Â鶹ÒùÔº

Rats successfully distinguished between two varieties of white wine, Riesling and Sauvignon Blanc, in a study led by researchers from the University of Trento, the University of Lincoln, the University of London, and the University of Vienna. Findings indicate that non-human animals can categorize complex olfactory stimuli without linguistic or human-specific cognitive abilities.

In the study, "Rats Can Distinguish (and Generalize) Among Two White Wine Varieties," in Animal Cognition, researchers applied an experimental olfactory discrimination method using a go/no-go scenario to assess whether rats could learn and generalize odor categories in a complex domain.

Nine adult male domesticated rats, previously trained on an unrelated odor categorization task, were trained to discriminate between wines of two distinct grape varieties.

Each rat was assigned one variety as the rewarded stimulus (S+), while the other served as the non-rewarded stimulus (S−). Training consisted of exposure to multiple wines of each variety across distinct vintages and geographic origins.

The experimental setup included an automated odor delivery system with chambers where rats were required to perform a nose poke to trigger a lever presentation. A correct response (lever press for S+) resulted in a food reward, while an incorrect response (lever press for S−) led to a timeout penalty. Training continued until the rats maintained over 80% correct responses for three consecutive sessions.

During testing, rats were presented with novel wines of the same varieties in unrewarded probe trials. Results showed that all nine rats learned to discriminate between the two wine varieties, with most generalizing their learned category distinctions to novel wines.

Rats pressed the lever more frequently for novel wines that matched their previously rewarded category, confirming their ability to generalize beyond specific training samples.

A significant difference in response rates between novel S+ and novel S− stimuli was detected through statistical analysis, with a generalized linear mixed model indicating successful category-based generalization. Rat sommeliers correctly distinguished 94% of trained wines and 65% of novel wines, though novel wine discrimination did trend upwards with repeat exposure. Eight out of nine rats were more likely to generalize novel wines, and six showed statistically significant levels of discernment.

A rat named Peanuts was the only one of the nine rats that completed the training phase but exhibited an unusual response pattern during the testing phase. Unlike the other rats, Peanuts did not successfully generalize the learned discrimination to novel wines. Instead, he showed a higher rate of incorrect responses, pressing the lever more often for the non-rewarded (S−) novel wines than for the correct category (S+).

Peanuts' results might reflect natural differences between individuals in olfactory processing or memory consolidation. Peanuts could also have been more of a connoisseur than the others, appreciating elements of each wine variety as its own reward, leading to inconsistent responses in the experiment.

Mixed training may have contributed to the rapid acquisition of discrimination skills, supporting its effectiveness in olfactory learning paradigms. Researchers implemented a mixed odor presentation approach, in which multiple wine samples were introduced within the same session, rather than training rats on a single wine at a time. This contrasts with the sequential training method often used in scent detection studies, where subjects learn each target odor individually before progressing to the next.

Findings suggest that complex olfactory categorization does not require linguistic mediation and may be more widespread among non-human animals than previously thought. Future research may explore whether similar olfactory discrimination abilities extend to other mammalian species, including dogs and primates, and whether rats can distinguish wines by terroir, winemaking style, or region of origin.

© 2025 Science X Network