What is the Extensive Form in Game Theory?

In game theory, the extensive form is a mathematical and visual framework used to represent games that unfold sequentially over time. Unlike the normal (or strategic) form, which uses a matrix to depict simultaneous decisions, the extensive form models a game as a tree structure to capture the exact order of players’ moves, the choices available at each step, the information players possess when making decisions, and the final outcomes. This article explains the core components of the extensive form representation, detailing how game trees, nodes, information sets, and payoffs combine to model complex strategic interactions.

The Structure of the Game Tree

The backbone of the extensive form is a directed tree, which consists of a series of nodes connected by branches:

• Root Node: The starting point of the game, representing the very first decision to be made.
• Decision Nodes: Points in the game where a player (or “Nature,” representing random chance) must choose an action. Each decision node is assigned to a specific player.
• Branches: The lines connecting the nodes, which represent the actions or moves available to the player at that specific decision node.
• Terminal Nodes: The endpoints of the game where play concludes. No further moves can be made from these nodes.

Information Sets and Knowledge

A key advantage of the extensive form is its ability to represent what players know when they make a move. This is achieved through information sets:

• Perfect Information: If a player knows exactly what choices have been made prior to their turn, every decision node sits in its own unique information set.
• Imperfect Information: If a player must make a choice without knowing the exact previous moves of other players, multiple decision nodes are grouped into a single information set. Visually, these nodes are often enclosed in a dashed line or connected by a dotted curve. The player knows they are at one of the nodes within the set, but they cannot distinguish which one.

Payoffs

At every terminal node, a payoff vector is specified. This vector lists the utility, reward, or payoff received by each player when the game ends at that particular node. The payoffs are written in a specific order, typically corresponding to Player 1, Player 2, and so on, allowing analysts to evaluate which sequence of moves yields the most desirable outcome for each participant.

Applications and Analysis

The extensive form is crucial for analyzing multi-stage games with sequential moves, such as chess, poker, or multi-round business negotiations. Because it details the chronological flow of the game, researchers can use analytical techniques like backward induction—working backward from the terminal nodes to the root node—to determine the optimal strategies and find the subgame perfect Nash equilibrium.