Octopuses Can Now Use Mirrors to Find Hidden Food — What This Tells Us About Animal Intelligence.

Science has a habit of humbling us. Just when we think we have drawn a clean line between human-like intelligence and the animal world, some creature arrives to quietly erase it. In June 2026, that creature was a small, eight-armed invertebrate from California's coastal waters — and the line it erased had stood unchallenged for decades. Researchers at Dartmouth College published a study in the journal Current Biology showing that octopuses can learn to use mirrors to locate food hidden from direct view. It sounds simple enough. But the implications are not simple at all. This is a capability that had only ever been documented in vertebrates — animals with backbones, centralised brains, and a shared evolutionary heritage that separated from octopuses roughly 350 to 500 million years ago. In other words, the octopus just became the first invertebrate in recorded science to demonstrate sophisticated mirror-mediated spatial reasoning, and in doing so it has forced researchers to fundamentally reconsider what intelligence is, where it comes from, and how many different ways evolution has found to build a thinking mind.

The Experiment: What the Researchers Actually Did

The study was led by Mary Kieseler, who conducted the research as a PhD student at Dartmouth's Department of Psychological and Brain Sciences and is now a postdoctoral researcher at the University of Fribourg in Switzerland. The senior author was Peter Tse, a cognitive neuroscientist and professor of psychological and brain sciences at Dartmouth. Their team worked with three California two-spot octopuses, a species known scientifically as Octopus bimaculoides, housed in Dartmouth's dedicated Octopus Lab.

The design of the experiment was elegantly simple. The researchers wanted to find out whether an octopus could do something that sounds straightforward but actually requires a sophisticated cognitive leap: see a reflection of a food source in a mirror, understand that the reflection represents a real object located somewhere else in space, and then navigate to that real location rather than attacking the reflection itself. That distinction matters enormously. An animal that simply reacts to a reflection — seeing movement and lunging at it — is displaying instinct. An animal that uses the reflection as information about where something actually is in the world is displaying something far more interesting.

The training process had two phases. First, the octopuses were given time to simply become familiar with a mirror placed in their habitat, allowing them to habituate to the object without any task attached. Then the real training began. A live crab — the octopus's preferred prey — was placed inside a glass jar and positioned so that the octopus could only see it by looking at a mirror. To reach the crab, the octopus had to understand what the reflection was showing it, turn 90 degrees, and navigate around a corner to the jar's actual location. It could not simply lunge forward at the image.

There was an added complication. Octopuses have chemoreceptors that allow them to smell and taste by touch, which means a live crab in the vicinity generates scent cues that could theoretically help an octopus locate it without using the mirror at all. To eliminate this confounding factor, the team replaced the live crab with a virtual crab image for the final tests. The octopus was placed in a start box open to the top and front, shown the virtual crab image in a mirror directly in front of it, and had to navigate to the correct location — the spot where the real crab would have been — based purely on the mirror information. It had no scent trail to follow. Only visual spatial reasoning would get it there.

The results were striking. After training, the octopuses correctly identified the food's location approximately 73% of the time. That figure alone is impressive, but overhead tracking of a small spot between the eyes on each animal's mantle revealed something even more telling: while the octopuses did not always take the absolute shortest physical path to the hidden stimulus site, they became progressively faster at calculating and reaching the correct location as training progressed. They were not just finding the food by luck or random searching. They were building a model of their environment, updating it using a mirror, and acting on it with increasing confidence and efficiency.

Why Using a Mirror Is Not as Simple as It Sounds

At first glance, using a mirror might seem like a party trick rather than a window into deep cognitive architecture. But the scientific community has long treated mirror use as a meaningful benchmark for a specific kind of spatial and representational thinking that is cognitively demanding in ways that are easy to underestimate.

To use a mirror as a navigational tool — rather than reacting to it as though it were a real object or another animal — an individual has to perform several sophisticated mental operations simultaneously. It has to understand that the mirror produces a representation of the world rather than being part of the world itself. It has to translate that two-dimensional reflection into a model of three-dimensional space. And it has to use that model to generate and execute a physical action in the right direction. Senior author Peter Tse made the comparison explicit and accessible: "We don't enter the world knowing how to use a mirror but learn how to use a mirror. Just as new drivers learn to use a rearview mirror to track other vehicles, octopuses can also learn how to use a mirror to infer where things are in the world."

Before the Dartmouth study, mirror-mediated localisation of hidden objects had been documented exclusively in vertebrates — specific mammals such as primates, dolphins, and elephants, and certain birds such as magpies and corvids. These are all animals with centralised nervous systems, relatively large brains relative to body size, and complex social lives that many researchers have long argued drive the evolution of higher cognitive abilities. The octopus fits none of these criteria in the conventional sense, yet it performed the same cognitive task. Kieseler was unequivocal about the significance of the finding: "Our findings are the first to demonstrate that invertebrates can use mirrors to understand their environment to find prey. It's a skill that previously has only been documented in vertebrates, such as in some mammals and some birds."

