Apparent sunk cost effect in rational agents.

Area of Science:

• Cognitive Science
• Behavioral Economics
• Neuroscience

Background:

• Rational decision-making aims to maximize gains, but observed behaviors in humans and animals often deviate, suggesting cognitive biases.
• A recent study interpreted certain economic decision-making patterns in humans, mice, and rats as evidence of the sunk cost fallacy.

Purpose of the Study:

• To challenge the interpretation of observed economic decision-making as the sunk cost fallacy.
• To propose an alternative explanation for the observed behaviors based on rational, reward-maximizing strategies.
• To suggest improved experimental designs for future research.

Main Methods:

• Analysis of a reported study on economic decisions in humans, mice, and rats.
• Application of a computational model simulating decision-making strategies.
• Identification of potential statistical fallacies, specifically attrition bias, in the original interpretation.

Main Results:

• The observed behavioral patterns, previously attributed to the sunk cost fallacy, can be reproduced by a computational model employing a reward-maximizing strategy.
• The interpretation of sunk cost fallacy in the original study is challenged due to a statistical fallacy (attrition bias).
• The study demonstrates that behaviors can be explained without invoking a sunk cost-dependent mechanism.

Conclusions:

• Observed decision-making behaviors in economic tasks may not necessarily indicate a sunk cost fallacy.
• Statistical confounds, such as attrition bias, can lead to misinterpretations of cognitive behaviors.
• Causal statistical inference and generative models are crucial for accurate interpretation of decision-making processes.