Do Cats Dream?

FĒLIS Editorial Feb 2026 18 min read

Late at night, a cat curls up in the corner of a sofa, paws twitching lightly, whiskers trembling at irregular intervals, throat occasionally squeezing out a muffled whimper. Anyone who has witnessed this scene thinks the same thing.

Based on existing neuroscience evidence, cats dream.

Jouvet, 1965

French neuroscientist Michel Jouvet performed targeted lesion surgery on cats' brainstems, destroying the sublaterodorsal nucleus in the pontine tegmentum. This nucleus sends inhibitory signals to spinal motor neurons during REM sleep, creating a state called REM atonia that keeps the body paralyzed while dreaming. Jouvet removed it.

Those cats stood up during REM sleep.

Arched backs, pouncing, chasing nonexistent prey in the air, executing complete predatory action sequences. Eyes closed the entire time, EEG confirming they were still in deep sleep.

Popular science articles usually stop here. What Jouvet and later Adrian Morrison recorded was far richer: complete territorial aggression sequences (arched back, sideways posture, hissing), defensive startle responses (sudden retreat, fur standing on end), and body language patterns that only appear in social contexts. Aggression, defense, social behavior, alternating across different REM cycles in the same cat. The brain during sleep was not running along a single track executing one fixed program, but switching between multiple behavioral domains.

This happened sixty years ago and remains the most direct set of behavioral evidence on the question of whether cats dream. Subsequent neuroimaging studies added extensive detail but did not surpass it in decisiveness.

Cat Dreams Are Probably Not Visual

Everything discussed below should be read with this premise in mind.

Human dreams are built around visual narrative. Cats have far lower cone cell density than humans, with visual resolution roughly one-tenth to one-fifth of ours. Their olfactory receptor gene count is approximately three times that of humans. The vomeronasal organ is highly developed. Hearing extends up to 64kHz, and the pinnae can rotate independently 180 degrees for sound localization.

Cats are animals that understand the world primarily through smell and hearing even when awake. Vision ranks lower in their sensory hierarchy. So when the brain reactivates daytime sensory memories during REM sleep, the channels called upon first are most likely olfactory and auditory. A cat dreaming of hunting may not have the prey's visual outline at the center of the dream, but rather the concentration gradient of scent molecules from the prey's body heat, or the directional shift of some high-frequency sound source.

REM in the Womb

Fetal cats already show substantial REM sleep activity inside the uterus. Newborn kittens have REM proportions exceeding 90% of total sleep time. Adult cats are around 25% to 30%.

A kitten that has not yet opened its eyes, has never seen prey, has never heard birdsong, already has a brain densely executing REM programs. There cannot be any "replay" in there.

The brain is using spontaneous neural activity during REM to calibrate motor circuits and sensory pathways. The cat's predatory action sequence involves millisecond-level coordination between proprioception, visual tracking, vestibular balance, and fine motor control, and the connections between these circuits need to establish correct synaptic weights. REM sleep allows the brain to generate signals on its own without external input, testing whether connections are working properly.

This part has been written about many times. Where few people continue further is this: kittens drop from 90% REM to the adult 25% to 30%, and this decline curve is not linear. It drops fastest in the first few weeks after birth, then gradually levels off. The shape of this curve closely overlaps with the timeline of synaptic pruning. Synaptic pruning is the process by which the brain actively eliminates excess neural connections during development, and its peak also falls in the first few weeks after birth.

Two curves overlapping. REM in the early period may not only be "testing" circuits but also providing the basis for synaptic pruning decisions: using spontaneous activation to tag which connections are active and which are redundant, so that the pruning process retains the former and eliminates the latter. Howard Roffwarg proposed the "ontogenetic hypothesis" in 1966, arguing that REM sleep promotes nervous system development in the neonatal period. Subsequent research has broadly supported this framework. The more specific sub-hypothesis that "REM provides tagging information for synaptic pruning" has not been directly verified. It is included here because it can explain why this decline curve has that particular shape.

Kittens' enormous sleep quantities are not laziness. The brain's hardware is undergoing a large-scale round of screening and reorganization, and REM sleep serves as both the tool and the venue for that screening.

The function of dreaming at the earliest stage of life is not recollection. It is construction.

The Cat in the Apartment

The domestic cat's brain was shaped by millions of years of life as a wild predator. Olfaction optimized for tracking prey, hearing tuned for locating the high-frequency calls of rodents, motor ability designed for ambush and short-distance bursts.

The daily sensory input of a modern indoor house cat is the airflow from climate control, the low-frequency hum of a refrigerator compressor, the artificial fragrance molecules in commercial cat food from a plastic bowl, a limited number of rooms, an occasional feather wand. Domestication history spans roughly ten thousand years. The evolutionary pressures that shaped the cat brain's sensory processing networks lasted millions of years. Ten thousand years is not enough for the nervous system to complete deep-level adaptation.

The raw material of the domestic cat's dreams comes from the apartment. The neural equipment processing that raw material comes from the wilderness. The high incidence of compulsive behaviors in indoor cat populations (excessive grooming, repetitive pacing, pouncing at nonexistent targets) has long been attributed to insufficient environmental stimulation. If memory consolidation during REM is also affected by impoverished raw material, the causes of these behavioral abnormalities extend beyond waking hours.

