readings coexist invisibly until they produce contradictory decisions.
Enron's stated values, integrity, communication, respect, and excellence, illustrate the gap that can open between abstract commitments and concrete conduct. The values did not cause the fraud, which ran through governance failure, incentives, accounting structures, deception, and weak oversight. What the abstract layer failed to provide was an enforceable behavioral standard.
Concrete standards can make contradictions easier to identify, though they cannot substitute for governance and enforcement, since determined fraud can coexist with detailed policies.
Detail without interpretation. A report listing 200 customer complaints without identifying the patterns running through them is informative and close to useless. Technical documentation that describes every step without the underlying logic has the same problem. Recipients have more information than they can organise and no framework for drawing conclusions.
Both failures contribute to misinterpretation: when speaker and listener operate at different levels, messages land in ways the sender never intended.
How to Move Between Levels
One effective sequence is concrete example, then abstract principle, then a new application. The first case grounds the idea, the principle enables transfer, and the new case tests whether the principle was understood.
It is not the only workable sequence. Depending on the audience and the goal, any of these can be better:
- overview, then example, then detail;
- problem, then model, then action;
- data, then pattern, then decision;
- worked example before theory, or theory before application;
- contrasting cases, or guided discovery.
The right choice depends on prior knowledge and task rather than a fixed rule.
Testing understanding with a single new case is also weaker than it looks. Correct application once does not establish general understanding. Stronger checks include a contrasting case, a boundary case, a counterexample, an example the learner generates independently, or an explanation in their own words.
And when someone answers unexpectedly, that does not by itself prove the abstraction was unclear: they may lack prerequisites, have misread the example, be working in a genuinely ambiguous case, or disagree with the framework. The useful response is to investigate how the principle was understood.
Martin Luther King Jr.'s "Letter from Birmingham Jail" (1963) can be read as moving effectively between levels. It opens concretely, with his confinement and the statement calling his activities unwise and untimely, ascends to broad principle about freedom and oppression, and returns to the concrete experience of endurance running over. Whether that movement was deliberate is an inference about authorial intent rather than something the text establishes.
Workplace Applications
Vague feedback. "Be more professional" sits well above the behaviors a manager actually has in mind, which might be responding within a working day or not interrupting in client meetings. Those are illustrative examples, not a definition of professionalism, and the point is that the abstraction floats above whatever specific conduct is meant. Naming the conduct is what makes the feedback actionable.
Values and mission. Alignment problems often include a gap between abstract statements and concrete decisions, though they also arise from incentives, leadership behavior, resource constraints, conflicting objectives, weak accountability, and lack of capability. Where the gap is the issue, the fix is to traverse the ladder explicitly: we say we value innovation, so what would that look like in the decisions you face, and what would count as evidence that we are doing it rather than saying it?
Strategy and execution. Strategy is usually expressed at a higher level than day-to-day action, which creates a need to map priorities onto concrete choices. This is a difference of degree, not kind: strategies can contain specific resource allocations and timelines, and execution routinely involves abstract judgment. Without explicit mapping, people apply the same words to different realities and decide inconsistently.
Data communication. The inverse problem. Numbers arrive without the mid-level interpretation that says what they mean. Supply it explicitly: here are the figures, here is the pattern they show, and here is the principle that follows.
Metaphors and Their Limits
A metaphor makes an unfamiliar concept easier to reason about by mapping it onto a more familiar domain. It is not inherently a movement between abstraction levels: metaphors can connect two abstract domains, two concrete ones, or domains at similar levels.
Later conceptual-metaphor research, particularly George Lakoff and Mark Johnson's Metaphors We Live By (1980), examined how familiar source domains shape reasoning about abstract topics.[6] When we treat argument as war, attacking positions and shooting down ideas, we import a whole framework from one domain into another, and it shapes how we approach the activity rather than merely describing it.
The risk is importing structure that does not hold. Describing the immune system as fighting disease suggests that intervention should always strengthen the attack. In practice immune responses can be insufficient, excessive, misdirected, or poorly regulated, and the warfare framing obscures those distinctions.
The same caution applies to the signal versus noise framing. It is a useful editorial metaphor, but the engineering analogy has limits, because human meaning depends on context, inference, and shared knowledge in ways that channel noise does not capture.
