Evidence (6/10): The 14% increased CTE risk per additional year in male rugby players is a specific statistic, and the gender research gap figures are cited with years. The study itself is not yet published, so the article is reporting on methodology rather than findings, which is acknowledged. However, two significant claims are left unsupported. The headline statement that “female brains are softer and more vulnerable” is attributed to the lead researcher without mechanistic explanation. Softer in what sense? Tissue density, white matter, skull thickness? These are meaningfully different things with different implications. Similarly, the claim that “women’s and men’s rugby are played quite differently”, which is central to the scientific rationale, is stated without any elaboration on biomechanics, impact forces, or tackle technique.

Balance (8/10): Responsibly avoids scaremongering while clearly communicating genuine risk. The goal of informing rather than discouraging participation is explicitly stated.

Clarity (7/10): Engaging narrative structure using the players’ experiences effectively grounds the science in human terms. However, the two unsupported claims actively create confusion rather than clarity for a general audience, pulling this score down.

Overall (7/10): Highlights an important and under researched area responsibly, but leaves two scientifically significant claims inadequately explained.

Evidence (9/10): Multiple named experts from legitimate institutions such as LSE, University of Sussex, NYU, and Cambridge are quoted. Referenced events like the Google engineer case are accurately reported. Relying on expert opinion rather than peer-reviewed literature is appropriate here, since no validated scientific framework for measuring consciousness in any system currently exists. The confident dismissal by Gary Marcus and Dawkins’ equally confident conviction are deliberately juxtaposed, with more moderate voices like Shevlin and Sebo implicitly framing the middle ground of genuine uncertainty as the most defensible position.

Balance (10/10): Exceptional for a topic this controversial. Dawkins is represented fairly in his own words, sceptical voices are included without dominating, and Shevlin’s warning against dogmatic dismissal reflects the most philosophically rigorous position available. The spectrum from “definitely not” to “we don’t know” to “will become more plausible” mirrors the genuine state of scientific and philosophical debate.

Clarity (8/10): Engaging and well-structured. The opening anecdote draws readers in effectively and the intelligence vs. consciousness distinction is clearly explained. Minor deduction for not defining “agentic AI” for a general audience.

Overall (9/10): Handles an unresolved scientific and philosophical question with rare intellectual honesty and sophisticated editorial structure.

Evidence (8/10): Based on a legitimate peer-reviewed study in Nature Neuroscience with a full citation and DOI. The Sox9-MEGF10-phagocytosis mechanism is accurately described and the experimental design is fairly represented, including the important detail that mice already had cognitive impairment before intervention. The six-month cognitive preservation finding is reported without exaggeration. Loses points for not mentioning sample sizes or effect sizes, and for not acknowledging that amyloid clearance in mice has repeatedly failed to translate to humans (a well-documented problem in Alzheimer’s research that deserves mention).

Balance (7/10): The article is appropriately cautious about the mouse-to-human translation gap, explicitly stating more studies are needed. However, it presents amyloid plaques as a central driver of Alzheimer’s without acknowledging that this remains contested. The amyloid hypothesis has faced significant challenges in recent years.

Clarity (8/10): Well-structured and accessible. The “vacuum cleaner” analogy for phagocytosis is effective, and the explanation of astrocytes as overlooked support cells is clearly explained. The logical progression from problem to mechanism to result to implication is clean and easy to follow.

Overall (8/10): A quite rigorous piece, held back by omitting the broader debate around the amyloid hypothesis and the historical difficulty of translating mouse Alzheimer’s models to humans.

Evidence (7/10): The 2018 study covering nearly half a million children is accurately represented, including the important caveat that only 66 blind children were included. The distinction between cortical and peripheral blindness is scientifically sound and central to the argument. The predictive processing framework for schizophrenia is legitimate but presented as more settled than it is. The attribution of the 1950 observation to “writer Hector Shevini” is unverifiable and unusual. The article may have been translated from another language, and names or credentials may have been distorted in the process. This undermines confidence in the historical framing even if the core science holds.

Balance (7/10): Appropriately cautious overall, explicitly stating blindness couldn’t serve as a practical safeguard and acknowledging early-stage research. However, predictive processing is one of several competing theories of schizophrenia, and that debate is absent.

Clarity (8/10): The clearest of the articles reviewed. The explanation of predictive processing and why the visual cortex matters is accessible without oversimplifying, and the logical chain from observation to hypothesis to treatment flows naturally.

Overall (7/10): Solid science journalism with accurate core findings, let down by unverifiable sourcing and presenting an emerging theoretical framework as established consensus.

Evidence (5/10): Almost no sources are named, cited, or linked despite repeated appeals to research. Claims like “neuroimaging studies have shown heightened amygdala activity” and “research shows strong effectiveness for CBT” are presented as established fact without attribution. The neuroscience of the freeze response and vagus nerve is plausible but presented with more certainty than the evidence warrants, particularly given the article’s own admission that “research on adult selective mutism specifically remains limited.”

Balance (5/10): Written by ReachLink, a therapy platform selling its own services, a significant conflict of interest that is never disclosed. Treatment recommendations consistently funnel toward their own platform, blurring the line between clinical guidance and advertising. Alternative perspectives on diagnosis or treatment are entirely absent.

