You have never perceived the present moment. Not once.
Light reaching your retina takes tens of milliseconds to be transduced into a neural signal. That signal travels to thalamus and then to cortex, accumulating delay at every synapse. Processing to the point where the information influences a decision adds more. By the time anything reaches the level of experience, the event that caused it has been over for something like a tenth of a second. Sound arrives on a different schedule, touch from your foot on a different one again, and the nervous system assembles all of it into an experience that feels simultaneous and current.
That is a manufactured impression, and manufacturing it is one of the more demanding jobs a nervous system does. The gap has to be hidden, the channels have to be aligned despite arriving at different times, and the animal has to act on information that is already stale by the time it exists. Every animal with a nervous system faces this, and neural time is the set of solutions.
The solutions vary enormously, because the physics does. A blue whale and a fruit fly are running the same basic hardware at wildly different scales, and the consequence is not that one is faster than the other in some vague sense. It is that they occupy measurably different temporal worlds, with different sampling rates, different delays, and different definitions of what counts as an instant.
Three separate things get called timing and they are worth keeping apart from the start. There is the delay problem, which is about how long information takes to arrive and what to do about it. There is temporal resolution, which is about how finely an animal can distinguish events in sequence. And there is timing proper, meaning the measurement of durations, which spans from microseconds to years and runs on several unrelated mechanisms. Almost every confusion in this subject comes from treating those three as one, and neural time is best understood as three problems that happen to share a vocabulary.
Why signals are slow, and what it costs to speed them up
An action potential is not electricity in a wire. It is a wave of ion flux across a membrane, regenerated at every point along the axon, and its speed is set by physics that does not scale kindly.
Conduction velocity in an unmyelinated axon increases with the square root of diameter, which is a terrible exchange rate. Doubling speed requires quadrupling diameter, and diameter costs volume, and volume is the scarcest resource in a nervous system. That relationship is why unmyelinated axons top out around one meter per second, or slower, meaning a signal from a large animal’s foot would take an appreciable fraction of a second to reach the brain.
Two solutions exist and both evolved more than once. The first is gigantism: build one enormous axon and accept that you can only afford a few. Squid escape responses run on giant axons over half a millimeter in diameter, visible without magnification, which is precisely why the squid became the preparation on which the action potential was first characterized. Nobody could have put an electrode inside a mammalian neuron in 1939. Essentially everything known about how any neuron works, in any animal, was worked out first on an invertebrate whose wiring happened to be thick enough to push a wire into, which is a good reminder that the history of neuroscience is partly a history of which preparations were technically tractable.
The second is insulation. Myelin wraps the axon in layers of glial membrane, leaving periodic gaps, and the impulse jumps between gaps rather than being regenerated continuously. Saltatory conduction reaches speeds comparable to giant axons in fibers less than a hundredth the diameter, and the comparative account of rapid conduction through giant axons and myelinated fibers establishes that it arose independently in vertebrates, in annelids, and in crustaceans. The fastest conduction ever recorded in any animal was measured in a shrimp.
The tradeoff is stark. A squid can have a handful of fast channels. A vertebrate can have millions, because myelin buys speed without volume. That single difference underwrites the possibility of a large brain with fast long-range connections, and it is a good example of a molecular innovation setting an architectural ceiling.
There is a cost nobody mentions. Myelin is metabolically expensive to build and maintain, it takes years to complete in humans, and it is what multiple sclerosis destroys. Speed is purchased and the payments continue.
The delays that remain are not small and they are not uniform. A signal from a giraffe’s foot has more than two meters to travel. Conduction speeds within a single brain vary by more than an order of magnitude depending on fiber type, which means information arriving at the same destination from different sources arrives at different times as a matter of routine. Some of that variation is exploited rather than tolerated: axon diameter and myelin thickness appear to be tuned so that signals from different distances arrive together where synchrony matters, which means conduction delay is a design parameter rather than only a nuisance.
