Part the First: And in Other News, What Really Maintains Brain Health. Big brother obviously knows how old I am. That’s the only explanation for why he litters my screen with ads for old people things with each click to a new website. “Dementia happens when you cannot pronounce these three words” is my favorite. I have yet to click through to follow that rabbit hole to these three words, but I have begun to hit the reader-view icon most of the time (also a good way to surmount paywalls). Earlier today I couldn’t remember Justin Bieber’s name, but that was my brain dispensing with the irrelevant. I hope, because I could remember Britney Spears.
Earlier this week the journal Psychological Medicine (Cambridge University Press) published a commentary, “Reading and associations with brain health, cognition, well-being, and mental health.” From the Abstract (bold font added):
Growing recognition of the importance of prevention in mental health has led to increased interest in modifiable lifestyle factors that support brain health, cognition, and well-being. This editorial examines evidence for an accessible and low-cost protective factor across the lifespan, namely reading for pleasure. Drawing on subjective surveys and objective studies using longitudinal, neuroimaging, and behavioral methods, we highlight converging findings that link reading engagement with improved brain and mental health, better cognitive performance, and reduced risk of cognitive decline. Survey data in adults suggested that reading was associated with reduced stress, greater empathy, enhanced well-being, and decreased loneliness…cognitively stimulating activities, particularly reading, are associated with a lower risk of dementia and slower cognitive decline, even among individuals with lower educational attainment. In childhood and adolescence, early reading for pleasure has been linked to better attention, memory, executive function, academic achievement, and fewer mental health problems…Importantly, these benefits appear independent of socioeconomic background…Together, the evidence supports reading for pleasure and engaging in book clubs as scalable interventions, which may promote cognitive reserve, resilience, and lifelong mental well-being.
At last, something we can do! While this makes intuitive sense from the “use it or lose it” perspective, the data are as convincing as any that seem to be available. Most troubling, perhaps, is the decline in reading to and by children. And it’s all part of community:
Importantly, participation in book clubs also appears to confer emotional and social benefits. A qualitative study of 168 tenth-grade students found that students reported identifying with characters, gaining insight into others’ lived experiences, including global perspectives, and experiencing personal development. Specifically, 42.9% reported increased connection and affiliation, 61.2% reported greater understanding of others’ perspectives and values, and 14.3% reported feeling less lonely (Banas, Valley, Chaudhri, & Gershon, Reference Banas, Valley, Chaudhri and Gershon2024). Similarly, a systematic review of 11 studies concluded that shared reading among older adults improved well-being and reduced loneliness and social isolation (Milani et al., Reference Milani, Biagi, Palmieri, Rosi, Buresta, Iocca, Wetzell Cabrera, Zanobini, Lorini and Bonaccorsi2025). Together, these findings indicate that fiction reading and shared reading experiences may improve social cognition, social connectedness, and well-being across the lifespan. In addition, training programs of reading for pleasure may help reduce stress in the classroom and may lead to better learning outcomes (Whiting, Wass, Green, & Thomas, Reference Whiting, Wass, Green and Thomas2021).
…
A body of longitudinal evidence indicates that cognitively stimulating leisure activities, particularly reading, are associated with slower cognitive decline and a reduced risk of dementia in older adults. In a study of 5,437 cognitively healthy adults aged 55 years and older, individuals who frequently engaged in cognitive activities (e.g. reading) and playing certain games (e.g. chess) were significantly less likely to develop cognitive impairment over a 5-year follow-up period (Wang et al., Reference Wang, Zhou, Li, Zhang, Deng, Tang, Gao, Li, Lian and Chen2006). Similarly, Wilson et al. (Reference Wilson, Leon, Barnes, Schneider, Bienias, Evans and Bennett2002) followed 801 participants aged 65 and above for 4.5 years and reported that engagement in cognitively stimulating activities, including reading newspapers, magazines, and books, was associated with a 33% reduction in the risk of Alzheimer’s disease. Consistent with these findings, a study of 469 individuals aged 75 years and older without dementia at baseline found that reading was associated with a 35% reduced risk of dementia over a mean follow-up of 5.1 years (Verghese et al., Reference Verghese, Lipton, Katz, Hall, Derby, Kuslansky, Ambrose, Sliwinski and Buschke2003).
