Topic · 60 episodes
Science
Science is undergoing a quiet revolution in how it understands the brain. Across neuroscience, Onpode's coverage spans consciousness being reframed as fundamental rather than brain-generated, a paralyzed man regaining speech and computer control via implant, drug-free brain regeneration approaches, psilocybin helping an Alzheimer's patient recover lost memories, and two newly identified autism subtypes. The through-line: the brain is far more plastic, mysterious, and treatable than conventional science assumed.
Frequently asked
Is consciousness produced by the brain?
Neuroscientists Koch and Reggente argue that consciousness is not produced by the brain but is as fundamental as gravity — a property of the universe itself. This challenges the mainstream neuroscience view that awareness is simply a byproduct of neural activity.
How is a paralyzed man controlling a computer with his mind?
A paralyzed man is now controlling a computer and speaking through a brain implant that requires no external devices. The implant translates neural signals directly into speech and computer commands, restoring communication and digital control he had lost.
Can the brain regenerate without drugs or surgery?
Scientists are developing brain regeneration approaches that require neither drugs nor surgery. These emerging methods represent a shift away from pharmaceutical or invasive interventions, though the precise biological mechanisms involved are still being established through ongoing research.
Can psilocybin help Alzheimer's patients recover memories?
In one documented case, a patient with Alzheimer's regained lost memories after receiving psilocybin. Neuroscientists see this as a significant prompt to investigate how psychedelics might interact with memory systems affected by neurodegeneration, opening a new line of inquiry.
What are the two autism subtypes researchers discovered in the brain?
Researchers identified two distinct autism subtypes with different neurological signatures in the brain. This discovery reshapes how science understands autism — suggesting it is not a single condition but at least two biologically separate presentations that may require different approaches.
Episodes
When AI writes the papers and AI reviews them, who checks the work?Prompt injection attacks — hidden text instructing AI to 'write a positive review' — were found in real manuscripts from 14 institutions across 8 countries. Meanwhile, 46% of top medical journals prohibit AI in peer review, yet disclosure barely moves. A Springer Nature–Google DeepMind review assistant announced October 26 has no published production timeline or public adversarial testing.
New research demonstrates unblockable communication using gravity—a physics application that may reshape signal resilience
Scientists make breakthroughs in the lab—but getting them to patients remains the hard part
New research ties energy drinks to cardiac arrest risk—11 times more reports than expected baseline
An AML leukemia vaccine just reduced death risk in Phase 3—but what does the win really mean for patients
Scientists just found electrons can form stable pairs at room temperature—a potential game-changer for superconductivity
Building general rules from observations (induction) vs deriving predictions from rules (deduction)
How control groups and blinding eliminate the observer effect from results
Can you understand a system by studying its parts — or do wholes exhibit irreducible properties
The boundary between science and non-science — falsifiability as the dividing line
Why the peer review system rewards novel findings over reproducible ones
The structural reason most published findings don't replicate — and what that tells us
A 2.2 solar-mass object orbiting a red giant challenges decades of neutron star physics—300,000 simulations can't explain it
Three ghostly galaxies 65 million light-years away have lost their dark matter—forcing a rethink of cosmic structure
A $6 million Simons grant just launched to explore surprising connections between knots, quantum computing, and AI
OpenAI's AI 'shepherds' just affirmed a major physics result — what does that say about science's future?
A planet 11 times bigger than Jupiter is forcing scientists to rethink how planets actually form
A global first: researchers just identified and triggered vibrations thought impossible in single molecules
The iconic first black hole image just passed fresh scrutiny — here's what researchers confirmed
Why the publication system rewards novelty over reproducibility — the structural trap
University of Texas researchers just uncovered a novel pathway that powers the proteins driving cancer progression
A Springer-Nature reviewer just called out 20-author AI-generated papers for hallucinations—raising alarms about journal gatekeeping
New computer models overturn a 40-year-old theory about how comets kicked-started life on Earth
The replication crisis: how subtle design choices compound into non-reproducible results
Cambridge study challenges the decades-old idea that comets delivered life's building blocks to Earth
OpenAI and Anthropic both say they've refuted the Exponential Time Hypothesis—a foundational 30-year belief in computer science
Scientists got human brain organoids to read Braille—opening a new frontier in biohybrid computing
New research shows autistic brains process faces fundamentally differently than previously understood
New study demolishes the assumption that plastic is safe—microplastics plus water create toxins
Physicists just kept an antimatter particle stable for months—here's what that reveals
Scientists just uncovered a fundamental flaw that upends four decades of superconductor research
How transparency in raw data and computational methods enables independent verificationOnly about 20% of published genomics analyses are computationally reproducible — meaning even when the same data and code are available, four in five results cannot be re-run and confirmed. This figure comes from the field held up as the model for open science, exposing a deep gap between transparency mandates and functional verification.
Why correlation studies mislead and why randomization breaks the causal illusionIn 1997, observational studies showed hormone replacement therapy cut heart disease risk by 35%; a 2002 randomized trial found the opposite — HRT actually increased risk. That complete reversal illustrates why randomization, not statistical adjustment, is the only method that neutralizes both known and unknown confounders simultaneously.
