Biological fine-tuning

 

 The Set of Amino Acids Used  in Life Is No “Frozen  Accident”

  https://scienceandculture.com/2026/04/the-set-of-amino-acids-used-in-life-is-no-frozen-accident/Jonathan McLatchie 

 

April 28, 2026

Evolution, Intelligent Design

Share

  

 Francis Crick described the conventional genetic code as a “frozen  accident,” given that once the mapping between codons and amino acids is  established it becomes extremely difficult to make significant changes without wreaking havoc on every polypeptide made by the cell.1 In recent  decades, scientists have come increasingly to the realization that the  genetic code is not random but highly optimized, on multiple levels.

Not only is the genetic code itself fine-tuned for error minimization, but it turns out that the set of amino acids used in life is also highly optimal — that is to say, the selection is not random. In 2011, a paper was published, in which the authors compared the coverage of the standard alphabet of 20 amino acids for “size, charge, and hydrophobicity with equivalent values calculated for a sample of 1 million alternative sets (each also comprising 20 members) drawn randomly from the pool of 50 plausible prebiotic candidates.”2 They found that,

…the standard alphabet exhibits better coverage (i.e., greater breadth and greater evenness) than any random set for each of size, charge, and hydrophobicity, and for all combinations thereof. In other words, within the boundaries of our assumptions, the full set of 20 genetically encoded amino acids matches our hypothesized adaptive criterion relative to anything that chance could have assembled from what was available prebiotically.

I wrote about this paper when it came out here.

A more recent paper, published in 2017 in The FEBS Journal, argued that the set of amino acids commonly used throughout biology is fundamentally non-random.3 The authors argue that, when compared to other sets of amino acids in relation to “component atoms, functional groups, biosynthetic cost, use in a protein core or on the surface, solubility and stability,” there are very good reasons why biology uses the conventional set of amino acids and not others. He observes that “Applying these criteria to the 20 standard amino acids, and considering some other simple alternatives that are not used, we find that there are excellent reasons for the selection of every amino acid. Rather than being a frozen accident, the set of amino acids selected appears to be near ideal.”

Functional Groups

Doig notes that “the choice of functional groups is rather limited in small molecules when using only C, H, O, N or S.” Carbon-nitrogen bonds, carboxyls, hydroxyls, amides, and amines are stable chemical groups that can form electrostatic interactions and hydrogen bonds. Alternative chemical groups (such as esters, anhydrides, and nitriles) are too prone to hydrolysis in an aqueous environment. Moreover, aldehydes and ketones are too chemically reactive.

Biosynthetic Cost

Another property that determines which amino acids are used by the cell is the energetic cost of their biosynthesis in terms of glucose and ATP molecules: “For example, Leu costs only 1 ATP, but its isomer Ile costs 11. Why would life ever therefore use Ile instead of Leu, if they have the same properties?” Doig further notes that “Larger is not necessarily more expensive; Asn and Asp cost more in ATP than their larger alternatives Gln and Glu, and large Tyr costs only two ATP, compared to 15 for small Cys. The high cost of sulfur-containing amino acids is notable.”

Burial and Surface

A close-packed core of a protein (where there are few empty spaces) maximizes weak attractions between atoms (van der Waals interactions), which make the protein more stable. Thus, “A solid core is essential to stabilize proteins and to form a rigid structure with well-defined binding sites.” This means that having nonpolar side chains is important to stabilize close-packed hydrophobic cores. On the other hand, polar and charged amino acid side chains, which are exposed on a protein surface, promote solubility in the aqueous environment.

Solubility

Doig further observes that “the least soluble amino acid at pH 7 in water is Tyr, so any less soluble than this may not be acceptable.”

Stability

Doig also notes that “even with stable functional groups, some amino acids are prone to unwanted reactions, such as cyclisation or acyl transfer, that can lead to decomposition or racemisation.”

The Implications

Doig proceeds to consider each of the twenty commonly used amino acids, evaluating for each its suitability for life relative to other amino acid sets. He concludes that “There are excellent reasons for the choice of every one of the 20 amino acids and the nonuse of other apparently simple alternatives. If all else fails, one can resort to chance or a ‘frozen accident’, as an explanation.” Curiously, he fails to consider an alternative explanation, which seems to fit the evidence better, and for which we already possess independent evidence — i.e., purposeful selection by an intelligent mind.

