Complex mind making choices.

Decoding Decisions: Can We Reverse Engineer Choices to Understand What Drives Them?

Nico Varela in Mind & Education July 2025 • 4 min read.

"Uncover the groundbreaking research exploring how observing choices can reveal the hidden factors influencing decision-making, from identifying key actions to understanding underlying beliefs and desires."

Imagine being able to understand exactly why someone makes a particular choice, not by asking them, but simply by observing their decisions. This idea, once confined to the realms of science fiction, is now being explored in groundbreaking research that seeks to decode the very essence of decision-making.

The core question at the heart of this exploration is: To what extent can we specify the decision problem someone faces simply by understanding their preferences? Can we truly 'hear the shape' of a decision problem by analyzing choices? This isn't just about predicting what someone will do; it's about understanding the underlying utility function – the internal calculation that drives those decisions.

Traditional economics tells us that if we know an agent's decision problem – their possible actions, potential outcomes, and their preferences – we can predict their choices. But what if we flip this around? What if we only see the choices and must infer the decision problem? This article explores the innovative approaches and potential limitations of this inverse problem, revealing just how much of our internal world is reflected in our external decisions.

The Quest to Map the Decision-Making Landscape

At the heart of this research lies the exploration of an agent's preferences through their 'ranking of information structures.' Imagine presenting someone with various options, each offering different pieces of information relevant to a decision. The way they rank these options – which information they value most – provides critical clues to their underlying utility function.

The research demonstrates that with a 'finite amount of ordinal data' – simply knowing the order in which an agent prefers certain information structures – it's possible to identify their set of 'undominated actions.' These are the key choices they would consider, irrespective of the specific scenario. Think of it as identifying the core strategies a person might employ, no matter the situation.

Finite Ordinal Data: By understanding the ranking of choices, key actions can be identified.
Belief Identification: Beliefs rendering each action as optimal can be pinpointed.
Utility Function: An additional comparison refines utility function.

But it doesn't stop there. The research goes on to suggest that 'an additional smattering of cardinal data' – comparing the relative value of different pieces of information – allows us to pinpoint their utility function with even greater precision. This moves beyond simply knowing what someone prefers to understanding how much more they value one option over another.

Peering Behind the Curtain of Choice

This research opens up a fascinating new perspective on how we understand decision-making. By shifting from predicting choices based on known preferences to inferring preferences from observed choices, it provides valuable insights into the hidden drivers of human behavior. This has implications far beyond academic circles, offering potential applications in fields ranging from marketing and policy-making to artificial intelligence and personalized medicine. As we continue to refine our understanding of how decisions are made, we move closer to a world where we can truly 'hear the shape' of the problems people face, and help them make choices that align with their deepest values and desires.

About this Article -

This article was crafted using a human-AI hybrid and collaborative approach. AI assisted our team with initial drafting, research insights, identifying key questions, and image generation. Our human editors guided topic selection, defined the angle, structured the content, ensured factual accuracy and relevance, refined the tone, and conducted thorough editing to deliver helpful, high-quality information.See our About page for more information.

This article is based on research published under:

DOI-LINK: https://doi.org/10.48550/arXiv.2403.06344,

Title: Can One Hear The Shape Of A Decision Problem?

Subject: econ.th

Authors: Mark Whitmeyer

Published: 10-03-2024

Everything You Need To Know

How can observing an agent's choices help us understand their preferences and motivations?

By analyzing an agent's choices, researchers aim to reverse engineer their decision-making process. This involves understanding how they rank different information structures, which provides clues to their underlying utility function. The way an agent values and prioritizes information reveals their preferences and motivations, essentially showing how much they value certain options over others. It's about understanding what drives their decisions by looking at the choices themselves, not just what they say.

What is the significance of 'finite ordinal data' in understanding decision-making, and what can we learn from it?

Finite ordinal data refers to the ranking of choices. By understanding the order in which an agent prefers certain information structures, researchers can identify their set of 'undominated actions'. These are the key choices or strategies an agent would consider regardless of the specific scenario. This provides insight into the agent's core decision-making strategies without needing to know the specific context or outcomes, offering a fundamental understanding of their approach to decisions.

How does the research approach the concept of the 'utility function', and why is it important?

The research explores the idea of understanding an agent's 'utility function' by observing their choices. The utility function represents the internal calculation that drives an agent's decisions, reflecting their preferences and values. By analyzing how an agent ranks information structures, researchers can infer the shape of this function. With additional 'cardinal data', the utility function can be pinpointed with greater precision. Understanding the utility function allows us to understand the underlying reasons behind choices, moving beyond simple predictions to grasp the 'why' behind each decision.

What is the difference between 'ordinal' and 'cardinal' data in this context, and what do we gain from each?

In this research, ordinal data refers to the ranking of information structures, providing information on the order of preferences (e.g., which information is preferred over another). Cardinal data, on the other hand, offers additional information about the relative value of different pieces of information – how much more one is preferred over another. Ordinal data helps in identifying undominated actions, while additional cardinal data refines the understanding of the utility function, allowing a more precise picture of the agent's preferences and the strength of those preferences.

Beyond academics, where could this research on decoding decisions have real-world applications?

The research opens up exciting possibilities across multiple fields. It could be applied in marketing to understand consumer behavior and tailor advertising. In policy-making, it can help design more effective interventions by understanding what drives citizens' choices. In artificial intelligence, it can improve the design of decision-making systems. In personalized medicine, it could assist in understanding patient preferences and making treatment decisions that align with their values. By understanding the drivers of decisions, this research promises to offer valuable insights that extend far beyond the academic sphere.