Why the Gap Is Easy to Miss
Communicators may underestimate how abstract their language appears to a less knowledgeable audience. With expertise, compressed terms become familiar enough that their hidden prerequisites are easy to overlook. This connects to the curse of knowledge, where knowing something distorts your estimate of what others know.
It is worth separating three things that often get treated as one:
- Unfamiliar terminology. Jargon and abstraction are different dimensions. "Femur," "M8 bolt," "IP address," and "sodium chloride" are technical and highly concrete.
- Compressed reasoning. Steps that have been collapsed because they are automatic for the speaker.
- Abstract generality. Genuine distance from observable instances.
"Algorithm" is a useful test case. It is an abstract category, but its meaning can feel immediate to an experienced engineer because the term activates many familiar examples and structures. That is familiarity, not concreteness, and confusing the two leads people to think they have descended the ladder when they have not.
Dialogue helps, but not automatically. A confused expression can prompt clarification, yet people also hide confusion, misread reactions, or stay quiet because of hierarchy or embarrassment. Spoken exchange creates opportunities to detect mismatch only when participants feel able to signal it and the speaker actively checks.
A Practical Checklist
- What level am I using?
- What level does this audience need, and how far apart are we?
- What concrete example anchors this abstraction?
- What principle connects these details?
- Where does the example or metaphor stop applying?
Two habits do most of the work. When an abstract statement is generating confusion, descend rather than adding more abstraction: asking what we specifically mean by a term is a useful move in disputes conducted entirely at high level. And when a concrete example has been shared, say what it illustrates, or it stays an isolated case.
Ground important or potentially ambiguous abstract terms when the audience may not share your meaning. Doing this for every abstract term would make expert writing repetitive; doing it for the load-bearing ones is what makes a principle usable.
Watch for abstractions that float free of any referent. "We need to be more agile" invites the question of what you would see differently if agility were present. Floating abstractions can conceal unresolved disagreement or missing operational detail, whether or not anyone intends that, and they can equally reflect uncertainty, diplomacy, brevity, or early-stage thinking.
Evidence and Its Limits
Hayakawa's ladder is a linguistic and pedagogical model rather than an experimental finding, and the research below is adjacent evidence rather than validation of the complete model.
Construal level theory. Nira Liberman and Yaacov Trope's work, beginning with a 1998 study in the Journal of Personality and Social Psychology and synthesised in their 2010 Psychological Review article, established that people represent psychologically distant events more abstractly than near ones.[4][3] A commitment for next year is described as staying healthy; the same commitment for tomorrow is going to the gym at seven.
This is genuinely related to the ladder, but it is not the same thing: construal level theory concerns psychological distance and mental representation, while Hayakawa's ladder concerns linguistic and semantic generality. Some studies suggest that matching message construal to psychological distance can influence persuasion and preference, though outcomes vary by task, goal, and audience, and the 2010 review does not report a single meta-analytic effect size for it.
It is not a universal communication rule.
Expert and novice differences. Cheryl Geisler's Academic Literacy and the Nature of Expertise (1994) examined reading, writing, and knowing in academic philosophy, finding that expertise involves working with texts at a different level of abstraction rather than simply understanding more of them.[5] Experts tend to extract principles where novices record cases. This does not mean experts have lost the capacity to work concretely: they may find it harder to reconstruct a novice reading because their knowledge is organised differently, but they can explain concretely, test with real readers, and develop teaching skill.
Stylistic observation. In The Sense of Style (2014), Steven Pinker describes how clear writers characteristically move between concrete detail and abstract point, using vivid examples to anchor general claims.[7] This is a stylistic observation rather than a quantified corpus study, and the specific example-principle-application pattern described earlier in this article is a WhenNotesFly framework rather than a structure Pinker established empirically.
Multiple contrasting examples can help learners work out which features of a concept are essential and which are incidental. That is a reason to use more than one case, not evidence that concept formation always requires at least two.
One Case: The Pentium FDIV Bug
In 1994 Intel's Pentium processor contained a flaw in its floating-point division unit that produced incorrect results for a narrow range of calculations. Intel initially described the problem in general terms, as affecting a very small fraction of users performing specific mathematical operations rarely encountered in practice. That description was technically defensible and gave individuals no way to judge whether it affected them.
