When food and shelter override fear

What wild mice teach us about predator scents, field assays, and the limits of simple conclusions

One of the most tempting stories in rodent research is also one of the riskiest: present a predator cue, watch what the animal does, and treat the result as a clean readout of fear.In the laboratory, that often looks straightforward. Predator scents can produce freezing, vigilance, startle responses, or avoidance.In the field, things are messier. Wild animals are not only deciding whether something feels dangerous.They are also deciding whether they are cold, hungry, sheltered, desperate, curious, reproductively motivated, or already committed to a risky foraging choice.This recently accepted 2026 manuscript asks exactly what happens when those competing motivations are allowed back into the picture.The paper examines two species of free-ranging wild mice near Warsaw, striped field mice ( Apodemus agrarius ) and yellow-necked mice ( Apodemus flavicollis ), in a winter field assay built from laboratory-inspired chambers.The researchers expected native predator scents from foxes and domestic cats to provoke stronger aversive responses than non-predator deer scent or procedural controls.Instead, they found almost no overt fear responses at all. No freezing. No flight. No withdrawal.Only trivial treatment differences, and a subtler pattern that turned out to be the real signal: the chambers with no animal-derived scent showed much greater behavioral variance than chambers containing predator or herbivore scent cues.The authors, Professor Rafal Stryjek and Dr. Michael H. Parsons argue that highly palatable food and winter shelter buffered measurable fear responses, making the assay itself more attractive than the risk cues were aversive.That result matters well beyond mice. It is a methodological warning for anyone trying to interpret risk-sensitive behavior in the wild, including urban rodent work.A weak response to predator cues does not necessarily mean the animals do not detect danger.It may mean that the experimental setup has embedded strong rewards that change the decision landscape.In other words, the apparatus may be answering a different question than the one researchers think they asked.

Why field fear studies so often disagree with laboratory studies

The paper begins from a familiar problem in behavioral ecology: laboratory findings and free-ranging field findings often do not line up neatly.Predator odors that reliably trigger defensive responses in the lab may produce muted or inconsistent effects in wild animals outdoors.Stryjek and Parsons locate part of that discrepancy in the usual strengths of laboratory design.By stripping away nuisance variables, the lab increases precision but also removes context.Wild animals, by contrast, make decisions under multiple competing drives at the same time.Hunger, exploration, social structure, season, prior experience, and access to shelter all shape how much weight a fear cue actually carries in the moment.The paper frames this as a dual-motivation problem. The same chamber can attract an animal because it contains highly palatable food and shelter, while also repelling the animal if it contains signs of predation risk.That creates an approach-avoidance conflict rather than a simple test of “fear yes or no.”This matters because many field assays are built precisely to draw animals into a test space.They often use bait that is far more attractive than baseline food in the environment, and they often provide a structure that becomes a windbreak or warm refuge.Under those conditions, failure to observe a clean aversive response should not be surprising.

What the researchers did

The study took place during winter, from mid-January to late March 2023, in a peri-urban landscape near Warsaw, on private land adjacent to forest and meadows.The chambers were pre-baited for about one and a half months before the experimental period.Each chamber was a wooden box with a plastic pipe entrance, monitored around the clock with infrared cameras and motion-triggered video recording.The setup is shown in Figure 1 on page 27, including the paired chambers, the two mouse species, and example chamber footage.The bait choice was not subtle. Each chamber received ten grams of chocolate-nut cream spread on a Petri dish every day after dusk.In winter conditions, that level of palatability matters. The chambers also provided cover from weather, and the paper explicitly notes that their attraction persists even without food because they function as a windbreak and are warmer inside than outside.To control the scent presentations, the team used standardized birch scent probes placed near the food bowls.Predator treatments used fresh fox and cat scent mixtures. A deer scent served as a non-predator animal control, and there were also wet and dry procedural controls.Scents were rotated across chambers using a Latin square structure so that animals always had a choice between one chamber containing an animal-derived or procedural treatment and another chamber containing a water-based control.This matters because the experiment was not merely “predator scent versus nothing.”It was a choice test in which all chambers remained attractive for reasons unrelated to scent.The mice were free to enter or ignore the chambers.The design therefore asked a more realistic but also more complicated question: what do wild mice do when risk cues are embedded inside a shelter-and-food opportunity during winter?

