Evolution found the same answer twice, in completely different brains.
At a Dartmouth laboratory, researchers have documented octopuses using mirrors to locate hidden food with 73 percent accuracy — a cognitive feat long assumed to belong only to vertebrates. The finding, published in Current Biology, does not merely add a footnote to animal behavior science; it asks us to reconsider the architecture of intelligence itself. When creatures separated from us by 600 million years of evolution arrive at similar solutions to similar problems, the question is no longer which animals are smart, but what intelligence truly requires to emerge.
- A foundational assumption in cognitive science — that sophisticated spatial reasoning belongs exclusively to vertebrates — has been quietly dismantled by eight-armed animals in a controlled laboratory setting.
- The 73% success rate is not a curiosity; it is a data point that forces a reckoning with how we have drawn the boundaries of mind across the animal kingdom.
- Researchers must now contend with the possibility that brain structure matters far less than brain function — that evolution can build comparable thinking tools from radically different biological blueprints.
- The concept of convergent evolution, already familiar in anatomy, is being extended into the realm of cognition, suggesting that complex environments reliably pressure nervous systems toward flexible problem-solving regardless of their origin.
- The field is now positioned to ask a harder question: if invertebrates can reason spatially, which other cognitive abilities presumed to be vertebrate-exclusive are waiting to be discovered elsewhere in the tree of life?
A team at Dartmouth has documented something that quietly overturns a long-held assumption: octopuses can learn to use mirrors — not merely to recognize themselves, but to understand reflections as a practical tool for locating food they cannot see directly. In controlled trials, they succeeded 73 percent of the time, a result now published in Current Biology.
Octopus intelligence has captured popular imagination before — Inky, the escape artist who fled New Zealand's National Aquarium in 2016, became a kind of folk hero. But there is a meaningful difference between anecdotal cunning and demonstrated cognitive ability. What the Dartmouth study shows is that octopuses can translate a two-dimensional reflection into a three-dimensional search strategy. That is not instinct. It is learning — the kind we associate with primates, corvids, and elephants, not with an animal whose brain is the size of a walnut and whose lineage diverged from ours roughly 600 million years ago.
Lead author Mary Kieseler framed the finding around a deeper question: how does intelligence evolve? If octopuses and vertebrates independently arrived at similar cognitive solutions, it suggests that certain environments — complex, three-dimensional, demanding of flexible thinking — can push very different nervous systems toward comparable abilities. This is convergent evolution applied not to fins or eyes, but to cognition itself.
The implications are unsettling in the best way. Our long-standing taxonomies of animal intelligence have been built around brain structure and evolutionary proximity to ourselves. This research suggests those categories may be the wrong ones. What matters, it seems, is not the architecture of the mind but the problems it is asked to solve — and evolution, given enough pressure, finds more than one way to build the capacity to solve them.
A team at Dartmouth has documented something that upends a long-held assumption about the animal mind: octopuses can learn to use mirrors. Not just recognize themselves in them—that's a different question entirely—but understand how reflections work as a tool to locate food they cannot see directly. In controlled trials, the octopuses found hidden meals with a 73 percent success rate, a result that matters because it suggests a form of spatial reasoning once thought to belong exclusively to vertebrates.
The study, now published in Current Biology, arrives at a moment when octopus intelligence is already well-established in popular imagination. There was Inky, the escape artist who fled New Zealand's National Aquarium in 2016, squeezing through a gap in his tank and making his way to the ocean. Stories like that have primed us to think of octopuses as clever. But there is a difference between anecdotal cunning and demonstrated cognitive ability—between a creature that seems smart and one that can be shown, in a laboratory, to solve a problem using abstract reasoning.
What the Dartmouth researchers found is that octopuses, when trained to understand how mirrors work, could use that understanding to solve a practical problem. They could see a reflection of food and translate that two-dimensional image into a three-dimensional search strategy. This is not instinct. It is learning. It is the kind of thing we associate with primates, corvids, elephants—animals with large brains and long evolutionary histories of social complexity. An octopus has a brain the size of a walnut, distributed partly through its arms, and it diverged from the human lineage roughly 600 million years ago.
Mary Kieseler, the lead author of the study, emphasized that this finding raises a particular question about how intelligence itself evolves. If octopuses and vertebrates arrived at similar cognitive solutions independently, it suggests that the pressures of certain environments—the need to hunt in three-dimensional space, to navigate complex terrain, to solve novel problems—can push different kinds of nervous systems toward comparable abilities. This is convergent evolution: the same answer emerging from different starting points.
The implications ripple outward. If invertebrates can develop sophisticated spatial cognition, then our taxonomies of intelligence may need revision. We have long sorted animals into categories based on brain structure and evolutionary proximity to ourselves. But this research suggests that the architecture of the brain matters less than what it needs to do. A coral reef or a seafloor presents problems that demand flexible thinking, and evolution has found multiple ways to build that capacity.
The octopuses in this study were not exceptional specimens. They were ordinary animals given an ordinary task. What made the difference was the experimental design—the patience to teach them, the precision to measure their success, the willingness to ask whether a creature so unlike us might possess abilities we had assumed were uniquely ours. That willingness to be surprised, to let the data reshape our categories, is where real science lives.
Citações Notáveis
This mirror-using ability in invertebrates demonstrates evolving intelligence, emphasizing the octopus's ancient divergence from humans.— Mary Kieseler, lead author
A Conversa do Hearth Outra perspectiva sobre a história
Why does it matter that octopuses can use mirrors? They're already known to be intelligent.
The difference is between reputation and proof. We knew octopuses were clever from stories and observation. This is the first time we've shown that an invertebrate can grasp the abstract relationship between a reflection and reality—something we thought required a vertebrate brain.
But 73 percent success—that's not perfect. Does that weaken the finding?
Not at all. It's actually more convincing. Perfect performance would suggest they were following a simple rule. Seventy-three percent suggests they're solving a problem, sometimes getting it wrong, learning from failure.
The researcher mentioned convergent evolution. What does that mean in this context?
It means octopuses and vertebrates didn't inherit this ability from a common ancestor. They evolved it separately, in response to similar environmental demands. That suggests the ability isn't tied to any one type of brain—it's a solution that works, and evolution found it twice.
So could other invertebrates do this too?
That's the open question now. We've only tested octopuses. But if the ability emerges from environmental pressure rather than brain type, then yes, we might find it elsewhere. We just haven't looked.