Clarity (7/10): The article’s strongest dimension. Empathetic, well-structured and accessible for a general audience. The comparisons between selective mutism, shyness, social anxiety and avoidant personality disorder are genuinely useful and clearly explained. The science, while poorly cited, is at least communicated in understandable terms.

Overall (5/10): Readable and empathetic, but significantly undermined by absent citations, a clear commercial agenda, and overconfident claims in an area where the evidence base is admittedly thin.

Evidence (8/10): Based on a legitimate study from the Institute of Cancer Research and Imperial College London. Causal language is mostly careful, “key factor” and “associated with” rather than “causes.” Specific prevention percentages are concrete and attributable. However, the study is observational, and the article could be clearer that association is being inferred as contribution rather than established causation. The acknowledgment that obesity doesn’t fully explain the rise is a responsible inclusion.

Balance (7/10): Multiple expert voices are included and the article avoids overselling obesity as the sole explanation. However, describing obesity exclusively as a “behavioural risk factor” (alongside smoking and alcohol) is scientifically reductive. Obesity has well-established genetic, hormonal, pharmacological and socioeconomic contributors that are entirely absent from this framing, which risks reinforcing stigmatising narratives around personal responsibility. Other hypothesised drivers of early-onset cancer such as ultra-processed foods and environmental exposures are also unaddressed.

Clarity (7/10): Generally accessible with concrete statistics. However, “behavioural risk factor” is used without explanation and is potentially misleading for a general audience, implying obesity is purely a lifestyle choice rather than a complex, multifactorial condition.

Overall (7/10): Responsibly reported at surface level, but the reductive framing of obesity undermines both.

Evidence (8/10): The scientific conclusions are well-supported by the sources cited. The article correctly concludes that personality is polygenic and poly-environmental, and this is backed by the 2015 meta-analysis, GWAS heritability data, and expert quotes. It appropriately avoids overreaching, and conclusions stay within what the evidence allows. Minor deduction for drawing on a non peer-reviewed study and for the warrior gene section, where the legal case framing risks implying stronger gene-behavior causation than the science supports.

Balance (9/10): The article presents nature and nurture evenhandedly and includes the important caveat around ancestry gaps in GWAS research. The trauma narrative is challenged without dismissing it entirely. Small deduction for not acknowledging ongoing debate around the Big Five model as a framework (accepting it uncritically as settled science is a mild bias).

Clarity (8/10): Most concepts are explained accessibly and accurately. However, the non peer-reviewed study is presented without flagging clearly to readers what peer review means and why its absence matters.

Overall (8/10): Scientifically responsible journalism whose conclusions are proportionate to the evidence, fairly presented and mostly clearly explained.

Evidence (7/10): Based on a single peer-reviewed study in JAMA Psychiatry with a solid sample size of 1,154 brain scans. The findings are accurately represented : the 4 subtypes, the 45 vs 26 abnormal regions distinction, and the prefrontal cortex/pallidum involvement are all grounded in the study. The article correctly flags that brain imaging is not yet clinically practical at the individual level. Minor deduction for presenting one study’s clustering results as broadly conclusive without mentioning replication needs.

Balance (8/10): Notably stronger than typical science journalism. It includes multiple expert voices — the lead researcher, a clinician, a psychiatrist, and a DSM working group neuroscientist — and represents both enthusiasm and caution. DelBello’s caveat about imaging being too expensive and imprecise for routine clinical use is a responsible inclusion. Small deduction for not acknowledging debate around ADHD overdiagnosis or the limitations of neuroimaging-based subtyping more broadly.

Clarity (9/10): The pop culture references (Veruca Salt, Angelica Pickles) and clinical analogies (“simmering volcanoes,” “super bouncy ball kids”) are effective for a general audience. The prefrontal cortex and pallidum are explained in accessible functional terms. The distinction between “more ADHD” and “meaningfully different” is clearly communicated.

Overall (8/10): A well-reported, responsibly hedged piece that accurately reflects the study’s findings and limitations.

Evidence (4/10): The article is based on a single peer-reviewed study in Molecular Psychiatry, which is a strength. However, conclusions are frequently overstated relative to what one study can support. Claims like shared brain-gene patterns pointing to “a common origin” go beyond what the evidence establishes. The gene-expression overlap is described as fact rather than as an association finding. No effect sizes, sample details, or methodological limitations are mentioned.

Balance (4/10): The article presents the overlap hypothesis enthusiastically without acknowledging that the categorical diagnostic model still has strong scientific supporters. It does not mention that autism and ADHD have well-documented distinct features and genetic architectures. The quote from Dr. Di Martino is the only expert voice, and no counterpoint or scientific debate is represented.

Clarity (5/10): The writing is accessible and uses good analogies (e.g. “cleaning up a messy room” for synaptic pruning). However, key concepts like resting-state connectivity and gene expression are glossed over in ways that could mislead. The article merges brain connectivity patterns with genetic causation without clearly distinguishing the two.

Overall (4.3/10): Readable and based on a legitimate study, but overreaches its conclusions, lacks balance, and sacrifices scientific precision for narrative flow.