SUNK COST EFFECT: CROSS-SPECIES BIAS

The first study examines the sunk cost effect in mice, rats, and human participants using complementary foraging paradigms—Restaurant Row for rodents and Web-Surf for humans. Participants were asked to decide whether or not to continue waiting for a reward (food or video) after already investing some time. The study found that all three species demonstrated increased commitment to obtaining the reward the longer they waited, exhibiting a clear sunk cost effect. Notably, the effect did not emerge until after the initial decision to wait, suggesting a specific vulnerability in reevaluation processes rather than initial deliberation. Researchers employed statistical simulations and tests to confirm that the suboptimal behavior aligned with sunk cost sensitivity. Furthermore, post-decision "liking" ratings and temporal changes in decision strategies reinforced the idea that time spent can irrationally influence subsequent actions, even when it should not logically affect decisions (Sweis et al.).

The second study challenges the interpretation of the sunk cost effect presented by studies such as that of Sweis et al. The authors argue that apparent sunk cost sensitivity may arise from attrition bias—a statistical illusion driven by fluctuations in motivation. Using a computational model with the same task parameters, they demonstrated that a rational agent—one not influenced by sunk costs—could still display the same behavioral patterns. The study emphasizes that designs like Restaurant Row cannot conclusively attribute behavior to sunk cost sensitivity unless sunk costs are decoupled from other valuation processes. The authors call for improved experimental designs incorporating causal inference and generative models to better distinguish between true sunk cost behavior and confounding variables (Ott et al.).

While the first study concludes that a genuine sunk cost effect exists across species and is driven by post-commitment reassessment, the second argues that the observed behaviors can be fully explained without invoking sunk cost mechanisms. Sweis et al. interpret their findings as evidence of a conserved neuroeconomic trait, while the second study contends that such behaviors result from statistical artifacts caused by uncontrolled task variables. The first relies on behavioral data and computational modeling to justify suboptimal decisions based on prior investments, whereas the second offers a theoretical framework grounded in rational behavior and calls for refined experimental methods to reinterpret those same results.

 WORK CITED: 

Ott, Florian, et al. "Revisiting the Sunk Cost Effect: A Rational Framework and Improved Task Design." bioRxiv (2020), https://doi.org/10.1101/2020.02.10.942086.

Sweis, Brian M., et al. “Sensitivity to ‘Sunk Costs’ in Mice, Rats, and Humans.” Science, vol. 361, no. 6398, 2018, pp. 178–181. https://doi.org/10.1126/science.aar8644.