Significantly, these data indicate that the space of usable amino acids is severely constrained. Evolutionary mechanisms would, therefore, need to explore a vast chemical space and converge on a highly optimized set. Once the canonical genetic code is established, it would be extremely difficult to change it over time, since each amino acid would be tied to specific codons, tRNAs, and aminoacyl-tRNA synthetases. Modifying the set of amino acids would thus require significant rewiring. Reassignments of the codons and amino acids would affect every polypeptide made by the cell and would wreak havoc on the translation of many different proteins. On the other hand, reducing the alphabet of amino acids significantly constrains the proteins that can be made. This, in turn, would constrain the chemistry and needed structural precision for primitive systems of DNA replication.

Intelligent Design

Intelligent agents are uniquely capable of purposefully selecting between options from a large search space. The fact that the set of twenty amino acids conventionally used in life is non-random but in fact highly optimized is not surprising on the hypothesis that their selection was by an intelligent mind. On the other hand, they are wildly surprising on the hypothesis that it arose by unguided processes. In view of this overwhelmingly top-heavy likelihood ratio, these findings point to teleology as being the best explanation.

Notes

1. Crick FH. The origin of the genetic code. J Mol Biol. 1968 Dec;38(3):367-79. doi: 10.1016/0022-2836(68)90392-6. PMID: 4887876.↩︎
2. Philip GK, Freeland SJ. Did evolution select a nonrandom “alphabet” of amino acids? Astrobiology. 2011 Apr;11(3):235-40. doi: 10.1089/ast.2010.0567. Epub 2011 Mar 24. PMID: 21434765.↩︎
3. Doig AJ. Frozen, but no accident — why the 20 standard amino acids were selected. FEBS J. 2017 May;284(9):1296-1305. doi: 10.1111/febs.13982. Epub 2017 Jan 13. PMID: 27926995.↩︎

Jonathan McLatchie

Resident Biologist and Fellow, Center for Science and Culture

Dr. Jonathan McLatchie holds a Bachelor’s degree in Forensic Biology from the University of Strathclyde, a Masters (M.Res) degree in Evolutionary Biology from the University of Glasgow, a second Master’s degree in Medical and Molecular Bioscience from Newcastle University, and a PhD in Evolutionary Biology from Newcastle University. Previously, Jonathan was an assistant professor of biology at Sattler College in Boston, Massachusetts. Jonathan has been interviewed on podcasts and radio shows including “Unbelievable?” on Premier Christian Radio, and many others. Jonathan has spoken internationally in Europe, North America, South Africa and Asia promoting the evidence of design in nature.

PET 2 LEVIN

 

Planaria Memory & Xenobots: Detailed Breakdown

---

PART 1: PLANARIA AND MORPHOLOGICAL MEMORY

Background: Why Planaria?

Planaria (Dugesia japonica and related species) are freshwater flatworms with an almost absurd capacity for regeneration. Cut one into 200 pieces, and you get 200 new worms. Every fragment — even a tiny tail segment with no brain tissue whatsoever — will regenerate a complete, properly oriented animal.

This alone is remarkable. But Levin's lab discovered something far stranger.

---

The Classic Setup: Polarity and Bioelectrics

Every planarian has a defined anterior-posterior axis — a head end and a tail end. This polarity is maintained by a gradient of bioelectric signals across the body, particularly involving:

• Gap junctions — protein channels that directly connect neighboring cells, allowing ions and small molecules to flow between them
• Ion pumps (especially V-ATPase, an H⁺ pump) that create voltage differentials across tissues
• Neurotransmitters like serotonin that act as patterning signals even outside the nervous system

The key insight: this bioelectric pattern is not just a readout of genetic instructions. It is itself instructive — it tells cells where they are and what they should become. It functions like a map or a blueprint held in electrical form across the tissue.

---

The Experiment: Erasing and Overwriting the Memory

Here is what Levin's lab did, step by step:

Step 1 — Cut the worm A planarian is cut transversely, producing a head fragment and a tail fragment. Normally:

• The head fragment regenerates a new tail at the wound site
• The tail fragment regenerates a new head at the wound site

This is normal polarity — the body "knows" which end is which.

Step 2 — Pharmacological intervention during regeneration While the tail fragment is regenerating (the critical window is the first ~24 hours after cutting), researchers bathe it in octanol — a drug that blocks gap junctions, severing the bioelectric communication between cells.

What happens? The tail fragment, unable to read its normal bioelectric positional information, sometimes regenerates a head at both ends — a two-headed worm. The chemical disruption confused the tissue's sense of axis.

