If terminal lucidity is real, it means that there are naturally occurring conditions under which even severely damaged brains can have their function restored.
And if a dying brain can indeed spontaneously access ‘lost’ memories, perhaps we could induce a living brain to do the same.
However, there are good reasons to think terminal lucidity might not be real.
Consider the case of a 12-year-old girl with cancer, reported in a 2024 study of terminal lucidity among children with severe illnesses.
What we still don’t have is any evidence of physiological changes during terminal lucidity.
An 81-year-old woman with advanced Alzheimer’s is living in a retirement home in Iceland. She hasn’t recognised her family or spoken a word to them in a year. One day, while her son Lydur is sitting beside her working on a crossword puzzle, she sits up and looks him directly in the face.
‘My Lydur,’ she says. ‘I am going to recite a verse to you.’
And then she says the following in a loud and clear voice:
Oh, father of light, be adored.
Life and health you gave to me,
My father and my mother.
Now I sit up, for the sun is shining.
You send your light in to me.
Oh, God, how good you are.
She then lies back down and never talks again. In a month, she dies.
This case, reported in a 2011 study by the biologist Michael Nahm and colleagues from the University of Virginia and the University of Iceland, is an example of terminal lucidity: an unexpected return of cognitive function in patients with severe dementia (or other brain damage) that occurs shortly before death.
As a neuroscientist, my first thoughts when encountering reports like this are that they can’t possibly be real. The main reason is that, by the time patients with severe dementia die, their brains are catastrophically damaged. For a protracted period before death, these patients typically show no signs of recognising family members. They often haven’t responded meaningfully to their environment in months or years. And they’ve lost 20 to 50 per cent of their synaptic connections – so much that their brains have visibly shrunk on MRI scans.
On top of this existing damage, the dying process itself is hardly optimal for brain function. Everything we know about neuroscience says this should make cognition worse, not better.
And yet, terminal lucidity keeps being reported.
If a dying brain can spontaneously access ‘lost’ memories, perhaps we could induce a living brain to do the same
In some reports, 60 to 70 per cent of dementia caregivers say they’ve witnessed at least one patient return to lucidity in the days before death. In a study that specifically followed deaths in hospice, researchers found examples of terminal lucidity in 6 per cent of cases. That makes the phenomenon neither ubiquitous nor rare – in the US alone, it would mean around 10,000 cases per year.
This potentially has profound implications for neuroscience and medicine.
Retaining memories depends on neural circuitry maintaining specific synaptic connection strengths. As dementia progresses, memory worsens because these neural structures physically deteriorate – synapses disintegrate, neurons die. So when a woman who hasn’t recognised her son in a year suddenly calls him by name and recites a poem she learned during her childhood, where have those memories been physically stored? If the neural substrate we thought was necessary for memory is degraded, but the memories can still be accessed, then either our understanding of memory storage is fundamentally wrong, or there’s some resilient substrate we haven’t identified yet.
Right now, around 60 million people have dementia worldwide, and their care costs more than $1 trillion each year. The newest treatments attempt to remove accumulations of abnormally folded proteins in the brain, known as amyloids, which are believed to contribute to Alzheimer’s. One such treatment is lecanemab (brand name Leqembi), which costs around $30,000 per year, in exchange for, at best, a minuscule reduction in the rate of disease progression (let alone actually reversing the decline).
If terminal lucidity is real, it means that there are naturally occurring conditions under which even severely damaged brains can have their function restored. And if a dying brain can indeed spontaneously access ‘lost’ memories, perhaps we could induce a living brain to do the same.
However, there are good reasons to think terminal lucidity might not be real. Primary among these is observer bias.
Watching a parent slowly lose themselves without ever having a final chance to say goodbye is devastating. People crave the deathbed scenes from books and movies: the elderly relative who pronounces meaningful final words to gathered family before peacefully expiring. But there’s often a long gap between when someone loses the ability to communicate and when they die – days, weeks, months or even years in cases of severe dementia. During this time, family members may rehearse what they wish they could say and dream of what they wish they could hear back, desperately hoping for one last sign, however small, that their loved one recognises them or is trying to communicate.
This creates exactly the conditions for observer bias to flourish.
Consider the case of a 12-year-old girl with cancer, reported in a 2024 study of terminal lucidity among children with severe illnesses. The study, from The Journal of Nervous and Mental Disease, was made by the paediatric haematologist-oncologist Philip Roehrs in the US and a team of researchers from around the world. They reported that the girl had been unresponsive for months and was eventually declared dead. However, according to her mother, just before the girl died, she had opened her eyes, mouthed ‘I love you’ and ‘I’m ready to go home,’ and blinked her eyes to answer yes/no questions. The researchers added that the mother ‘appeared to consider these moments as signs of life, despite her daughter’s medical condition.’
Reports of terminal lucidity come from emotionally invested family members or caregivers
No one wants to ask a grieving mother about how confident she truly is that her unresponsive daughter briefly regained awareness before dying. But from a scientific perspective: were the words she mouthed meaningful speech or just involuntary muscle movements? The same kind of question applies to a grandmother who seems to smile and nod (postural reflex?) or a dying spouse who squeezes a hand when told ‘I love you’ (muscle spasm?).