What the study found

Across the study, animals visited the chambers 912 times. Visitation rate did not differ significantly by treatment.The most striking negative result was the absence of overt defensive behaviors.The authors did not observe withdrawal, flight, or freezing during any treatment. That is not a small detail.It means that even with ecologically familiar predator scents from native or sympatric predators, the mice did not produce the classic visible fear behaviors many readers would expect.There was one statistically significant treatment effect in overall time spent in the chamber, but the effect size was tiny.Pairwise differences existed between some treatments, such as deer versus fox or dry control versus fox, yet the paper treats these as trivial in practical terms.Food consumption timing also failed to show meaningful preference patterns.When the researchers compared how often food was consumed first from a chamber with deer, fox, or cat scent versus the water control, the results did not differ significantly from a 50:50 distribution.If the analysis stopped there, the study might be misread as “predator scents do not matter.”The authors do not make that claim. Instead, they pay close attention to the shape of the data.The boxplots on page 28 show that the greatest outlier spread and the highest behavioral variability occurred in the non-animal controls, particularly the dry and water conditions.By contrast, animal-derived scents, including deer as well as predator scents, produced tighter distributions.Levene’s tests supported some of these differences in variance. The authors interpret this pattern as a sign that the mice may actually have perceived the chambers without animal scent as safer, allowing longer and more variable stays, including anecdotal extended sleep episodes that occurred only during control conditions.That is a subtle but important shift. The absence of overt fear does not mean the absence of risk processing.It may mean that the field assay only allowed risk to show up in muted, indirect ways, because food and shelter had already changed the cost-benefit balance.

Why native predator scents still failed to produce a clean fear signal

Part of the study’s logic was to strengthen the cue side of the equation.In earlier related work, the researchers had observed only modest reactions to non-native predator scents and wanted to test whether sympatric predator scents would be more evocative.That prediction did not hold. Native fox and cat scents still failed to generate clear defensive behavior in the chambers.The paper offers a plausible explanation: familiar predator cues may have been treated as part of the ordinary landscape and discounted because the chamber itself was perceived as relatively safe and highly rewarding.By contrast, in earlier work, the novelty of non-native scents may have elicited more exploratory inspection.The deer treatment adds another layer. If the system were only about predator fear, herbivore scent should have looked clearly safer.It did not. The authors suggest that animal-derived scents more generally may carry information about parasites, pathogens, or predator attraction, so even non-predator scent can reduce the sense of safety.This helps explain why the major contrast in the data was not “predator versus herbivore” but “animal-derived scent versus non-animal controls.”

What this means for rodent ecology and field design

This paper is valuable because it forces a more disciplined interpretation of negative results.A field assay does not only measure the intended stimulus. It also measures everything the apparatus contributes to the decision.In this case, the apparatus included highly palatable Nutella bait, enclosed chambers, winter conditions, and free access.Those are not background details. They are part of the treatment environment.If they raise the positive value of entering and remaining in the chamber, then predator scents have to compete against strong shelter and food incentives before any visible aversion appears.For urban rodent work, that lesson travels well even though the study species here were wild mice rather than urban rats.Any real-world assay that uses especially attractive food, warm protected enclosures, tunnel-like structures, or other refuge-like contexts may be testing a composite of risk and reward rather than fear alone.That does not make such assays useless. It means teams should stop pretending they are simple.A site can be “dangerous” and still be worth entering.A cue can be detected and still not be strong enough to override shelter in winter.The behavioral output may therefore say more about tradeoffs than about pure aversion.The authors explicitly conclude that future work should reduce the attractiveness of baits relative to the natural nutritional baseline and minimize the appeal of the test apparatus when the goal is to assess risk perception more cleanly.That recommendation is directly useful. Better field assays will not come only from stronger cues.They will also come from more honest control over what the chamber gives the animal in return.

What this means for EUREKA

For the EUREKA team, the paper is best read as a design principle.Rodent data systems should not log only “response” or “no response.”They should also log the context that may be buffering or amplifying response: season, shelter value, bait attractiveness, baseline food availability, apparatus type, and weather.A negative result without contextual metadata is weak evidence. A negative result with those variables attached becomes interpretable.The paper also argues for paying more attention to variance, not only means or medians.The authors show that the most meaningful pattern in their dataset was not a dramatic shift in central tendency but a difference in how behavior spread across treatments.For EUREKA, that is a useful reminder that monitoring systems should be built to preserve behavioral richness rather than collapse everything into a yes-no event.Variation can be the signal.

Closing

This new manuscript does not tell a simple story about fear. It tells a more useful one.Wild animals make decisions inside conflicting motivational landscapes, and experimental assays can easily smuggle those motivations into the result.In this case, winter shelter and highly palatable food appear to have buffered measurable responses to predator scents, even when the scents came from familiar predators and even when the researchers built the assay specifically to make fear more likely to appear.That makes the paper important not because it proves wild mice are unafraid, but because it shows how easily fear can be hidden inside reward.For anyone studying rodent behavior in natural settings, that is a methodological warning worth taking seriously.Better assays will come from treating ecological context as part of the mechanism, not as noise to explain away after the fact.

 

Reference

Stryjek, R., Parsons, M. H., Bebas, P., Fendt, M., Kiyokawa, Y., Chrzanowski, M., Socha, K., & Blumstein, D. T. Buffering effects of shelter and palatable foods mitigate fear responses in foraging wild mice. Scientific Reports (2026, accepted manuscript / article in press).