Minimal Cognition

Where does cognition begin? Not where you think. Bacteria sense, remember, learn, communicate, and make decisions — and they've been doing it for billions of years before the first nervous system evolved. The line between "cognitive" and "merely metabolic" keeps getting pushed further down the tree of life, and each push makes it harder to maintain that consciousness requires a brain, or that there's a sharp boundary between the minded and the mindless.

The Bacterial Cognitive Toolkit

Lyon's review of bacterial behavior is a systematic demolition of the "no nervous system, no cognition" assumption.¹ E. coli cells have upward of 10,000 chemoreceptors per cell, clustered at the leading pole in a formation some researchers call a "nanobrain." These receptors integrate multiple information channels with non-linear responsivity, habituation, and adaptation — the same dynamical features that define neuronal sensory processing in animals.

The toolkit includes sensing, valence assignment (evaluating whether a situation is good or bad for survival), memory (bacterial memory was discovered in 1972 — Macnab & Koshland — yet cognitive scientists largely ignore it), learning, anticipation, decision-making, and communication. Each capacity has parallels not just functionally but at the molecular mechanism level — signal transduction pathways in bacteria provided the foundation for the information-processing approach that launched the cognitive revolution in psychology.¹

Lyon's key philosophical move is to define cognition biologically rather than anthropocentrically: cognition is what an organism uses to become familiar with, value, and interact with its environment in order to stay alive. Under this definition, bacterial chemotaxis isn't metaphorically cognitive — it's the same kind of process that, in creatures more like us, we'd readily label cognition without hesitation. As Popper quipped: "from the amoeba to Einstein is just one step."¹

There's even a "bacterial IQ" formula based on genome size and proportion of signal transduction genes, plus an introversion/extroversion measure based on how many of those systems face outward versus inward. Simplifying bacterial behavior to "run and tumble" and averaging over populations hides enormous cognitive complexity — like reducing all human behavior to "walking and standing" and measuring population averages.¹

Umwelt: The World as Experienced

Uexküll's concept of Umwelt adds a complementary lens, and Feiten's paper does important work distinguishing two very different things that go by the same name.²

Type 1 Umwelt (the deflated version): the subset of physical reality that an organism can detect. A tick's type 1 Umwelt is butyric acid, warmth, and hairless skin — three features it can sense. This is an engineering concept. It doesn't imply subjective experience. An elevator has a type 1 Umwelt (the buttons and weight sensor are its sensory interface).

Type 2 Umwelt (what Uexküll actually meant): a subjectively experienced world actively constructed by the organism. The organism is a subject, not a machine. The tick doesn't just detect butyric acid — it lives in a world where butyric acid means something. This is the radical claim, and it's routinely misrepresented. Dennett equates Umwelt with an elevator's "ontology," which is precisely the machine-theory view Uexküll was attacking.²

The tension between ecological psychology (Gibson: affordances are real features of a shared environment, waiting to be discovered) and enactivism (the organism constructs its meaningful world through interaction) maps onto the type 1/type 2 distinction and remains genuinely unresolved. Does the organism discover meaning in the world, or create it? This is one of the deepest questions in philosophy of mind, and it's not just about bacteria — it's about whether perception is fundamentally receptive or fundamentally creative.²

The Plant Kingdom

Plants are a fascinating test case for where to draw the cognitive boundary. Daniel Chamovitz argues that plants can see, smell, feel, communicate, and remember — and he's not being metaphorical.³

A plant "sees" not just for photosynthesis but to regulate development — the same group of genes Chamovitz discovered for light-sensing in plants turned out to regulate circadian light responses in animals too. Plants "smell" by detecting volatile chemicals: a ripe fruit releases ethylene, neighboring fruits detect it and begin ripening in synchrony. The parasitic plant dodder literally sniffs out its preferred host, choosing tomato over wheat by scent. Plants communicate through airborne pheromones (a maple under insect attack releases a chemical that primes neighboring trees to mount defenses) and through root-to-root signaling (a drought-stressed plant "warns" neighbors through shared root networks).³