The animals that see faster than you
The most legible measure of temporal resolution is critical flicker fusion frequency: the rate at which a flashing light stops looking like flashes and starts looking continuous. Below your threshold you see flicker; above it you see a steady lamp. Humans sit around sixty hertz, which is why film at twenty-four frames per second works with the right shutter and why old fluorescent lighting bothered some people.
The comparative range is roughly two orders of magnitude. Leatherback sea turtles come in near fifteen hertz. Common cuttlefish around forty-two. Crayfish around fifty-three. Chickens run somewhere in the high eighties to a hundred. Peregrine falcons have been measured near one hundred and twenty-nine. Some flying insects reach into the hundreds.
The pattern turned out to be predictable. A comparative analysis linking metabolic rate and body size to temporal resolution found that critical flicker fusion frequency rises with mass-specific metabolic rate and falls with body mass, across a phylogenetically broad vertebrate sample. Small, fast-metabolizing animals sample the world more finely. Large, slow ones sample it more coarsely.
The mechanism is not mysterious. Temporal resolution is expensive: sampling faster means more photoreceptor turnover, more transduction cycles, more neural throughput, more energy. It only pays if the animal can act on the extra information, and reaction time is constrained by body size, since a large animal cannot change direction quickly no matter how fast it perceives. A whale gains nothing from a two-hundred-hertz visual system because it cannot do anything in five milliseconds. A fly gains everything.
Light level does substantial work in the same equation. Faster sampling means integrating photons over shorter windows, which means fewer photons per sample and a worse signal-to-noise ratio, which is only tolerable in bright conditions. Deep-sea and nocturnal species accordingly show low thresholds, trading temporal resolution for sensitivity, and the same animal’s threshold drops as light falls. That tradeoff between speed and sensitivity is one of the most reliable in sensory biology and it is the same exchange rate that governs every other design decision in a sensory system.
That reframes the fly-swatting problem correctly. A fly is not fast because its muscles are fast, though they are. It is fast because your hand’s approach is being sampled at a rate that makes it look, from the fly’s side, like a slow deliberate gesture with plenty of time to evaluate.
What flicker fusion does and does not tell you
Here is where the popular version outruns the evidence, and the overreach is worth dismantling because the underlying finding is good.
The Healy result got reported as small animals experience time in slow motion, which is a claim about subjective experience derived from a measurement of visual sampling rate. Those are different things, and the inference has real problems.
First, flicker fusion is a property of a sensory system, not of a whole animal. Different modalities in the same animal have different temporal resolutions, and auditory temporal resolution in many animals vastly exceeds visual. There is no single clock speed.
Second, the threshold varies within an individual with light level, with the region of the retina being tested, with temperature in ectotherms, and with arousal. A number quoted for a species is a number from particular conditions.
Third, and most importantly, the step from processing rate to felt duration is a leap. It is possible that an animal sampling at two hundred hertz experiences each second as longer. It is also possible that subjective duration is set by something else entirely, or that the question does not have a determinate answer for animals whose capacity for experience is itself unresolved. This connects directly to the problem of measuring anything about animal experience from the outside, and flicker fusion is a physiological measurement being asked to carry a philosophical load.
The defensible claim is narrower and still striking: animals differ enormously in how finely they can resolve events in time, that variation is predictable from body size and metabolism, and it constitutes a dimension of ecological niche that nobody thought to measure until recently. What it means from the inside is a separate question, and it belongs with the rest of the architecture of sensory worlds rather than being settled by a threshold measurement.
The practical consequences are real regardless of how the experiential question resolves. Flicker from artificial lighting invisible to humans is visible to many birds and insects, and lighting installations designed against human thresholds are perceptibly strobing to the animals living under them. Screen refresh rates that look continuous to us do not to a parrot in the room. Poultry housing lit at frequencies below avian flicker fusion has been implicated in stress responses. That is an entire category of environmental effect that was invisible until somebody measured the threshold in the animal rather than assuming ours.