The Octopus Brain: A Completely Different Architecture for Thinking

To appreciate why this result is so scientifically revolutionary, it helps to understand just how different the octopus's nervous system is from the brains of the vertebrates it has now matched in this cognitive task. The octopus brain is shaped like a donut — literally a ring of neural tissue wrapped around the animal's oesophagus. It contains approximately 180 million neurons in the central brain. But the truly extraordinary feature is what lies outside the central brain: roughly two-thirds of the octopus's total neurons are distributed throughout its eight arms, with each arm containing neural machinery complex enough to taste, touch, and make local decisions without waiting for signals from the central brain. When an octopus arm reaches into a crevice to feel for prey, it is largely thinking for itself. Each arm is, in effect, a semi-autonomous agent operating under the loose supervision of a central coordinator.

This architecture is so different from the vertebrate model that neuroscientists have spent decades debating whether the octopus's cognitive abilities can meaningfully be compared to those of mammals and birds at all. In vertebrates, intelligence tends to be associated with the expansion of the cerebral cortex — the outer layer of the brain responsible for complex thought, memory, and planning. Octopuses have no cerebral cortex. They evolved their cognitive abilities through a completely different set of molecular and structural mechanisms, over hundreds of millions of years of independent evolution.

What is now emerging from the growing field of cephalopod neuroscience — accelerated by findings like the Dartmouth study — is a picture of an animal that has arrived at sophisticated cognitive outcomes through a radically different biological route. Cuttlefish, squid, and octopuses have excellent memories, use tools, are capable of delayed gratification, and can learn by watching others. They have dopamine receptors that function differently from those in vertebrates, yet appear to support similar reward-learning processes. The molecular story is different, but the behavioural results are strikingly convergent.

What This Tells Us About the Evolution of Intelligence

The deepest implication of the Dartmouth findings lies in what they reveal about the nature of intelligence as a biological phenomenon. Humans and octopuses share a common ancestor — but that ancestor lived roughly 350 to 500 million years ago and was probably a worm-like creature with a rudimentary nervous system and primitive eye-like patches of light-sensitive cells. Everything the octopus brain does today, it evolved entirely independently from anything the vertebrate brain does. There is no shared cognitive heritage to explain the overlap. The only explanation is convergent evolution: the independent development of similar solutions to similar problems by organisms on completely separate evolutionary branches.

Kieseler addressed this directly, noting that because the two lineages split so long ago, the independent development of spatial processing in octopuses indicates that complex cognitive solutions evolve symmetrically across disparate branches of life. Senior author Tse's suggestion that octopuses likely possess sophisticated internal representations of space — what he called an internal map — adds to a body of evidence that building a mental model of one's environment is not a vertebrate invention, but a general strategy that evolution has converged on multiple times across the tree of life.

This view challenges one of the most persistent assumptions in the history of cognitive science: that sophisticated intelligence is a product of specific neurological structures, particularly the vertebrate cortex, and that the cognitive capacities we associate with those structures are therefore exclusive to the lineages that possess them. The Dartmouth octopus study is one of a growing number of findings suggesting that intelligence is better understood as an adaptive strategy — something that emerges wherever the ecological pressures are strong enough and the evolutionary time is sufficient — rather than as the exclusive property of any particular brain architecture.

A Broader Rethinking of What Intelligence Means

The scientific community is increasingly grappling with the implications of findings like this one for how we define and measure intelligence across species. For most of the twentieth century, the study of animal cognition was heavily influenced by behaviourist traditions that were sceptical of attributing anything resembling thought to non-human animals, let alone invertebrates. The last several decades have gradually eroded that scepticism in vertebrates — it is now widely accepted that many mammals and birds have rich cognitive inner lives. What the Dartmouth study and related cephalopod research are now suggesting is that the same expansion of our understanding needs to extend to invertebrates as well.

The octopus represents an alternative architecture for cognition — distributed, decentralised, built from entirely different molecular components — that works brilliantly for the ecological niche it occupies. That such a different architecture can produce the same ability to translate a mirror reflection into successful spatial navigation suggests that the universe of possible minds is larger than science has traditionally assumed, and that the cognitive abilities we have been measuring may be far more widespread across the animal kingdom than our vertebrate-centric frameworks have been equipped to detect.

Conclusion

Three California octopuses in a laboratory at Dartmouth have quietly changed the science of intelligence. By learning to use a mirror to find hidden food — correctly 73% of the time, with improving speed — they have done something no invertebrate has ever been documented doing before. They have joined a very short list of animals, all previously vertebrates, that can take a reflection and use it to reason about the real world. The evolutionary distance between an octopus and a dolphin or a crow is almost incomprehensibly vast. And yet here they are, solving the same spatial problem the same way. That is not coincidence. That is convergent evolution doing what it does best: finding the same elegant answer to the same hard question, no matter what kind of brain is doing the thinking. The octopus did not just find the crab. It found something much bigger.

*This article draws on the peer-reviewed study "Octopus bimaculoides can learn to utilize a mirror to localize a reward outside the line of sight," published in Current Biology (2026) by Mary Kieseler, Peter Tse, and colleagues at Dartmouth College.*