Neuroscience has a convenient assumption baked into most animal research: brains work pretty much the same way across species. Study a rat, learn something about humans. Test a drug in mice, predict what it'll do in people. It's a reasonable assumption, right? Evolution tends to preserve solutions that work. Except when it doesn't. A review in Neuroscience & Biobehavioral Reviews takes a hard look at one of the simplest learning paradigms in neuroscience and finds that even the basics aren't as conserved as we thought.
The paradigm in question is eyeblink conditioning, and the results should make every translational researcher a little nervous.
Eyeblink Conditioning: The "Simple" Learning Task
Let's start with what eyeblink conditioning actually is, because it's deceptively simple. You give an animal a cue (a tone, a light, something neutral), then a moment later you puff air at its eye. The animal blinks reflexively. Do this enough times, and eventually the animal starts blinking before the air puff arrives, just from hearing the tone. It's learned to anticipate the annoying thing and protect its eye preemptively.
This is about as basic as learning gets. It's associative. It's cerebellar. It's been studied for decades across multiple species. The cerebellum (that wrinkly structure at the back of your brain that handles motor coordination) is clearly involved, and the circuits have been mapped in exquisite detail. If any form of learning should be conserved across mammals, it's this one.
Spoiler: it's not.
When You Actually Compare Species, Things Get Messy
The review looked at eyeblink conditioning across rabbits, rats, mice, and primates. These aren't exotic, obscure comparisons. These are the workhorses of neuroscience research, the species in which most of our basic knowledge about cerebellar learning was established.
And they don't do the task the same way.
The timing is different. How quickly animals learn varies. How strongly they respond differs. How the learning extinguishes when you stop delivering air puffs shows distinct patterns. Even the specific cerebellar circuits involved aren't identical. Same basic task, same general brain structure, different implementations.
This is like discovering that the recipe for chocolate chip cookies varies not just between households but in ways that fundamentally change whether the cookies are chewy, crispy, or somewhere in between. The end product looks similar enough, but the process isn't the same.
What Exactly Differs?
Let's get specific about what varies. First, optimal training parameters. The timing between cue and air puff that produces the best learning isn't the same in a rabbit versus a rat. The number of trials needed differs. The way you have to structure sessions varies.
Second, the cerebellar circuitry. Yes, the cerebellum is involved in all these species, but the precise pathways and their relative importance shift. Some nuclei matter more in some species than others. The granular details of which cells do what aren't carbon copies.
Third, the relationship between learning and memory. How long the learned response lasts, how it's consolidated, and how it interacts with other cognitive processes shows species-specific patterns.
This is supposed to be simple associative learning. Motor learning. The kind of thing cerebella have been doing for millions of years. And yet, when you actually look closely, "conserved" is a generous description.
Why This Should Worry Translational Researchers
Here's where this gets uncomfortable for anyone trying to develop therapies based on animal research. The whole point of using animal models is that findings should translate to humans. If a drug enhances cerebellar learning in mice, we hope it'll do something similar in people.
But if the underlying mechanisms aren't identical, translation gets dicey. A compound that targets a specific circuit in mice might be targeting a circuit that matters less in humans. Or matters more. Or works completely differently. The superficial similarity of the behavior masks real mechanistic differences underneath.
This isn't an argument against animal research. It's an argument for humility about what animal research can and can't tell us. If even eyeblink conditioning varies significantly across mammalian species, what about more complex learning? What about memory consolidation? What about the dozens of cognitive processes we study in rodents and assume will translate to clinical populations?
Maybe Evolution Is Messier Than We Assumed
There's a certain elegance to the conservation story. Evolution finds a good solution and sticks with it. The basic wiring for cerebellar learning got established early in mammalian evolution, and subsequent species inherited it more or less intact. Elegant, parsimonious, and apparently incomplete.
What seems to actually happen is more nuanced. The general principle is conserved, but the implementation details shift. Different species face different ecological pressures. They have different body sizes, different motor demands, different predators, different lifespans. A learning system optimized for a rabbit might not be optimal for a mouse or a primate.
Evolution doesn't care about making neuroscientists' jobs easier. It cares about organisms surviving and reproducing. If tweaking the cerebellar circuit improves fitness in a particular niche, that tweak gets selected for, regardless of whether it breaks the assumption of cross-species conservation.
What Do We Do With This Information?
The review doesn't suggest throwing out animal research or dismissing everything we've learned from eyeblink conditioning studies. That would be an overreaction. But it does suggest some adjustments in how we think about translation.
First, more direct human work where possible. Eyeblink conditioning can be studied in humans, and doing so lets us check whether findings from animal models actually apply.
Second, being explicit about species differences when they exist. If a mechanism works differently in mice versus rats, that's worth knowing and reporting, not sweeping under the rug.
Third, understanding that translation is not automatic. A finding in one species is a hypothesis about another species, not a conclusion. The burden of proof for clinical relevance should be higher than "it worked in mice."
Simple isn't always simple. Conserved isn't always conserved. And the more closely we look, the more we find that biology's tendency to complicate things extends even to the basics.
Reference: Bhattacharyya S, et al. (2025). Interspecies variations in eyeblink conditioning. Neuroscience & Biobehavioral Reviews. doi: 10.1016/j.neubiorev.2025.106067 | PMID: 41067328
Disclaimer: The image accompanying this article is for illustrative purposes only and does not depict actual experimental results, data, or biological mechanisms.