Compensating for your own lag
A nervous system that simply acted on delayed information would be catastrophically bad at anything involving a moving target. The solutions are worth cataloguing because they are the most underappreciated engineering in neuroscience.
Prediction is the main one. Rather than reporting where a moving object is, visual systems extrapolate where it will be, using velocity to project forward by roughly the amount of the delay. The consequence is a family of illusions, including the flash-lag effect, in which a briefly flashed object appears to trail a moving one that is physically alongside it, because the moving object was extrapolated and the flash could not be.
Efference copy handles the self-motion problem. When a motor command is issued, a copy goes to sensory areas so the resulting sensory change can be predicted and discounted. That is why the world does not appear to lurch when you move your eyes, why you cannot tickle yourself effectively, and why electric fish can distinguish their own discharge from a neighbor’s.
Delay lines are the elegant solution for computing time differences directly. The classic case is sound localization in the barn owl, where axons of different lengths from the two ears converge on coincidence-detector neurons, so that a particular neuron fires only when signals from each ear arrive simultaneously, which happens only for a particular interaural delay. The owl computes direction by turning a timing difference into a place code, and it does so with microsecond precision using components that are individually far slower. That last clause is the general principle worth extracting: a population of imprecise elements can compute with precision no single element possesses, because averaging across many noisy detectors sharpens the estimate. Microsecond timing in a system whose action potentials last a millisecond is not a paradox. It is what populations are for.
Alignment across modalities is the last problem and the strangest. Light arrives essentially instantly; sound takes about three milliseconds per meter. Neural processing is faster for sound than for vision. The two errors partially cancel, producing a horizon of roughly ten to fifteen meters within which audiovisual events are perceived as simultaneous, and outside which they separate. The nervous system also recalibrates: expose someone to a consistent lag between an action and its consequence and the perceived simultaneity shifts.
Neural time also gets manipulated deliberately in one place worth noting, which is signal design. Animals that communicate with rhythmic displays are transmitting into a receiver with a known temporal resolution, and signals evolve to sit inside it. Fireflies flash at species-specific intervals. Cockatoos that drum with manufactured sticks produce individually distinctive rhythms. Any display running faster than the receiver’s sampling rate is wasted, and any running slower is inefficient, which means a signaller’s tempo is a readout of its audience’s temporal grain.
Every one of these is a workaround for a delay that cannot be eliminated. The unified present is an achievement, not a given.
Interval timing, and the clock nobody can find
Estimating seconds to minutes is a separate system from both circadian rhythm and millisecond timing, and it appears everywhere from insects to primates.
Its signature is scalar variability. The error in an animal’s estimate scales proportionally with the interval being estimated, so timing ten seconds is roughly twice as variable as timing five. That proportionality holds across species and across enormously different durations, which is a strong constraint on mechanism and the reason a single explanation has been sought for decades.
The classical account posits a pacemaker emitting pulses, an accumulator counting them, and a comparison against stored values. It reproduces scalar variability under some assumptions and has never been located anatomically. The striatal beat frequency model proposes instead that populations of cortical neurons oscillating at different frequencies produce a pattern that is unique at each moment after a start signal, with striatal neurons learning to detect the pattern corresponding to a reinforced duration. Other accounts hold that timing emerges from the natural dynamics of recurrent networks without any dedicated clock, so that time is read off the trajectory of a population’s state rather than counted.
The evidence pulls in several directions at once. Dopamine manipulations shift timing systematically, which supports a pacemaker-like component. Time cells that fire at particular moments during a delay have been found in hippocampus, tiling elapsed intervals the way place cells tile space, which supports the population-trajectory view and connects timing directly to the machinery that handles spatial representation. That overlap is not incidental. The same structures that tile space appear to tile elapsed duration, and the same offline replay that compresses spatial trajectories runs at roughly twenty times real speed, which means a system built to represent where things are is also representing when, at a rate decoupled from the events themselves. Timing deficits appear with cerebellar damage for short intervals and with basal ganglia damage for longer ones, suggesting multiple systems rather than one.
