Chapter 1: Foundations of Choice: Preferences and Utility#

Every day you make choices—what to eat, where to study, how to spend your free time. Economics starts by building a clear language to describe those choices. This chapter introduces the building blocks of that language: preference relations and utility functions.

The Big Picture#

We want to build a model of how people decide, one that is logical yet flexible enough to reflect real desires. The core idea is that a person can compare any two options and say which they prefer, or that they are equally good. From a few simple rules (called axioms) we can build a rich theory. That theory leads to the idea of a utility function: a number that represents a person’s tastes. This chapter will walk you through those axioms and show how they translate into tools economists use every day.

Modeling Choice with Preferences#

Imagine you are at a fruit stand. You can pick an apple, a banana, or a bunch of grapes. To describe your tastes, we need a way to compare any two bundles of fruit—not just these three, but any amounts you could imagine. Economists do this with a preference relation, a precise way of saying “bundle A is at least as good as bundle B.”

We work with a set of all possible alternatives, often called the consumption set . For now, think of as all non‑negative amounts of two goods, like apples and bananas, shown as points on a plane. A preference relation is a way to compare any two items in . We write to mean “ is at least as good as ” (or “ is weakly preferred to ”).

Preference relation: A way to compare any two alternatives in . We write to mean “ is at least as good as ” (or “ is weakly preferred to ”).

The symbol works like a “greater than or equal” sign for satisfaction. But it’s not a number—it only orders things, without telling you how much better one bundle is.

Rationality: Completeness and Transitivity#

For a preference relation to describe a reasonable decision‑maker, we need two basic properties. These are often called the rationality axioms.

Completeness means that for any two bundles and in , you can always compare them. Either you find at least as good as , or at least as good as , or both. There are no “I can’t decide” gaps.

Completeness: For all , either or (or both).

Transitivity makes sure your rankings hang together logically. If you think an apple is at least as good as a banana, and a banana is at least as good as a bunch of grapes, then you must think an apple is at least as good as grapes. Without transitivity, your preferences could cycle: you might prefer A to B, B to C, and yet C to A. That would make it impossible to pick a “best” choice.

Transitivity: For all , if and , then .

Together, completeness and transitivity make a rational preference relation. This is the minimal logical structure we need to model a decision‑maker who can consistently rank any set of options.

A helpful analogy: think of a tournament where every pair of teams plays, and the results are recorded as “team A is at least as good as team B” based on some rule. Completeness says every pair has a result; transitivity says there are no rock‑paper‑scissors cycles in the overall ranking.

📝 Section Recap: Rational preferences satisfy two simple rules: any two bundles can be compared (completeness), and rankings never cycle (transitivity). These axioms give us a consistent way to talk about “better than” or “equally good.”

From Weak Preference to Strict Preference and Indifference#

The weak preference relation is our starting point, but we often want to say “strictly better” or “exactly as good.” Both can be built directly from .

Strict preference, written , means one bundle is definitely better than the other. Formally:

In words: is at least as good as , and is not at least as good as . So is strictly preferred.

Indifference, written , means the two bundles are equally satisfying:

You can think of indifference as a tie: you would be equally happy with either bundle.

These derived relations inherit the logical structure of . For example, strict preference is transitive (if and , then ), and indifference is also transitive under rational preferences.

📝 Section Recap: From the single idea of “at least as good as,” we can cleanly define “strictly better” and “exactly as good.” This keeps our language precise and avoids confusion.

Desirability: Monotonicity and Local Nonsatiation#

Most economic models assume that, all else equal, more of a good is better. This is not always true in real life (too much ice cream can make you sick), but it is a useful starting point. Two related conditions capture this idea.

Monotonicity (or strong monotonicity) says that if bundle has at least as much of every good as bundle , and strictly more of at least one good, then is strictly preferred to . In symbols, for two goods, if , , and at least one inequality is strict, then .

Monotonic preferences: More of any good, with no less of any other, always leads to a strictly better bundle.

A weaker but very powerful condition is local nonsatiation. It says that no matter which bundle you have, there is always another bundle arbitrarily close to it that you strictly prefer. You are never completely satisfied; a tiny improvement is always possible.

Local nonsatiation: For any bundle and any distance , there exists some bundle within distance of such that .

Local nonsatiation does not require that “more is always better” in all directions—it allows for goods you might dislike. But it does rule out “bliss points” where you are perfectly content and any small change makes you worse off. This condition turns out to be crucial for many results in consumer theory.

📝 Section Recap: Monotonicity says more of a good is always better; local nonsatiation says you can always find a slightly better bundle nearby. Both capture the idea that people generally want more, not less.

Convexity: A Taste for Diversification#

Imagine you like both coffee and tea. If you have a cup of each, you might enjoy a mixture of the two—say, half coffee and half tea—more than either extreme alone. That is the intuition behind convex preferences.

Formally, preferences are convex if, whenever you are indifferent between two bundles and , any mixture (weighted average) of them is at least as good as the originals. For a number between 0 and 1:

The bundle is simply a mix: you take a fraction of and the rest () from . Convexity says you never dislike mixing; you either prefer the mix or are indifferent.

If the mix is strictly preferred whenever and are different, we say preferences are strictly convex. Graphically, this means indifference curves (sets of bundles among which you are indifferent) bulge toward the origin. Convexity captures a taste for diversification—a balanced basket is often more desirable than a lopsided one.

📝 Section Recap: Convex preferences mean you like averages at least as much as extremes. This formalizes the common‑sense appeal of a diversified bundle, like mixing two goods rather than consuming only one.

Continuity and Indifference Sets#

We want preferences to be “well‑behaved” in the sense that small changes in a bundle don’t cause wild jumps in your ranking. Continuity of preferences captures this idea.

