Two-Ball Ellsberg & Savage Models
Our research confronts Savage's decision-making models with experimental evidence, revealing crucial discrepancies. By demonstrating that individuals' choices in our Two-Ball Ellsberg Paradox experiment defy predictions made by Savage's framework, we highlight fundamental challenges in modeling decisions under uncertainty.
Savage's Framework and Its Limitations
Savage's seminal work (Savage 1972) presents a robust framework for decision-making under uncertainty, advocating for subjective probabilities as the basis for individuals' choices. Unlike models that rely on objective probabilities, Savage's theory encapsulates a broader range of uncertainties by allowing for subjective belief formation. However, our experimental findings expose limitations in this approach, particularly when faced with complex, ambiguous scenarios like the Two-Ball Ellsberg Paradox.
Experimental Evidence Against Savage's Models
Our experiments, particularly the Two-Ball Ellsberg Paradox, present scenarios where individuals' choices systematically deviate from predictions made by Savage's subjective probability model. Contrary to the expectation that individuals form probabilities congruent with the experimental setup, we observe a persistent preference for known probabilities (the gamble $R$) over ambiguous ones (the gamble $AA$), even when the latter offers objectively higher or equivalent winning chances. This contradicts the notion that subjective probabilities in Savage's framework align with the realities of the decision context.
Refuting Key Assumptions
The core assumptions of Savage's model, particularly the alignment of subjective probabilities with experimental realities, are challenged by our findings. We argue that while Savage's model allows for a wide range of subjective belief formations, it fails to account for the cognitive biases and heuristics that play a crucial role in real-world decision-making. Our experiments suggest that individuals may not always form subjective probabilities that reflect the actual structure of uncertainty they face, particularly in ambiguous situations.