When mathematician Thomas Nicely publicised the flaw with specific division problems and the size of the resulting errors, the issue became immediately comprehensible to a general audience.
Concrete examples helped make the defect understandable, but the controversy also concerned Intel's initial replacement policy, its credibility, the level of error different users considered acceptable, and IBM's decision to halt shipments of affected machines. Intel eventually took a $475 million charge to replace affected chips, and the episode became an influential case in technology product communication and customer-response policy.
Sources & Further Reading
- Hayakawa, S. I. Language in Thought and Action. Harcourt Brace, 5th ed., 1990. View source(The ladder and the Bessie example.)
- Korzybski, A. Science and Sanity. International Non-Aristotelian Library, 1933. (Map and territory.)
- Liberman, N., & Trope, Y. "The Role of Feasibility and Desirability Considerations in Near and Distant Future Decisions." Journal of Personality and Social Psychology, 75(1), 5-18, 1998. DOI: 10.1037/0022-3514.75.1.5(Distance and abstraction.)
- Trope, Y., & Liberman, N. "Construal-Level Theory of Psychological Distance." Psychological Review, 117(2), 440-463, 2010. DOI: 10.1037/a0018963(The synthesising review.)
- Geisler, C. Academic Literacy and the Nature of Expertise: Reading, Writing, and Knowing in Academic Philosophy. Lawrence Erlbaum, 1994. (Expert and novice abstraction levels.)
- Lakoff, G., & Johnson, M. Metaphors We Live By. University of Chicago Press, 1980. View source(Conceptual metaphor.)
- Pinker, S. The Sense of Style. Viking, 2014. View source(Concrete and abstract prose.)
Further Reading
- Williams, J. M., & Bizup, J. Style: Lessons in Clarity and Grace. Pearson, 2014. (Sentence-level craft.)
Frequently Asked Questions
What is the ladder of abstraction?
It is S. I. Hayakawa’s model of how words sit at different distances from observable things, introduced in Language in Action (1939). At the bottom are specific, checkable references such as a particular cow in a particular field; at the top are broad categories such as livestock, assets, or wealth. Each step upward selects some features as relevant and discards others. The model is a linguistic and pedagogical framework rather than an experimental finding, and its value is in making the level you are speaking at visible so you can choose it deliberately.
Is abstract language just imprecise language?
No. A well-defined abstraction can express category-level relationships precisely, which is exactly what makes scientific, legal, and financial categories useful. Terms like vehicle or mammal have reasonably firm boundaries. But abstract terms can become vague when their boundaries are unclear or contested: wealth, professionalism, fairness, and innovation are defined differently across contexts and disciplines. So abstraction is not inherently imprecise, though it is where vagueness tends to appear.
Should I always start concrete and then move to the principle?
Concrete example, then principle, then a new application is one effective sequence, not the only one. Depending on the audience and the goal, an overview before examples, a problem before the model, a worked example before theory, theory before application, or contrasting cases may all work better. The right choice depends on prior knowledge and task. Move between levels according to what the audience already knows and what they need to do with the idea.
Is jargon the same as abstraction?
They are different dimensions that often get treated as one. Plenty of technical terms are highly concrete: femur, M8 bolt, IP address, sodium chloride. What makes expert language hard for others is usually some mix of three things: unfamiliar terminology, reasoning that has been compressed because it is automatic for the speaker, and genuine abstract generality. Algorithm is a useful test case, since it is an abstract category whose meaning feels immediate to an experienced engineer. That is familiarity rather than concreteness, and confusing the two makes people think they have descended the ladder when they have not.
Is concrete communication always better than abstract?
No, and treating it that way is the most common misreading of the model. High abstraction is essential for scientific theory, law, strategy, categorisation, comparison, transfer, and planning. Staying too low produces detail with no organising pattern: a report listing 200 complaints without identifying what runs through them is informative and close to useless. The principle is fit between level and purpose, not a general preference for the specific.
How do I check whether an abstraction was understood?
Applying it correctly to one new case is weaker evidence than it looks. Stronger checks include a contrasting case, a boundary case, a counterexample, an example the other person generates themselves, or an explanation in their own words. If they answer unexpectedly, that does not by itself prove your explanation was unclear: they may lack a prerequisite, have misread the example, be working in a genuinely ambiguous case, or disagree with the framework. Treat it as a prompt to find out how the principle was understood.