Continuity of preferences: For any bundle , the set of bundles that are at least as good as (the upper contour set) and the set of bundles that are no better than (the lower contour set) are both closed sets.

A closed set contains its boundary points. So if a sequence of bundles all weakly preferred to comes closer and closer to some limit bundle, that limit is also weakly preferred to . No sudden reversals at the limit.

Continuity guarantees that indifference sets (the collections of bundles all indifferent to a given bundle) are well‑behaved curves without holes or jumps. They separate the consumption set into regions of strictly better and strictly worse bundles. This property is essential for drawing smooth indifference curves and for proving that a utility function exists.

📝 Section Recap: Continuity means your preferences don’t have abrupt breaks; if a sequence of bundles is all at least as good as , then the limit point is too. This ensures indifference curves are connected and well‑defined.

Utility Functions: Representing Preferences Numerically#

Working with binary comparisons for every pair is clumsy. We would much rather assign a number to each bundle so that higher numbers mean more preferred bundles. That is exactly what a utility function does.

Utility function: A function such that for all ,

If such a function exists, we say represents the preference relation . The utility number itself has no meaning beyond the ordering—it is purely a label for ranking. A bundle with utility 10 is better than one with utility 5, but it is not “twice as good” in any objective sense.

When can we find a utility function? Rationality (completeness and transitivity) alone is not enough. For example, lexicographic preferences—where you care infinitely more about one good than another, like comparing words in a dictionary (the first letter decides everything)—cannot be represented by a real‑valued function. But with continuity, a classic result kicks in.

📝 Section Recap: A utility function translates the abstract preference relation into numbers, preserving the ranking. It lets us use calculus and optimization tools, but the actual numbers are only meaningful for their order.

Ordinality: Why Utility Numbers Are Just Rankings#

The utility function we build is ordinal—only the order of the numbers matters, not their size or differences. Any strictly increasing transformation of represents exactly the same preferences.

Suppose is a strictly increasing function (so implies ). Then the composite function also represents , because

For example, if represents some preferences, then represents the same preferences. The two functions give different numbers, but they rank every pair identically.

This ordinal property is why we cannot interpret the difference in utility between two bundles as a measure of “how much” better one is. Utility is like a runner’s finishing place (1st, 2nd, 3rd), not like their race time.

📝 Section Recap: Utility is ordinal, not cardinal. Any monotonic transformation of a utility function represents the same preferences, so only the ranking of numbers has economic meaning.

Existence of a Continuous Utility Function (Debreu)#

A fundamental theorem by Gérard Debreu tells us exactly when a continuous utility function exists. The key ingredients: the consumption set is a well‑behaved, connected set (like all non‑negative amounts of goods in ), and the preference relation is complete, transitive, and continuous. Under these conditions, there exists a continuous utility function that represents .

We won’t prove the theorem here, but the intuition is powerful. Continuity of preferences ensures that the sets of “better than” and “worse than” bundles change gradually, so we can assign numbers in a way that respects the ordering without jumps. This result justifies the standard practice of drawing smooth indifference curves and using differentiable utility functions in most applied work.

📝 Section Recap: If preferences are rational and continuous on a nice consumption set, Debreu’s theorem guarantees a continuous utility function exists. This bridges the abstract preference world and the mathematical world of functions.

Quasiconcavity of Utility from Convex Preferences#

When preferences are convex, the representing utility function has a special shape: it is quasiconcave. A function is quasiconcave if its upper contour sets—the bundles with utility at least some level —are convex sets. In symbols, for any , the set

is convex. This is exactly the geometric translation of convex preferences. If you prefer averages to extremes, then the set of bundles you like at least as much as a given bundle has no “indentations”: it contains the whole line segment between any two bundles in it.

Strict convexity of preferences corresponds to strict quasiconcavity of the utility function (the upper contour sets are strictly convex). Quasiconcavity is weaker than concavity; a function can be quasiconcave without being concave. For example, is quasiconcave (its upper contour sets are convex) but not concave.

📝 Section Recap: Convex preferences imply that any utility function representing them is quasiconcave—its “better‑than” sets are convex. This property is the mathematical mirror of a taste for diversification.

Marginal Rate of Substitution and Its Invariance to Monotonic Transforms#

When a utility function is differentiable, we can measure how much of one good a person is willing to give up to get a little more of another, while staying equally happy. This trade‑off is the marginal rate of substitution (MRS).

For two goods, the MRS of good 2 for good 1 at a bundle is the absolute value of the slope of the indifference curve:

It tells you how many units of good 2 you would sacrifice for one extra unit of good 1, remaining indifferent. Because utility is ordinal, the MRS does not change if we apply a monotonic transformation. If we replace with where is strictly increasing and differentiable, then by the chain rule:

The factor cancels out. So the MRS is a purely ordinal concept—it reflects the shape of preferences, not the arbitrary scaling of the utility function. This invariance is a direct consequence of ordinality.

📝 Section Recap: The marginal rate of substitution measures the trade‑off between goods along an indifference curve. Because utility is ordinal, the MRS does not change if we apply any increasing transformation to the utility function.

Summary#

We began with the simple idea that people can compare options and built a clear framework for rational choice. From the axioms of completeness and transitivity, we defined strict preference and indifference. We then added natural properties—monotonicity, convexity, and continuity—that make preferences well‑behaved and allow us to represent them with a continuous utility function. That utility function is ordinal, so only the ranking matters, not the numerical values. The shape of preferences is captured by quasiconcavity and the marginal rate of substitution, which stays the same under any order‑preserving change of the utility numbers. These ideas are the foundation for everything that follows in consumer theory and equilibrium analysis.