The honest summary is that interval timing is behaviorally well characterized and mechanistically unresolved, and that the search for a single clock has probably been the wrong search.
The comparative breadth is the part that constrains any eventual answer. Scalar timing has been demonstrated in pigeons, rats, primates, fish, and honeybees, which means whatever produces it either evolved once very early or is a generic property of the kind of dynamics nervous systems run. Bees time intervals well enough to exploit flowers that produce nectar at particular hours, returning at the right time on subsequent days. Caching birds track how long ago each cache was made and adjust recovery accordingly, which requires timing on the scale of days rather than seconds and is a different system again.
The clock that runs without a brain
Circadian rhythm is the timing system with the cleanest mechanism, and it is the only one that runs at the level of individual cells.
The molecular core is a transcription-translation feedback loop: clock genes produce proteins that accumulate, enter the nucleus, and suppress their own transcription, with degradation eventually releasing the suppression and restarting the cycle. The loop takes roughly twenty-four hours, and it is self-sustaining, continuing in constant darkness with a period near but not exactly a day, which is why it needs daily resetting by light.
The homology is partial and the convergence is instructive. Animals, fungi, plants, and cyanobacteria all run circadian clocks, but the specific genes differ substantially between kingdoms, meaning the twenty-four-hour loop has been built more than once. The cyanobacterial clock is the most striking, since its core can be reconstituted in a test tube from three purified proteins plus energy, producing a twenty-four-hour phosphorylation rhythm with no cells, no transcription, and no translation involved. That is a biological clock reduced to chemistry, and it is the strongest available demonstration that timekeeping is not a nervous-system capability that simpler organisms approximate. It is a chemical capability that nervous systems inherited.
In mammals a hypothalamic nucleus acts as master pacemaker, entrained by light through a dedicated retinal pathway, and it synchronizes peripheral clocks running in essentially every tissue. Liver cells keep time. So do skin cells. The organisms with no nervous system at all run these clocks too, which is a reminder that timekeeping is a cellular capability that nervous systems coordinate rather than invent.
Non-circadian biological clocks exist alongside it: tidal rhythms in intertidal animals, lunar rhythms governing coral spawning, and annual rhythms driving migration and hibernation that persist in constant conditions for years. Ground squirrels held in constant temperature and darkness continue to enter and exit hibernation on an approximately annual schedule, which is a clock with a period so long that no individual molecular loop could plausibly be counting it directly.
The circadian system also does something that matters for the rest of this subject, which is set the temporal frame within which everything else happens. Learning, memory consolidation, sensory thresholds, and reaction times all vary systematically across the day, which means any experiment measuring an animal’s capability is measuring it at a phase. Nocturnal animals tested during their subjective night perform differently from the same animals tested at the other end of the cycle, and a substantial amount of older comparative work did not control for it.
Neural time as the medium, not the message
The deepest point about neural time is that timing is not just something nervous systems measure. It is frequently what they use to represent things.
Rate coding, where information is carried by how often a neuron fires, was the default assumption for decades and is genuinely used. But temporal coding, where the precise timing of individual spikes carries information, is now well established in several systems. Auditory neurons phase-lock to sound waveforms, firing at a consistent point in each cycle, which preserves fine temporal structure that a rate code would discard. Some sensory systems appear to encode stimulus intensity in latency, with stronger stimuli producing earlier spikes, which is a remarkably fast code since the answer is available from the first spike.
Oscillations organize the traffic. Neural populations oscillate at characteristic frequencies, and the phase of an ongoing oscillation determines when inputs are likely to be effective, which effectively gates communication between regions in time. The hypothesis that structures communicate by aligning their oscillatory phases is one of the more productive ideas in current systems neuroscience and one of the less settled.