Step 3 — The crucial discovery: the memory persists Here is where it gets extraordinary. After the two-headed worm is fully regenerated, researchers:

• Remove the drug (restore normal chemical environment)
• Allow the worm to live normally for weeks
• Then cut it again — no drugs, no intervention

A normal worm, cut in this state, would regenerate with correct polarity. But these worms regenerated two-headed worms again — even though no drug was present during the second regeneration.

The altered bioelectric state had become self-sustaining. The tissue had "learned" a new body plan and perpetuated it across generations of regeneration without any genetic change and without ongoing chemical intervention.

---

What Is Actually Being "Remembered"?

This is not memory in the neural sense — there are no neurons storing an engram. What's happening is:

Bioelectric state as memory: The pattern of ion flows, membrane voltages, and gap junction connectivity across the tissue constitutes a stable attractor state — a self-reinforcing electrical pattern that cells maintain and propagate when they divide and communicate.

Think of it like a standing wave in a medium. The wave pattern persists not because any single molecule holds it, but because the dynamics of the system continuously recreate it. When you cut the worm, the new cells inherit the electrical environment of their neighbors and reproduce the same pattern.

The "Body Plan Morphogenetic Field": Levin frames this in terms of a morphogenetic field — a distributed representation of target anatomy encoded not in DNA sequence but in the bioelectric state of the tissue. DNA provides the parts list (what proteins are available); the bioelectric field provides the blueprint (how those parts should be assembled).

The planaria experiments show this blueprint can be:

• Altered (by disrupting gap junctions)
• Stabilized in its new form (self-reinforcing dynamics)
• Transmitted across regenerative events (new cells adopt the electrical state of existing cells)
• Maintained without ongoing external signals

---

The Deeper Implication

This means the worm's body plan is not simply computed fresh from the genome each time. There is a second layer of information — above the genome, encoded in bioelectric dynamics — that stores and propagates the target morphology.

Levin uses the analogy of computer hardware vs. software: the genome is the hardware (fixed, inherited), but the bioelectric pattern is the software running on top — and that software can be edited, corrupted, and rewritten independently of the hardware.

This is teleological in a deep sense: the tissue is not just following chemical gradients passively. It is actively comparing its current state to a stored target state and making corrections. When that target state is altered, the corrections drive toward the new target.

---

---

PART 2: XENOBOTS — LIVING MACHINES AND KINEMATIC SELF-REPLICATION

What Are Xenobots?

Xenobots were first reported in 2020, created collaboratively by Levin's lab at Tufts and Josh Bongard's computational lab at the University of Vermont.

They are made from Xenopus laevis — the African clawed frog — specifically from ectodermal cells (cells that would normally become skin) harvested from early embryos.

Crucially: these cells are removed from their normal developmental context. They are no longer receiving the chemical signals that would tell them "you are part of a frog embryo, become skin." Freed from that context, they exhibit novel, emergent behaviors that no frog has ever exhibited — and that were never programmed.

---

Stage 1: Self-Assembly and Spontaneous Organization

When ectodermal cells are harvested and placed in a dish as a loose aggregate, something remarkable happens over ~24 hours:

• The cells do not die (as isolated cells typically would)
• They reorganize — moving, adhering, and arranging themselves into a compact, roughly spherical or ovoid body
• Cells with cilia (hair-like projections, normally used to move mucus across skin) reorient those cilia outward, creating a coordinated propulsion system
• The resulting structure, roughly 500–700 micrometers across, moves through fluid in a directed way

No blueprint was given. No morphogen gradient was applied. The cells used their existing genetic toolkit — developed for one purpose (being frog skin) — to build a novel functional organism suited to their new context.

This is a profound demonstration of context-dependent agency: the cells are not executing a fixed program, they are solving the problem of "what should I be?" given their current situation.

---

Stage 2: The AI-Designed Body Plans

In the 2020 paper, Bongard's team used an evolutionary algorithm running on a supercomputer to design optimal xenobot body shapes for specific tasks (like moving in a particular direction or pushing objects).

The process:

1. Simulate thousands of random arrangements of frog cells in silico
2. Evaluate which arrangements best accomplish a target behavior
3. Select the best performers, mutate them, repeat
4. Take the winning design and build it in real life using microsurgery on real frog cells

Strikingly, the real xenobots closely matched the behavior of their simulated counterparts — confirming that the cells were indeed executing coherent, predictable behaviors, not random activity.

But the spontaneously self-assembled xenobots (without AI design) were arguably more interesting, because they required no external design at all.

