Causal Decision Theory

Causal Decision Theory (CDT) is a school of thought within decision theory which posits that rational agents should choose actions based on their causal effects. In other words, an agent should select the action that produces the best expected outcome by considering how their actions cause subsequent events.

The fundamental principle of CDT is expected utility. An agent evaluates the potential outcomes of different actions by calculating the expected utility of each action. This involves summing the utilities of all possible outcomes, each weighted by the probability that the outcome will occur, given the action chosen.

CDT emphasizes the importance of the causal influence of actions. It argues that an agent should consider how their actions will directly influence outcomes, rather than basing their decisions on historical data or correlated information. This is in contrast to Evidential Decision Theory (EDT), which recommends actions based on the evidence they provide about what the likely outcomes are.

One of the most famous thought experiments associated with CDT is Newcomb's Paradox. Newcomb's Paradox involves a game with two boxes, Box A and Box B, and a highly reliable predictor. The predictor has already placed money in these boxes based on its prediction of the player's choice:

• Box A is transparent and always contains $1,000.
• Box B is opaque and contains either nothing or $1,000,000.

The player can either take only Box B or both Box A and Box B. If the predictor has predicted that the player will take both boxes, Box B will be empty. If the predictor has predicted that the player will take only Box B, it will contain $1,000,000.

According to CDT, the rational decision is to take both boxes. This is because the contents of Box B have already been determined by the predictor, and taking both boxes causally dominates taking only one, as it ensures at least $1,000 more. The predictor's decision is seen as fixed and not causally influenced by the player's current choice.

CDT has been criticized for its approach to Newcomb's Paradox, with some arguing that it fails to consider the predictive nature of the situation. Evidential Decision Theory (EDT), for instance, suggests that the player should take only Box B, as doing so provides strong evidence that Box B contains $1,000,000.

Functional Decision Theory (FDT) is another alternative to CDT that has been proposed. FDT argues that decisions should be based on the function of the agent's decision-making process itself and how it correlates with the actions of similar agents or predictors. In Newcomb's Paradox, FDT would recommend taking only Box B because the agent's decision-making process is functionally correlated with the predictor’s predictions.

CDT is applied in various fields such as economics, artificial intelligence, and philosophy. In economics, CDT helps in understanding how agents make rational choices that maximize their utility. In artificial intelligence, CDT frameworks are used to design agents that can make decisions based on causal relationships rather than mere correlations. In philosophy, CDT provides a structured way to evaluate ethical decisions based on their outcomes.

Related Topics

• Decision Theory
• Evidential Decision Theory
• Newcomb's Paradox
• Functional Decision Theory
• Expected Utility

Causal Influence