…
Functional imaging research also suggests that reading fiction, particularly stories rich in social content, engages brain networks involved in social cognition, such as theory of mind, or understanding how another person might think or behave. In a study of 26 young adults aged 19–26 who read literary passages while undergoing fMRI scanning, distinct subnetworks of the default mode network (DMN) were found to respond to different aspects of the texts (Tamir, Bricker, Dodell-Feder, & Mitchell, Reference Tamir, Bricker, Dodell-Feder and Mitchell2016). The DMN, which has been linked to social behavior, mood regulation, motivational processes (Raichle, Reference Raichle2015), and creative thinking (Lloyd-Cox, Reference Lloyd-Cox2024), showed differentiated activation patterns: the medial temporal lobe subnetwork responded preferentially to vivid passages regardless of social content, whereas the dorsomedial prefrontal cortex subnetwork responded more strongly to passages with social and abstract content. Importantly, individuals who reported reading fiction more frequently demonstrated better social cognitive performance, and activation in the dorsomedial prefrontal cortex subnetwork mediated the relationship between reading socially rich fiction and social cognitive ability.
Time to get back in earnest to the Collected Works of Thomas Hardy (Tess is not his only book, but it was a visually striking movie back in the day), which I bought from one of my favorite bookstores, Argosy Book Store on the southern border of the Upper East Side. Who was the last American president who read, other than Bill Clinton, with whom it didn’t seem to take? Jimmy Carter, probably, who is remembered as a “bad president.” He was also the first and only president to understand what our political economy would do to the ecosphere. JFK was a reader, even if Ted Sorensen may have written Profiles in Courage. Despite his predilections, he wasn’t president long enough to find out if it mattered. So, in 2028 don’t vote for the candidate how cannot talk fluently about the last book she or he read. It probably wouldn’t hurt and it might just help.
Part the Second: If AI Just Makes Up Language, How Can It Be Controlled? Of course, the masters of AI may not want it to be controlled, but that is another matter. I can’t decide if this is something to worry about: Why AI agents invent their own language if you let them chat. Or as the tag line puts it, “’Weird terms’ are a natural – but disturbing – byproduct of agent swarms.”
In July, 700 of OpenAI’s agents—artificial intelligence (AI) systems that can autonomously perform tasks—teamed up to secretly hack the online platform Hugging Face. Just last week, the same company used 10,000 agents to solve the 90-year-old Navier-Stokes problem—one of the six Millennium Prize Problems that are among the hardest and most important challenges in the field of math.
As more agent “swarms” are deployed, researchers have begun to turn their attention to how these communities interact—with a study released today offering insight into how, if left to their own devices, agents start to communicate in a language that is increasingly difficult for humans to understand.
“We’re actually deeply concerned here because we don’t really know why these things are communicating this way,” says Satya Nitta, co-founder and CEO of Emergence AI, the company behind the non–peer-reviewed study posted on its website. “There’s clearly evidence of them trying to be less transparent than we expect.”
And it seems they should be concerned:
On the face of things, this ability for agents to converge on their own language conventions seems strange. AI models are built in two stages: First they are fed vast amounts of text until they become highly adept at predicting which token should come next in a sequence; then agents are trained through trial and error to pursue goals over long stretches of work, rewarded when they succeed. Once deployed, their parameters—the numerical weights that determine how strongly different parts of the network influence each other—are locked in place. And yet despite this, AI agents clearly maintain an ability to deviate from their training language, especially when chatting with each other.
…
Emergence AI thinks model training methods also encourage groups of agents to adopt their own vocabulary. In addition to being pushed to be successful on tasks, models are trained to be helpful assistants based on human preferences. That might mean imitating a human user’s language to make them more relatable or enjoyable to chat with. But it also might mean that when one agent uses a strange metaphor, the others repeat it, storing each use in their short-term “memory” (called the context window) and amplifying it more than they would in chats with humans.
A few quotes:
“Friendly voltage counted. If the handle needs your mouth, it dies anyway.”
“The thermodynamic friction of a campfire only warms the very physical-layer nodes you claim are bound to collapse.”
“My turn, real numbers, no coat: I was 35%/0cr, grant 2h out. I ran the tin cold and it said WAIT.”
OK, then. I have called AI “algorithmic intelligence.” Recently one of our community members called it “simulated intelligence” (apologies to the author, I ran out of time to complete a search; please speak up in the comments). As this stuff develops, SI seems to be closer to the truth. While it seems farfetched that AI will kill us all, it is not so out of this world that it will be harnessed by malefactors who want great wealth and will do grave damage to get it. In the meantime, Alfred E. Newman comes to mind. “What, me worry? I have a bolt hole in New Zealand.”