---

Stage 3: Kinematic Self-Replication (2021)

The 2021 PNAS paper reported something that genuinely shocked the scientific community.

The setup: Researchers placed xenobots in a dish containing a large supply of loose, dissociated frog cells — essentially raw cellular material.

What happened: The xenobots spontaneously began to gather the loose cells, sweeping them together into piles using their ciliary motion.

Those piles of gathered cells then self-assembled into new xenobots.

Those new xenobots could, in turn, gather more loose cells and produce another generation.

This is kinematic self-replication — replication through motion and behavior rather than through chemical template copying (like DNA replication). It is an entirely different category of reproduction from anything previously documented.

---

Why "Kinematic" Is the Key Word

In biology, replication almost always means template-directed chemical copying: DNA unzips, complementary bases are added, you get two copies. The information is in the molecule.

Kinematic self-replication is different: the information is in the behavior and shape of the organism. The xenobot "replicates" not by copying its molecules but by doing things in the world — moving, gathering, organizing — that result in a new entity similar to itself.

This is closer to how a whirlpool "reproduces" (by creating conditions for new whirlpools) than how a bacterium reproduces. But xenobots are vastly more complex and directed.

Von Neumann had theorized kinematic self-replicators in the 1940s as a thought experiment. Xenobots are the first biological instantiation.

---

The Shape Discovery: C-Shape Matters

A critical finding from the 2021 work: the default spherical xenobots replicate poorly — they gather some cells but the process is inefficient and quickly dies out after 1–2 generations.

When the AI evolutionary algorithm was again applied — this time to optimize for replication efficiency — it discovered that a C-shaped or pac-man-shaped xenobot was dramatically more effective at gathering and corralling loose cells into the interior of the C, where they could aggregate and self-organize.

This C-shape does not exist in frog biology. It was discovered by an algorithm, then built, then confirmed to work. The frog cells — never having "intended" to replicate this way — executed the replication faithfully when given the right body shape.

---

The Philosophical Payload

The xenobot work raises several deeply unsettling and important questions:

1. What is an organism? Xenobots are made of frog cells, carry frog DNA, but are not frogs. They have never existed in nature. They exhibit goal-directed behavior (moving, replicating) that no frog exhibits. Are they a new species? A new category of being?

2. What is reproduction? Xenobots don't pass on genetic information to their "offspring" — the daughter xenobots have the same genome as the parent, inherited from the original frog embryo. What's being transmitted is form and behavior, not genetic novelty. This blurs the line between reproduction and construction.

3. Where does the "goal" come from? The xenobots were not programmed to gather cells and replicate. This behavior emerged from the interaction of cell-level properties (cilia, adhesion, motility) with the physical environment. The goal-directedness is real — measurable, reproducible — but it wasn't designed in. It emerged.

Levin argues this shows that agency and teleology can be genuine properties of matter — not imposed from outside by a designer or genome, but arising from the dynamics of sufficiently complex, self-organizing systems.

---

Connecting Planaria and Xenobots: The Unified Theme

Both experiments point to the same conclusion:

	Planaria	Xenobots
Substrate	Adult flatworm tissue	Embryonic frog cells
Context change	Chemical disruption of polarity signals	Removal from embryonic context
Response	Adopt and maintain new body plan	Self-assemble novel body plan
Memory/Goal	Bioelectric target state	Emergent morphogenetic goal
Key finding	Morphological memory above genome	Goal-directed behavior without programming

In both cases, cells are doing something more than executing a genetic program. They are reading their environment, computing a response, and pursuing a target state — even when that target state has been altered or is entirely novel.

This is Levin's core claim made flesh: biology is cognitive all the way down, and the gap between "mere matter" and "minded agent" is not a gap at all, but a continuum we are only beginning to measure.

Michael Levin - teleology in biology

 Michael Levin is a developmental biologist at Tufts University whose work is reshaping how we think about intelligence, agency, and goal-directedness in living systems. Here's an overview of his revolution:

The Core Idea: Intelligence Is Not Brain-Specific

Levin argues that cognition and agency are substrate-independent — they don't require neurons or brains. Instead, any system capable of pursuing goals, remembering states, and problem-solving in a flexible way deserves to be called "intelligent" in some meaningful sense.

Teleology Rehabilitated

Classical science largely banished teleology (goal-directedness) from biology after Descartes and the mechanistic revolution. Levin is helping bring it back — not as mysticism, but as a legitimate, measurable property of biological systems.