Plant memory is real and multi-layered. A Venus flytrap remembers that one trigger hair has been touched and waits for a second — but only for about 20 seconds. Wheat seedlings remember winter before they'll flower (vernalization). Some stressed plants produce offspring that are more resistant to the same stress — transgenerational memory through epigenetic modification. The short-term memory in the flytrap is electricity-based, "much like neural activity." The longer-term memories are epigenetic — changes in gene activity that pass from parent to offspring without altering DNA sequence.³

Plants even share molecular machinery with animal nervous systems. They produce and respond to neuroactive chemicals, including glutamate receptors — the same receptors necessary for memory formation in the human brain. The same drugs that inhibit human glutamate receptors affect plants. Darwin proposed the "root-brain" hypothesis: that the root tip acts like the brain of lower animals, receiving sensory input and directing movement. Modern research groups are following up.

Chamovitz draws the line at thinking: "To me thinking and information processing are two different constructs." He dismisses "plant neurobiology" as ridiculous terminology — "plants do not have neurons just as humans don't have flowers." But he acknowledges that plants exhibit "anoetic consciousness" and that the cell-to-cell communication throughout the plant is functionally analogous to neural processing. The question isn't whether plants process information — they obviously do, with enormous sophistication. It's whether the processing involves anything we'd recognize as experience.

Umwelt Extended: The Social Sphere

Bueno-Guerra's extension of Uexküll's framework proposes a third component alongside the Merkwelt (perceptual world) and Wirkwelt (motor world): the Sozialwelt — the social sphere of each species.⁴ The proposal arose from a practical problem in comparative cognition research. Cooperative experimental setups that seemed natural from a human perspective failed consistently with chimpanzees. The breakthrough came when researchers switched to competitive paradigms, which matched chimpanzee social dynamics far better. The species hadn't failed at cognition — the experimenters had failed at Umwelt.⁴

The most revealing example: an experiment on moral evaluation in chimpanzees presented "good" and "bad" experimenters — someone who consoled a victim versus someone who hit a third party. Adult chimps chose as humans would, preferring the consoler. But young males consistently chose the aggressor. The researchers realized that adolescent male chimps, who need strong allies for future dominance contests, were correctly reading the "bad" experimenter as a powerful potential ally. The same behavior looked immoral from one Sozialwelt and perfectly rational from another.

This connects back to the bonobo "jerk preference" finding elsewhere in this article. The lesson is the same: cognitive evaluation can't be separated from social ecology. What looks like a failure of prosocial cognition may be a success of status cognition operating in a different social world. The Sozialwelt proposal formalizes what comparative psychologists have been learning the hard way — you can't test an organism's mind without understanding the social world that mind evolved to navigate.

Cognition Before Neurons

If bacteria can be cognitive, what about multicellular animals that never evolved a nervous system? Trichoplax adhaerens — a millimeter-wide blob with no neurons, no muscles, and practically the smallest genome in the animal kingdom — turns out to be a stunning test case.⁵

Prakash and Bull at Stanford spent a decade studying how this creature moves, and the answer is pure physics. Millions of cilia on its underside don't swim through fluid — they walk, adhering to surfaces and popping off in a mechanical gait that emerges naturally from the interplay between ciliary driving forces and surface adhesion. No central controller tells them what to do. The walking behavior is an excitable system, with dynamics that map directly onto models of action potentials in neurons. Small perturbations in height (rather than voltage) trigger cascading activity changes, nonlinear and self-amplifying — "neuroscience without neurons," as Prakash calls it.⁵

The ciliary coordination is even more remarkable at scale. Mechanical interactions transmit information across the organism as waves of synchronized orientation. The cilia flock like starlings — reorienting collectively in fractions of a second — despite being locked into fixed cellular positions (a "solid flock," unlike the fluid flocking of birds or fish). And the system is selective: some stimuli propagate and alter the organism's behavior, while others simply dissipate. The organism effectively filters signals, responding to some and ignoring others — a property we usually attribute to nervous systems.⁵

This is philosophically potent. It suggests that the functional properties we associate with neural systems — excitability, signal propagation, selective filtering, coordinated behavior — can be achieved by mechanics alone. The nervous system, when it eventually evolved, may not have invented these capabilities so much as inherited them from an older, mechanical substrate. As one researcher put it: "What the nervous system is doing may not be what we thought it was doing."

