Researchers at the University of California, Los Angeles, demonstrated that rodents can display the same kinds of cognitive missteps that people show. Their findings appear in the journal Psychonomic Bulletin and Review as part of a broader effort to map decision making across species.
In the reported work, two behavioral experiments were crafted so that mice would estimate how likely an event would occur before earning a food reward, described by the authors as candy balls. This setup was designed to probe how probability judgments guide action in animals under controlled conditions.
In one training regime, the animals learned to associate a light with a reward when paired with a sound, and in another regime they learned that a sound alone could signal a reward. The mice were exposed to cues and trained to respond to the signals that correctly predicted reward availability, a basic test of cue learning that also established baseline motivation through regular sugar reinforcement.
During a key manipulation, researchers hid the light bulb behind a metal cover while continuing to play sounds. The animals then faced a choice about whether the hidden lamp was on, with a potential food reward contingent on that judgment. Remarkably, the rodents showed a tendency to infer that the lamp was illuminated whenever sound cues were present, an inclination toward assuming a light when a sound was heard. This behavioral bias suggests an internal expectation that certain senses reinforce others, even when direct evidence is missing.
Another well known set of ideas in human psychology is the Linda problem, introduced to illustrate a conjunction error in probabilistic reasoning. The classic scenario describes Linda as a young, engaged thinker with strong interests in philosophy and social justice, who participated in anti nuclear demonstrations during her student years. The challenge, translated for readers, asks people to weigh two options: the probability that Linda works as a bank teller, versus the probability that Linda works as a bank teller and is actively involved in the feminist movement. Although the latter is statistically less probable, many respondents choose it because they naturally align Linda’s description with the more detailed scenario, effectively rating the second option as more likely because it better matches the imagined profile.
The UCLA study therefore adds to a growing picture of how both humans and animals can rely on contextual cues to judge likelihoods, occasionally making errors that appear systematic rather than random. The researchers interpret these results as evidence that a form of linkage error may be more widespread across species than previously acknowledged. They also challenge a specific claim that such probabilistic mistakes in the Linda task arise solely from language features in humans, instead pointing toward cognitive processes that operate independently of linguistic structure and conscious deliberation.
Taken together, the experiments underscore a broader point about decision making: cue association and expectation can shape choices even when the available data are incomplete or misleading. For researchers, this raises intriguing questions about the neural mechanisms that support probabilistic inference and how these mechanisms compare across species. For readers, the findings offer a reminder that cognitive shortcuts can be surprisingly pervasive, influencing judgments in everyday contexts as well as laboratory tasks.