His key insight: organisms at every scale — from single cells to tissues to whole animals — exhibit goal-directed behavior toward specific target states (like a body plan or organ shape). This isn't metaphor; it's observable, manipulable, and quantifiable.

The Planaria & Xenobot Experiments

Levin's lab produced some of the most striking evidence:

• Planaria (flatworms) remember their head-tail polarity even when cut and regrown in reversed chemical environments — their tissues "remember" a target morphology.
• Xenobots — living robots made from frog stem cells — spontaneously self-assembled, moved, and even exhibited a crude form of kinematic self-replication, a behavior no one programmed.
• Anthrobots (2023) — made from human lung cells — self-assembled and promoted neural repair in lab dishes, again without explicit programming.

These organisms pursue goals that weren't hardcoded — they compute solutions to problems using biological hardware.

The Cognitive Light Cone

One of Levin's most important theoretical contributions is the "cognitive light cone" — the idea that different agents have different scales over which they integrate information and pursue goals:

• A bacterium has a tiny cognitive light cone (milliseconds, micrometers)
• A brain has a larger one (years, the whole body)
• A society or ecosystem might have an even larger one

This gives a continuous, non-binary framework for intelligence — dissolving the sharp line between "dumb matter" and "minded beings."

Bioelectricity as the Medium

Levin's lab focuses heavily on bioelectric signaling — electrical patterns across cell membranes (not just in neurons) that encode positional and morphogenetic information. These signals function like a body-wide "blueprint memory":

• Disrupting bioelectric patterns causes radical morphological changes (e.g., making flatworms grow two heads)
• Restoring correct bioelectric patterns can cure cancer in tadpoles or induce limb regeneration

Bioelectricity appears to be a proto-cognitive medium — a way that non-neural tissue computes and communicates goal-states.

Implications Across Fields

Philosophy of mind: Mind and agency exist on a continuum; we need new frameworks beyond neurocentric consciousness theories.

AI & robotics: True intelligence may require goal-directedness and embodiment, not just computation. Levin collaborates with Daniel Dennett and others on these questions.

Medicine: Cancer may be a failure of cellular agency — cells "forgetting" their role in the larger collective and reverting to unicellular selfishness. Treating it bioelectrically rather than just chemically is an active research direction.

Ethics: If intelligence is a spectrum, where does moral consideration begin? Levin takes this seriously — his lab has published on the ethics of creating novel cognitive entities (like xenobots).

The Deeper Philosophical Bet

Levin is essentially arguing that teleology is real, physical, and measurable — that "what is this system trying to do?" is a legitimate scientific question at every scale of biology. This cuts against both naive mechanism and mystical vitalism, carving out a rigorous third path: computational teleology, where goal-states are physically encoded and causally efficacious.

It's one of the most ambitious research programs in contemporary science — simultaneously experimental, theoretical, and deeply philosophical.

OW LITTLE WE UNDERSTAND. READ THE NEXT TWO POSTS TO ENTER A NEW WORLD OF BIOLOGY

Worth investigation

 Terminal lucidity

From Justapedia, where neutrality meets objective truth

Jump to navigationJump to search

Terminal lucidity, also known as paradoxical lucidity, rallying or the rally, is an unexpected return of mental clarity and memory, or suddenly regained consciousness that occurs in the time shortly before death in patients with severe psychiatric or neurological disorders.^[1][2][3] This condition has been reported by physicians since the 19th century.

History

Several case reports in the 19th century described the unusual condition of an improvement and recovery of the mental state in patients days or weeks before death. William Munk, for instance, in 1887 called the phenomenon "lucidity before death".^[4] According to historical reviews headed by the biologist Michael Nahm, who also has an interest in mediumship and near-death experiences,^[5] the phenomena have been noted in patients with diseases which cause progressive cognitive impairment, such as Alzheimer's disease, but also schizophrenia, tumors, strokes, meningitis, and Parkinson's disease.^[6][7][8] However, terminal lucidity is not currently listed as a medical term.^[9]

According to Nahm, it may be present even in cases of patients with previous mental disability.^[10] Nahm defines two subtypes: one that comes gradually (a week before death), and another that comes rapidly (hours before death), with the former occurring more often than the latter. There may be plenty of cases reported in literature, although the phrase terminal lucidity was coined in 2009.^[11] Interest in this condition, which dwindled during the 20th century, has been reignited by further studies.^[4] A 2020 research screened for what the authors preferred to call "paradoxical lucidity", a general term for unexpected remissions in dementias, independent of whether it followed a terminality process or not; it found strong association of the condition as a near-death phenomenon and stated that it can overlap the concept of "terminal lucidity" in some cases.^[6] Such a paradoxical condition is considered a challenge to the irreversibility paradigm of chronic degenerative dementias such as Alzheimer's.^[12]