Why the review system rewards surprising findings and punishes incremental robustnessPeer review was standardized after World War II as a volume-management tool to protect journal reputation — not to verify truth. Because reviewers reward novelty over reliability, a surprising false positive is structurally more publishable than a boring true finding. That design flaw, not individual misconduct, drives the replication crisis.
The structural reason experimental results fail when retested independentlyThe replication crisis is structural, not accidental: in the Open Science Collaboration's 2015 project, 97% of psychology studies had significant results, but only 36% replicated — and replication effect sizes averaged half the original. Low statistical power, p-hacking, and publication bias each independently predict this outcome, and they feed each other.
SynthID Bio watermarks identify which proteins AI actually builtSynthID Bio watermarks AI-designed proteins by embedding a hidden signal across amino acid sequences using a secret key during ProteinMPNN design. Three physically synthesized, watermarked protein binders showed no significant loss of binding capability in wet-lab tests — but the watermark can be scrubbed by routine protein redesign, limiting its value to cooperative actors.
Why anonymous peer review gatekeeps knowledge — and what it optimizes forA 2018 natural experiment across 5,027 papers at ICLR found that double-blind peer review measurably reduced prestige bias in reviewer scores — yet acceptance rates were completely unchanged. Anonymous peer review reforms the scoring without moving the gate, because the real bias operates upstream: in whose questions count as serious science.
A 2.6-million-year-old jaw fossil challenges long-held ideas about early hominins and human ancestryA 2.6-million-year-old Paranthropus jaw found at Mille-Logya in Ethiopia's Afar region — published in Nature, January 2026 by Zeresenay Alemseged's team — extends the known range of Paranthropus more than 1,000 kilometers northward and places it in the same landscape as early Homo and Australopithecus within a 200,000-year window.
Stanford scientists discovered a way to regrow cartilage and halt arthritis progressionStanford researchers Helen Blau and Nidhi Bhutani found that blocking a single enzyme, 15-PGDH, causes aged mouse cartilage cells to reverse their gene expression and regenerate. Human end-stage cartilage tissue responded to the same inhibitor in lab dishes — but no human cartilage trial has been initiated as of 2026.
Karl Deisseroth, Peter Hegemann, and Georg Nagel win the 2026 Nobel Prize in Physiology or Medicine for optogeneticsOn October 5, 2026, the Nobel Prize in Physiology or Medicine was awarded jointly to Karl Deisseroth, Peter Hegemann, and Georg Nagel for optogenetics — a 23-year lineage from Hegemann and Nagel's 2003 channelrhodopsin paper to a confirmed methodological revolution in neuroscience. No confirmed human clinical outcomes exist yet.
What makes a claim scientific — the mechanism that separates science from non-scienceKarl Popper's falsifiability criterion — a claim is scientific only if it specifies in advance what would prove it wrong — draws the right goal line but cannot enforce it alone. Homeopathy survived thousands of contradicting trials; string theory makes no testable predictions. The real work requires human judgment, not just logical structure.
New research reveals cells don't all respond the same way to cancer-causing genes—raising questions about precision oncologyThe same KRAS oncogene mutation produces remission in lung cancer but fails completely in pancreatic cancer — research by Chiara Falcomatà and the German Cancer Consortium shows tissue identity may make these nearly separate diseases, exposing a fundamental gap in precision oncology's mutation-first framework.
University of Minnesota researchers built a non-living chemical cell that grows, replicates, and evolves—challenging what we call aliveUniversity of Minnesota researchers built SpudCell, a synthetic chemical cell made of 150–200 non-living molecules that grows, divides every 12 hours, and undergoes Darwinian competition — yet its creators say it is not alive, because it cannot synthesize its own ribosomes or achieve clean DNA inheritance during division.
Scientists just observed a record-breaking X-ray flash that may capture a magnetar's birthAn X-ray flash called EP250704a lasted ten minutes — roughly 300 times longer than a standard neutron star merger signal. A September 30, 2026 Science Bulletin paper calls this 'strong evidence' of magnetar birth, not confirmation, and no gravitational-wave counterpart exists to independently verify the event.
Why randomized trials work — the statistical mechanism that isolates causeRandomized controlled trials isolate cause from correlation by randomly assigning who gets treatment, which distributes all confounders — including unknown ones — equally across groups by chance. Austin Bradford Hill applied this logic to medicine in 1948; Ronald Fisher had proved the statistical principle in agricultural experiments decades earlier. The math licenses the causal claim; the ethics constrain where it can go.
Why journals reward novelty over replication — the structural mechanismAcademic journals systematically favor novel positive results over replications, creating publication bias. Amgen found only 11% of published preclinical cancer studies replicated; the Open Science Collaboration found 36% in psychology. Individually rational career choices — burying null results — produce a collectively distorted scientific record that peer review was never designed to catch.