The High End: Dolphins and the Play Mystery

The continuity from bacteria to neurons gets tested differently at the other end — in animals whose cognitive sophistication embarrasses simple explanations. Kelly the dolphin at the Institute for Marine Mammal Studies hides paper under a rock, tears off small pieces to maximize fish rewards, and then uses leftover fish to bait gulls for even more fish. She taught the strategy to her calf, who taught other calves. This isn't stimulus-response. It's multi-step planning with deferred gratification, innovation, and cultural transmission.⁶

Dolphins use tools (sponges on snouts to protect against stonefish), invent feeding strategies adapted to specific habitats (mud-ring fishing in Florida, kelp-ambush hunting in Patagonia, shrimp-net raiding in Texas), and respond correctly to novel sentences in an artificial sign language — including sentences they've never heard before, which suggests genuine comprehension rather than trained associations. They pass the mirror self-recognition test, which was thought to indicate self-awareness.⁶

And then there's play, which remains one of the deepest unsolved problems in animal behavior. The intuition that play-fighting teaches combat skills, or that play builds social bonds, or that play "prepares for adulthood" — none of it holds up to rigorous testing. Lynda Sharpe spent five years tracking 45 wild meerkat pups and found that play-fighting didn't predict real fighting success, play didn't reduce aggression, play didn't strengthen social bonds, and dispersing meerkats didn't preferentially team up with favorite playmates.⁷

Yet play persists across species at enormous cost — frolicking bighorn lambs impale themselves on cacti, fur seal pups get snatched by sealions. Natural selection should have eliminated it if it were useless. The most compelling evidence comes from rats: play-deprived rats become adults who can't regulate their social behavior, either erupting in rage or cowering in corners. Play-enriched rats develop larger cerebral cortices with more neural connections and learn faster. The mechanism may be stress calibration — play activates the same neurochemical pathways as stress, and experiencing manageable stress during play tunes the stress response for later life.⁷

The Alien Mind: Octopus Intelligence

If dolphins represent cognitive sophistication within the vertebrate lineage, octopuses represent something more philosophically unsettling: convergent evolution of intelligence from an entirely alien starting point. Our last common ancestor, half a billion years ago, had at most a few neurons. "The same thing that got them their smarts isn't the same thing that got us our smarts," says Mather, "because our two ancestors didn't have any smarts."⁸

The octopus brain has about 130 million neurons, but three-fifths of them are in the arms, not the brain. Each arm operates semi-autonomously — a severed arm will crawl away, seize food, and try to pass it to where the mouth would be. The cognitive architecture is distributed in a way that has no vertebrate parallel. Peter Godfrey-Smith suggests we may "have to change the way we think of the nature of the mind itself to take into account minds with less of a centralized self."⁸

The behavioral evidence for flexible intelligence is hard to dismiss. Octopuses in labs learn to open childproof pill bottles, recognize individual humans months apart, and hold grudges — one at the New England Aquarium would soak a specific volunteer with its water jet every time she visited, while being gentle with everyone else. One octopus bounced a pill bottle back and forth with directed water jets twenty times, meeting all scientific criteria for play behavior.⁸

Mather's explanation for why octopuses evolved intelligence is elegant: it was the loss of the ancestral shell. No shell means versatile hunting but also vulnerability to everything. Each prey demands a different strategy (camouflage? speed? ambush?), each predator a different escape (color change? ink? rock fortifications?). The cognitive demands of being simultaneously a versatile predator and an unprotected soft body drove general-purpose intelligence — not through social complexity, as in our lineage, but through ecological complexity.⁸

This matters for the continuity question because it shows that the road from bacteria to flexible intelligence doesn't run through a single neural architecture. Centralized vertebrate brains and distributed octopus nervous systems converge on the same abstract capability — flexible problem-solving — from radically different starting points. If minds can be non-neural (bacteria), mechanical (Trichoplax), centralized (dolphins), or distributed (octopuses), then what cognition is can't be defined by its substrate. It's something about the functional problem being solved.