Spike-timing-dependent plasticity is where timing becomes structural. Whether a synapse strengthens or weakens depends on the order and interval of pre- and postsynaptic firing, with a window of tens of milliseconds: fire before the target and the connection strengthens, fire after and it weakens. That is causality detection implemented in chemistry, and it means the temporal precision of the whole system is what allows learning to encode which events led to which. The window is the interesting constraint: it is roughly the width of a few tens of milliseconds, which happens to be about the timescale on which physically causal events in an animal’s environment actually occur. A system with a much wider window would associate things that had nothing to do with each other. A much narrower one would miss real causal chains that take time to unfold.
The synaptic changes that store what an animal learns are therefore not merely fast. They are timing-dependent in a way that makes the millisecond structure of activity the thing that determines what gets remembered.
Time in the sensory periphery
Different senses run on different temporal scales, and the differences shape what each one is good for.
Auditory temporal resolution is the finest in most animals, because sound is a temporal signal by nature. Humans detect interaural time differences down to around ten microseconds, which is a smaller interval than the duration of an action potential and is achieved by populations rather than single cells. Bats push further, resolving echo delay differences on the order of a microsecond, which corresponds to distance differences under a millimeter and is the active sensing case where timing is the entire perceptual channel. Toothed whales run the same computation in a medium where sound travels four and a half times faster, which compresses every delay accordingly and means their timing machinery has to be correspondingly finer for the same spatial resolution.
Vision is comparatively slow, limited by the photochemical cascade in photoreceptors, which is why flicker fusion tops out where it does and why fast events blur.
Olfaction is slower still and operates on a completely different temporal logic, since molecules arrive by diffusion and turbulence rather than by propagation. A chemical signal encodes history rather than instantaneous state, which is why a dog reading a scent trail is reading the past and can determine direction of travel by comparing the age of successive marks.
Electroreception is essentially instantaneous, which sounds like an advantage and creates a specific problem: with no travel time there is no delay to measure, so distance has to be inferred from the shape of a field distortion rather than read off a clock. The animals that generate a probe and wait for it to come back have the opposite arrangement, getting range for free from delay, which is the single largest advantage of an acoustic probe over an electrical one.
The consequence is that an animal integrating across senses is combining channels that disagree about when things happened, and the unified perceptual world it experiences is stitched from streams with incompatible temporal geometries.
Living on different schedules
Zoom out from milliseconds and the temporal structure of a life is itself a variable.
Metabolic scaling produces a striking regularity: across mammals, lifespan and heart rate vary inversely such that total heartbeats over a lifetime cluster loosely around a billion. A shrew’s heart runs at over a thousand beats per minute for a couple of years. A large whale’s runs at single digits for the better part of a century. The relationship is loose and has real exceptions, with bats and naked mole rats living far longer than their metabolic rate predicts and humans exceeding the primate trend, but the pattern is strong enough to have driven a great deal of aging research.
The developmental consequence matters more for cognition than the lifespan number does. An animal with a long life and slow development gets an extended window in which experience can shape circuitry, which is what makes accumulated knowledge worth acquiring. Elephants gestate for twenty-two months and remain dependent for years, great apes have prolonged juvenile periods, and large parrots live for decades. Human synaptic pruning in frontal regions continues into the late twenties, which extends the window in which circuitry remains substantially shapeable by experience far past the point of physical maturity.
The contrast case is the one that makes the point. Most octopuses live one to two years, reproduce once, and die on an endocrine schedule, which forecloses accumulation entirely and is a decent part of why that lineage bet on within-lifetime proteomic flexibility instead.
The claims that do not hold up
An audit, because temporal claims about brains circulate with unusual confidence.
Small animals see the world in slow motion overstates a real finding, for the reasons above. They sample faster. What that is like is not established.
Humans use vision at thirty frames per second, or sixty, or any single number, misunderstands the system. There is no frame rate. Different aspects of vision have different temporal properties, flicker sensitivity varies with conditions, and people detect single-frame anomalies at rates well above any quoted figure.
Reaction time measures nerve speed conflates conduction with processing. Nerve conduction is a small fraction of a typical reaction time; the rest is decision and motor preparation.