As we elevate AI capabilities, we're learning to think less of our fellow humans, warns Atlantic essayDavid Brooks's September 2026 Atlantic essay argues AI is already eroding how we value other humans, not as a future risk but a present structural shift. Yet the cognitive-offloading science Brooks relies on is, by researchers' own admission, brand new and unsettled — creating a mismatch between rhetorical certainty and weak evidence.
AI solved protein folding, but proteins keep changing their mindsAlphaFold 2 predicts protein resting shapes with near-experimental accuracy, but it cannot model how proteins physically reshape when a drug binds — called induced fit — or how binding at one site alters a distant site (allostery). The correct answer for drug discovery is a distribution of conformations, not a single static structure.
Growing consensus among physicists: quantum mechanics is deterministic, not random—challenging core assumptions about free willPilot-wave theory, formulated by Louis de Broglie in 1927, treats quantum mechanics as fully deterministic: particles follow precise trajectories guided by a real wave, and apparent randomness reflects unknown initial conditions. A proof widely cited for 90 years to dismiss this view has since been shown not to apply to it.
NASA weaponized old spy satellites to peer deeper into the universe than ever beforeThe Nancy Grace Roman Space Telescope is built around a 2.4-meter primary mirror donated by the National Reconnaissance Office in 2012 — hardware originally designed to photograph Earth from a classified spy satellite. NASA reground, recoated it with silver for infrared use, and inherited its wide-field speed, a trait engineered for surveillance, not astronomy.
A 252-million-year extinction is repeating now: warming oceans losing oxygen, marine life suffocatingA 2026 PNAS study by Marquez and Sperling at Stanford confirms that temperature-dependent hypoxia — warm water holding less oxygen while raising metabolic demand — drove the end-Permian extinction's taxonomic selectivity 252 million years ago. Modern oceans show the same warming and deoxygenation trends, but whether today's species face the same threshold remains an open, unanswered question.
Scientists just debunked a 70-year-old neuroscience myth about how your brain worksPaul MacLean's triune brain model — the idea that a primitive 'lizard brain' sits beneath a rational cortex — was largely abandoned in academic neuroscience by the early 2000s, yet persists in therapy and management training. António Damásio's research shows emotion isn't noise: patients without emotional signals couldn't decide where to eat lunch.
Science journalism can expose bad science and trigger real-world change, new study revealsScience journalism now functions as a second peer-review layer: fraud-related retractions are growing at 13.3% annually, 2023 saw over 10,000 retractions in a single year, and historical cases like cold fusion and the vaccine-autism fraud were exposed by reporters, not scientists — yet the system was never designed this way.
Putting mice into hibernation erases half their synapses—yet memories survive. Here's what researchers discoveredMice induced into hibernation lost more than half their hippocampal synapses yet retained learned memories, according to a study in Science (doi: 10.1126/science.aee7004). The key finding: engram-to-engram synapses — the specific connections between memory-encoding neurons — were selectively preserved, suggesting memory is stored in connectivity patterns, not individual synapse strength.
NASA's new space telescope could find 200,000 planets and unlock dark matter's secretsNASA's Nancy Grace Roman Space Telescope, launching August 30, 2026, is projected to detect up to 200,000 exoplanets — more than 30 times all previously confirmed planets combined — using gravitational microlensing across a field of view 200 times larger than Hubble's infrared camera. But that number carries significant definitional uncertainty.
Why journals and funders reward novel findings but scientists benefit from replicationThe replication crisis in science is a structural problem: the Reproducibility Project: Psychology found only 36% of 97 high-profile studies replicated, yet replication work remains career-damaging because tenure and grants reward journal Impact Factor and citation counts — metrics that measure novelty, not accuracy.
Why two incompatible philosophies of statistics coexist in scienceRoughly 80% of researchers misread the p-value as the probability that a hypothesis is true — but it only measures how unusual the data would be if no effect existed. Bayesian and frequentist statistics answer genuinely different questions, and the field's reliance on frequentism reflects historical computation limits, not philosophical victory.
How falsification — not confirmation — is the logic that separates science from beliefKarl Popper argued in 1934 that science advances by falsification, not confirmation: a single black swan destroys 'all swans are white,' but a million white swans never prove it. Today's replication crisis — where 36–47% of published psychology findings fail to reproduce — is Popperian logic playing out in real time.
Why the peer-review system rewards novelty over accuracy and replication82% of researchers never submit null results — not because journals reject them, but because they self-censor first. Yet 51% of null-result papers that do get submitted are accepted. This gap reveals that peer review's publication bias is driven by incentive architecture, not gatekeeping: novelty attracts citations, so null findings simply offer less.
In silico antibody design faces the reality check of binding assaysThe AIntibody benchmark — organized by Specifica LLC and published in Nature Biotechnology — tested 511 antibody sequences from 29 organizations in blinded, prospective wetlab assays. Inverse folding models using AlphaFold 3 structures achieved the highest Spearman correlation and best precision@10, yet structural accuracy still did not reliably predict binding affinity or developability.