Social Cognition and Its Limits

One of the sharpest tests of where general cognition shades into something specifically human comes from social evaluation. Human infants as young as three months prefer individuals who help others over those who hinder them. This looks like a deep, possibly innate feature of human social cognition.⁹

Bonobos — our closest relatives along with chimpanzees — fail this test spectacularly. At the Lola ya Bonobo Sanctuary, bonobos consistently chose the "jerks": characters who hindered others, hogged resources, or stole toys. They liked the mean individuals more after watching them behave badly. The researchers suspect bonobos read dominance displays as signals of social power and prefer to ally with powerful partners regardless of prosocial behavior.⁹

If that's right, something specific happened in the human lineage — an aversion to antisocial behavior that functions as both partner selection (avoid bad collaborators) and social enforcement (the threat of rejection keeps potential cheaters in line). This preference may be "at the heart of why we're so cooperative," enabling humans to work in large groups with strangers in ways no other primate does.⁹

This is a useful corrective to the temptation to see human-like cognition everywhere. Lyon's bacteria have genuine cognitive capacities. Octopuses have genuine intelligence. Dolphins plan and innovate. But the specific kind of social cognition that enables large-scale human cooperation appears to be genuinely unusual. The continuity thesis is right that cognition is everywhere, but wrong if it implies that all cognition points in the same direction. Bacteria optimize for survival. Octopuses optimize for ecological flexibility. Bonobos optimize for social power. Humans optimize for cooperative reputation. Same underlying capacity — valence, learning, decision-making — channeled by very different selection pressures into very different cognitive styles.

The Gene-Behavior Layer

There's a dimension of cognition that the neuron-centric view mostly misses: genes aren't just the blueprint for building neural hardware; they're running software in real time. Gene Robinson's genomic research reveals that emotions and behavior are shaped by a second layer of organization — gene activity patterns that change dynamically in response to experience.¹⁰ This gene-behavior coupling has been demonstrated in organisms far simpler than humans: in honeybees, the transition from nurse to forager involves thousands of gene expression changes in the brain, and these changes precede the behavioral shift rather than following it. The genes aren't just building the hardware; they're part of the operating system.

This matters for the continuity question because it pushes cognition down another level. If Lyon's bacteria are cognitive because they process information through signal transduction pathways, and if those same pathways evolved into the gene-regulatory networks that shape behavior in animals, then the divide between "genetic" and "neural" cognition is artificial. The brain is a two-layer system — neurons and genes working together like hardware and software — and the gene layer connects seamlessly back to the molecular cognition that bacteria have been doing for billions of years.¹⁰

The Continuity Challenge

What connects Lyon's bacteria, Prakash's Trichoplax, Montgomery's octopuses, and Kelly the dolphin is a shared rejection of the "machine theory of living beings." At every level — bacterial chemotaxis, ciliary mechanics, distributed arm intelligence, cetacean innovation — organisms aren't stimulus-response automata. They're flexible responders that assess the significance of signals relative to their own internal state. The concept of valence — assigning existential value to a summary of current circumstances — appears at the bacterial level as chemotactic preference, at the Trichoplax level as selective signal propagation, at the octopus level as ecological problem-solving, and at the dolphin level as multi-step strategic reasoning.¹

If cognition really is continuous from bacteria to humans, the Hard Problem Of Consciousness gets both easier and harder. Easier: maybe experience is everywhere, and the mystery isn't how it arises from non-experience but how it complexifies. Harder: now you need to explain experience in a cell with no nervous system at all. And if mechanical systems can produce the functional signatures of neural computation — as Trichoplax suggests — and if radically different neural architectures can converge on flexible intelligence — as octopuses demonstrate — then the question isn't just "where does cognition begin?" but "what is a nervous system actually adding?" The cognitive spectrum isn't a ladder. It's a bush, branching in many directions, with humans occupying one twig among many.