The brain processes information at a fixed rate treats a heterogeneous system as a single processor. Different circuits run at wildly different speeds in the same brain, and the notion of a single clock speed is imported from computing rather than derived from biology.
Neural time runs at one speed in a given animal is wrong in the same way. Conduction velocity varies with temperature, which matters enormously for ectotherms, so a lizard’s nervous system is measurably slower in the cold and its behavioral repertoire narrows accordingly. Timing in fish and reptiles is a function of ambient temperature in a way it simply is not for a mammal, which is an underappreciated advantage of endothermy.
Time slows down in emergencies is not supported by the strongest test. Experiments dropping people in free fall and testing perception of a rapidly changing display found no improvement in temporal resolution, indicating the effect is a memory phenomenon, where denser encoding makes the interval seem longer in recall, rather than a perceptual one.
Everyone has an internal clock accurate to the second overstates interval timing, which is scalar and therefore proportionally worse for longer intervals.
The heartbeat-lifespan constant is the most quoted version of the metabolic scaling story and it is a loose regularity rather than a law. Bats, birds, and naked mole rats live several times longer than their metabolic rates predict, humans exceed the primate trend substantially, and the exceptions are numerous enough that treating a billion beats as an allowance is a metaphor rather than a finding.
Circadian rhythm is exactly twenty-four hours is false in constant conditions, where the free-running period differs from twenty-four hours in most individuals, which is precisely why entrainment by light is required.
What neural time is actually telling us
The unifying point is that a nervous system does not have access to the present, and every animal is solving that problem with the materials available.
The physics is unforgiving. Signals travel at speeds set by membrane biophysics, they can be accelerated only by spending volume or by wrapping insulation around them, and every synapse adds delay. That constraint shaped the architecture: myelin permitted large brains with fast long connections, giant axons permitted fast escape in animals that could not afford myelin, and both solutions were found independently more than once because the physical problem is the same everywhere.
The compensation is where the sophistication lives. Prediction, efference copy, delay lines, and cross-modal recalibration are all mechanisms for producing a coherent, apparently current, apparently unified experience out of stale information arriving on incompatible schedules. The seamlessness is the artifact.
It also has a clinical corollary that makes the machinery visible when it breaks. Disruptions of temporal processing show up in specific disorders rather than as general slowness, with timing deficits documented in Parkinson’s disease, cerebellar damage, and schizophrenia, and with the sense of agency depending on the perceived interval between an action and its consequence. Stretch that interval artificially and people stop feeling that they caused the outcome. Agency, on the evidence, is partly a timing judgment.
And the scale-dependence is the part with the widest implications. Temporal resolution tracks body size and metabolism, which means a predator and its prey frequently occupy different temporal grains, and the mismatch is part of what determines who wins. The predator-prey pairings that look evenly matched on any other axis are frequently not evenly matched on this one, and temporal resolution has been proposed as a dimension of niche differentiation that comparative ecology mostly has not measured. It also means that comparing animals on any cognitive dimension without accounting for the timescale they operate on is comparing measurements taken with different instruments.
The 24-lecture Neurozoology course works the tree of life on exactly that principle, alongside the study of how knowledge moves between animals, the first edition’s survey of nervous systems, and the working animals whose capacities were discovered by the people relying on them. The insects and reef animals operating at temporal grains no vertebrate can match and the collective systems whose timing runs on chemical rather than neural propagation sit at opposite ends of the same continuum.
A fly evades your hand because the two of you are not in the same moment. The thing worth carrying out of this is that neural time is not a detail of implementation sitting underneath the interesting questions. It is the constraint that produced the architecture. Myelin exists because signals are slow. Prediction exists because signals are slow. Populations exist partly because precision beyond any single element requires them. And the differences between animals on this axis are large enough that comparing two species on any cognitive measure, without asking what temporal grain each one operates at, is comparing readings from instruments with different sampling rates and pretending the numbers are commensurable.
A fly evades your hand because the two of you are not in the same moment. Neither of you ever was, and the fact that it feels otherwise is a hundred milliseconds of very good engineering.