Replay Speed

In 2001, Matthew Wilson and Kenway Louie at MIT recorded that during REM sleep in rats, hippocampal place cell firing sequences closely matched those from daytime maze running, precise to the single-neuron level. Replay speed was approximately six to seven times the speed of waking experience. Cats and rats share the same basic hippocampal structure and REM sleep mechanisms. At this rate, a single 20-minute REM cycle in a cat may process an amount of information equivalent to two hours of waking experience on a neural level.

How to Tell Dreaming from Ordinary Twitching

NREM stage: EEG dominated by high-amplitude slow waves (delta waves, 0.5 to 4Hz), muscle tone present. Limb twitches during this stage are spinal reflexes, unrelated to dreams.

REM stage: EEG shows low-amplitude fast waves, very similar to the waking state, which is why it is called "paradoxical sleep." Hippocampal theta rhythm (4 to 8Hz) appears, PGO waves propagate from the pons to the lateral geniculate nucleus and occipital cortex. Except for the extraocular muscles and diaphragm, skeletal muscle tone throughout the body is almost completely suppressed.

Practical identification: if the cat's eyeballs are moving rapidly under closed eyelids (sometimes you can see the eyelids flickering), while most of the body's muscles are relaxed with only the paws, whiskers, and corners of the mouth moving, it is most likely in REM. If the body maintains a degree of muscle tension, a leg suddenly kicks once and then stops, that looks more like an NREM spinal reflex.

A cat's full sleep cycle is about 25 to 30 minutes, with the REM phase lasting six to eight minutes. A cat sleeping 14 to 16 hours a day spends roughly three to four hours in REM. Roughly calculated, a domestic cat spends close to twenty thousand hours dreaming over its lifetime.

The Exemption of Purring

Some cats purr during REM sleep. Purring requires the laryngeal muscles to contract rhythmically at 25 to 50 times per second, and REM atonia should theoretically suppress these muscles. The purring circuit has an exemption from REM atonia, same as breathing. During waking hours, purring is usually associated with a sense of safety and self-soothing. The fact that this circuit can still be activated during dreams becomes important when we get to nightmares.

Nightmares

The cat's amygdala remains significantly active during REM sleep. Memories tagged with fear participate in REM replay as usual. Prefrontal cortex activity drops markedly during REM. In humans, the prefrontal cortex handles logical judgment and reality testing; its going offline is the neurological reason humans lose critical thinking in dreams. The cat's prefrontal cortex is already less developed than a human's; with further reduced activity during REM, the capacity to buffer fear becomes negligible.

Humans have a process called "fear extinction during sleep": fear memories gradually fade in intensity across multiple REM replays. This depends on the prefrontal cortex's top-down regulation of the amygdala. The cat's prefrontal cortex probably cannot sustain this regulation, so fear memories replay at roughly the same intensity each time.

Cats that have experienced severe trauma maintain extremely strong fear responses to specific stimuli even after living in a safe environment for years. The usual explanation focuses on conditioned reflex consolidation during waking hours. Adding high-fidelity fear replay during REM to the picture, these cats are not only trapped in fear memories during the day but go through another round at night, and they lack the physiological mechanism to make each round progressively weaker. Shelter interventions for traumatized cats are almost entirely focused on desensitization training during waking hours. Sleep-period interventions (optimizing sleep environments, reducing nighttime stressors) have not been systematically incorporated.

Back to purring. With the prefrontal cortex essentially absent, the purring circuit may be the only physiological means a cat has for countering fear during a nightmare. A brainstem-level, automatically running soothing circuit that does not depend on higher cognitive functions. Entirely different from the human mechanism of relying on the prefrontal cortex to "come back online" to restore a sense of reality during nightmares. Older, more fundamental.

PGO Waves Persist Under General Anesthesia

General anesthesia shuts down consciousness, shuts down perception, shuts down movement. The PGO wave generator keeps working.

It is embedded at the brainstem level, in the same neuroanatomical region as the respiratory center and cardiac regulation center. After chemical agents strip away the brain's higher functions layer by layer, what remains running are the most ancient circuits that sustain life, and PGO waves.

Jouvet himself leaned toward the view that REM sleep plays an irreplaceable role in maintaining "the genetic programming of instinctive behavior." Roffwarg emphasized neural development. Mark Blumberg's recent work focuses on sensorimotor calibration. There is no consensus on why this mechanism was anchored by evolution at such a deep level. The part that is agreed upon is only this: it was indeed anchored there.

Twenty Thousand Hours

Dreaming means the brain generated coherent internal experience while completely disconnected from the outside world. Rocks do not dream. Thermostats do not dream. The existence of dreaming points to an inner space that can run on its own when external input is absent. What can be confirmed is that the cat's brain generates temporally sequenced behavioral commands in this space, draws on memory material from multiple sensory channels, accompanied by amygdala-mediated emotional activity. What cannot be confirmed is whether the cat "experiences" any of this, something that has not been resolved even in humans.

Twenty thousand hours. Evolution spent millions of years protecting the hardware that sustains all of this to the same level as the heartbeat.