Causes

The earliest attempt at explanation was issued by Benjamin Rush in 1812, which proposed the hypothesis that a reawakening could be due to a nervous excitation caused by pain or fever, or else because of dead blood vessels, released by a leakage of water in the brain chambers. Johannes Friedreich, in 1839, proposed that the factors causing impairments may be reversed shortly before death, analogous to the reabsorption in terminal patients with hydrocephalus, and that high fever may be a cause of it. According to Macleod (2009)^[13] in his observations, explanative causes could not be found for the variety of cases, but it was suggested that due to the modern pharmacology in terminal cases, the condition may be less common today.^[4] A recent proposed mechanism includes a non-tested hypothesis of neuromodulation, according to which near-death discharges of neurotransmitters and corticotropin-releasing peptides act upon preserved circuits of the medial prefrontal cortex and hippocampus, promoting memory retrieval and mental clarity.^[14]

References

1.   Koczanowicz, Leszek (2020). "Chapter 12 - The Anxiety of Clairvoyance: Terminal Lucidity and the End of Culture". Anxiety and Lucidity: Reflections on Culture in Times of Unrest. Routledge. pp. 162–198. ISBN 978-0367218232.

2.   Mendoza, Marilyn A. "Why Some People Rally for One Last Goodbye Before Death". Psychology Today (blog). Retrieved 26 August 2019.

3.   Bursack, Carol Bradley. "When Loved Ones Rally Before Death". AgingCare. Retrieved 26 August 2019.

4.   Chiriboga-Oleszczak, Boris Alejandro (2017-03-28). "Terminal lucidity". Current Problems of Psychiatry. 18 (1): 34–46. doi:10.1515/cpp-2017-0003. ISSN 2353-8627.

5.   Michael Nahm

6.   Batthyány, Alexander; Greyson, Bruce (2020-08-27). "Spontaneous remission of dementia before death: Results from a study on paradoxical lucidity". Psychology of Consciousness: Theory, Research, and Practice. 8: 1–8. doi:10.1037/cns0000259. ISSN 2326-5531. S2CID 225192667.

7.   Nahm, Michael; Greyson, Bruce (December 2009). "Terminal lucidity in patients with chronic schizophrenia and dementia: a survey of the literature". Journal of Nervous and Mental Disease. 197 (12): 942–944. doi:10.1097/NMD.0b013e3181c22583. ISSN 1539-736X. PMID 20010032.

8.   Nahm, Michael; Greyson, Bruce; Kelly, Emily Williams; Haraldsson, Erlendur (July–August 2012). "Terminal lucidity: a review and a case collection". Archives of Gerontology and Geriatrics. 55 (1): 138–142. doi:10.1016/j.archger.2011.06.031. ISSN 1872-6976. PMID 21764150.

9.   Webster Medical Dictionary

10.                     Nahm, M.; Greyson, B. (2014). "The Death of Anna Katharina Ehmer: A Case Study in Terminal Lucidity". OMEGA. 68 (1): 77–87. doi:10.2190/OM.68.1.e. PMID 24547666. S2CID 1265185.

11.                     Bering, Jesse (2017). "One Last Goodbye: The Strange Case of Terminal Lucidity". Scientific American Blog Network.

12.                     Mashour, George A.; Frank, Lori; Batthyany, Alexander; Kolanowski, Ann Marie; Nahm, Michael; Schulman-Green, Dena; Greyson, Bruce; Pakhomov, Serguei; Karlawish, Jason; Shah, Raj C. (2019-06-19). "Paradoxical lucidity: A potential paradigm shift for the neurobiology and treatment of severe dementias". Alzheimer's & Dementia. 15 (8): 1107–1114. doi:10.1016/j.jalz.2019.04.002. ISSN 1552-5260. PMID 31229433.

13.                     Macleod, AD (December 2009). "Lightening up before death". Palliative & Supportive Care. 7 (4): 513-516. doi:10.1017/S1478951509990526. PMID 19939314.

14.                     Bostanciklioğlu, Mehmet (January 2021). "Unexpected awakenings in severe dementia from case reports to laboratory". Alzheimer's & Dementia. 17 (1): 125–136. doi:10.1002/alz.12162. ISSN 1552-5279. PMID 33064369. S2CID 222840626.