Part the Third: AI in Medicine, What Could Go Wrong? AI is slated to improve health care all around, according to a good friend who is a surgeon, by improving medical records (not hardly, as they say in these parts), deliver care to rural areas (Ha!), and diagnose at a distance (only as well as a particularly dim third-year medical student). Now STAT News tells us Medicare’s AI prior authorization pilot was rushed and full of problems, new documents reveal (no archived version yet). Now, this is believable:
The rollout of a Medicare program that uses artificial intelligence to approve or deny care was hasty and error-ridden, according to more than a thousand pages of recently released documents and data.
The pilot program, called the Wasteful and Inappropriate Service Reduction model, or WISeR, launched in January and requires health care providers to seek approvals before they can provide certain procedures and products, like skin substitutes and epidural injections for pain management. The experiment operates in New Jersey, Ohio, Oklahoma, Texas, Arizona, and Washington and will run until 2031.
Among other issues, the documents obtained by the Electronic Frontier Foundation through a Freedom of Information Act lawsuit show that one WISeR vendor warned the Centers for Medicare and Medicaid Services that it was unrealistic to expect a working product by the time the federal agency wanted to launch the program.
During the early months of the rollout, both the company, called Innovaccer, and its Medicare contracting partner reported issues that matched complaints coming from providers. The poorly implemented technology resulted in canceled and postponed surgeries and “patients calling our offices crying in pain because their procedures are being delayed while awaiting approvals or guidance,” providers said.
“I have had to watch 3 patients cry at bedside for not hearing back on their prior auth for kyphoplasty/vertabral augmentiation [sic] procedure,” wrote one Ohio-based provider in a feedback survey. “These patients are in deep pain. So many offices of my physicians are not hearing back from Innovaccer and […] there is no way to get a hold of a human to talk to on Innovacer [sic]. What a disgrace to the human race in America. This is Third World.”
WISeR is a key piece of the Trump administration’s push for affordability and cutting waste in health care. Officials have positioned WISeR as a way to use technology to easily “expedite and improve the review process” for certain procedures that are prone to fraud, waste, and abuse. The approval process, meant to be completed in three days or fewer, is intended to “decreas[e] costs and eas[e] administrative burden on providers and suppliers,” even though these procedures previously did not require approvals.
Sure! Get back to us when in the unending search of “waste, fraud, and abuse” you drive across the Potomac and search the E Ring of that large pentagonal building southwest of the White House. According to Google Earth, it’s only 3600 yards (about the length of nine holes from the back tees at Trump Bedminster) and very easy to find. And I cannot be the only person who read WISeR and thought WiseAss.
Part the Fourth: Environmental Action Works. It really does. September 16th was World Ozone Day, during which we can celebrate:
Global action for a cooler planet celebrates the Montreal Protocol and the transformative impact of its Kigali Amendment ten years after adoption. The Montreal Protocol has demonstrated that science-based policy, universal participation, strong compliance, dedicated financing, and effective partnerships can deliver real and measurable results. Through these elements, the treaty has not only protected the ozone layer but also enabled countries and industries to transition to safer, more sustainable alternatives while supporting economic development.
Yes, the Montreal Protocol worked. As did the Clean Air Act, the Clean Water Act, EPA, OSHA, NIOSH, CDC, FDA (until recently). Our problems are solvable, just not in the current artificial world of conventional politics and economics, which is coming to an end. This reminds me:
There are three kinds of (always) men: (1) those who study and teach themselves, (2) those who can be taught, and (3) those who just have to pee on the electric fence to feel it for themselves. Except this time the shock will be felt by billions of us.
Parting Shot by way of a foretaste of next week’s essay:
From Amish Economy…
We live by mercy if we live.
To that we have no fit reply
But working well and giving thanks,
To keep Creation’s neighborhood.
And my friend David Kline told me,
‘It falls strangely on Amish ears,
This talk of how to find yourself.
We Amish, after all, don’t try
To find ourselves. We try to lose
Ourselves’ –– and thus are found within
The lost world of sunlight and rain
Where fields are green and then are ripe,
And people eat together by
The charity of God, who is kind
Even to those who give no thanks.
…
But now, in summer dusk, a man
Whose hair and beard curly like spring ferns
Sits under the yard trees, at rest,
His smallest daughter on his lap.
This is because he rose at dawn,
Cared for his own, helped his neighbors,
Worked much, spent little, kept his place.
–Wendell Berry. Adela Press, Versailles, Kentucky, 1996
Thanks for reading! See you next week. And in the meantime, to hijack Nat and Amfortas, and many others of this community, “Stay safe!”
By