This is the problem with nearly all reports of terminal lucidity: they come from emotionally invested family members or caregivers with strong motivations to perceive communication where there might be none. Those witnessing changes in their loved ones often have no scientific training to distinguish meaningful behaviour from reflexes, and no blinding to prevent bias.
In 2024, Joan M Griffin and other researchers surveyed family caregivers who reported witnessing lucid episodes in dementia patients. They found 51 per cent of these episodes involved communication that ‘made complete sense’, with another 32 per cent making ‘some sense’. Though, keep in mind that 48 per cent of episodes occurred at least six months before death, with only 18 per cent happening in the final week. That means that most of these weren’t ‘terminal’ lucid episodes at all – they’re potentially just part of the normal cognitive fluctuations that occur as dementia progresses.
You may like How to talk to someone who has dementia Nathaniel Chin & Caoilfhinn Rauwerdink Save It’s not always clear when someone’s lost the ability to decide Jessica M Choplin Save What moments of insight have in common Sam Dresser Save
The scientific evidence we currently have for terminal lucidity leaves us almost entirely with anecdotal reports from biased observers, collected retrospectively. These have not been objectively verified and mostly describe ambiguous behaviours that could arguably be explained by wishful interpretation.
So, how can we explain this phenomenon that has been consistently reported across centuries and cultures? Anecdotal reports of patients who haven’t spoken in years suddenly engaging in complex conversation are hard to dismiss as mere muscle spasms or observer bias. Something appears to be going on beyond pure confabulation.
No one has yet published any neuroscientific recording of terminal lucidity as it happens, so we still have no record of what the brain is doing during an episode. Although any explanation is necessarily speculative, we can make educated guesses based on known and related phenomena.
One explanation relates to brain-swelling reduction. Dying patients often develop brain swelling for multiple reasons, such as electrolyte imbalances, low blood protein, kidney or liver failure, inflammation, or damage to the blood-brain barrier. This swelling compresses brain tissue and reduces blood flow, starving neurons of oxygen and glucose. The result is the familiar progression of terminal delirium: confusion, then sleepiness, then coma, then death.
But in their final days, many patients stop eating and drinking entirely. The resulting dehydration could reduce brain swelling, allowing blood flow to increase, which could temporarily restore some cognitive function – a brief window of lucidity before the dying process continues. It’s a simple mechanism: lower intracranial pressure means better movement of blood through the brain, which means more oxygen and glucose in the brain, which in turn means more functional neurons.
Maybe damaged neural networks can spontaneously reorganise in ways that temporarily restore function
Another explanation, perhaps working in tandem with brain-swelling reduction, relates to how brain networks respond to extreme conditions.
Normal consciousness requires a precise balance of neurochemicals. What is referred to as ‘wakefulness’ – the daily state when you are conscious and your brain is actively responding to the world – depends on brainstem neurons continuously releasing neuromodulators like acetylcholine, noradrenaline and histamine into the cortex. When this system breaks down in a damaged or dying brain, people typically become progressively less responsive.
But sometimes, perturbing a broken system can temporarily fix it. If you’ve ever fixed a glitchy electronic device by hitting it, you’ve seen how pushing a malfunctioning system into an unusual state can accidentally restore function.
Maybe something similar happens in dying brains, where under extreme or unusual conditions, damaged neural networks can spontaneously reorganise in ways that temporarily restore function. One research paper speculates that dying brains might generate ‘spontaneous network integration manifesting as lucid behaviour’ through ‘rapid and nonlinear synchronisation’. Translated from Academic into English, this means something like: ‘Weird things might happen in a dying brain that briefly improve function, but we don’t really know what those things are.’
Even after reading many of the studies about terminal lucidity, I genuinely don’t know whether it really exists. Given this, the appropriate response is neither to credulously accept every report nor to entirely dismiss the phenomenon either. Instead, we need to begin studying terminal lucidity properly. For a long time, almost no one was, though that’s beginning to change.
In 2019, the US National Institute on Aging funded a wave of research, and results are arriving. The most important so far, by a team from the University of Wisconsin-Madison, has captured potential lucid episodes on video. Across 1,433 hours of footage from 20 hospice patients, the team identified nine lucid episodes in three participants – events where communication and social engagement spiked far above each patient’s measured baseline. One man, after a two-week decline, spoke in full sentences for two days, named his family members, said ‘I’m leaving soon,’ and died days later.
This isn’t definitive proof that terminal lucidity is real – the study still suffered from potential observer bias. But it’s the best evidence so far that something is happening that can’t be waved away as grief-soaked confabulation.
What we still don’t have is any evidence of physiological changes during terminal lucidity. There have been no brain scans during an episode, nor blood work to accompany them. A group of researchers at New York University Langone Health, led by Sam Parnia, has been planning continuous EEG monitoring of 500 dying dementia patients, which might someday tell us what brain activity looks like during lucid episodes.
If the video evidence holds up, and if some future study finds a physiological signature to match that evidence, the picture of dementia we currently have would need rethinking. Memory and identity would need to be understood as more distributed, more redundant, and more recoverable than the standard neurodegenerative story allows. And instead of pouring billions into another hopeless amyloid-clearing drug, we might finally have a treatment path worth pursuing – if only we can discover how to stretch a lucid episode into a lasting return.
