
Sign up to save your podcasts
Or


Two studies this year asked how common the brain disease CTE is among former NFL players, and they counted in very different ways. CTE can only be definitively diagnosed after death, so the hard part is deciding who gets counted.
In this episode, Emma and Ezra look at a 2026 study in JAMA Network Open by Gallini and colleagues, and a 2026 study in the BMJ by Daneshvar and colleagues. Both study deceased players, with brain findings from autopsy, so they show links, not causes. The first adjusts for the fact that families of players with symptoms may be more likely to donate a brain, and estimates that roughly three in four deceased former players had CTE at death. The second refuses to guess about players who didn't donate and reports a range, with a minimum of nearly one in four for players who died from 2016 to 2021. Along the way: why the two studies aren't fully independent, the difference between "went with" and "caused," and three questions worth asking of any claim.
This podcast uses AI-generated host voices. Emma Reed and Ezra James are not real people. The research, scripts, and fact-checking are directed by a human producer.
What to take away
For practitioners and policymakers
The research
Both papers draw largely on the same brain bank (UNITE, Boston University) and share several authors, so they are not independent datasets.
In the news
The fine print
Credits and links
In late October, New York enacted a law limiting what AI companion chatbots can do with users under eighteen. Behind it sits a simpler question: do these chatbots make people less lonely, or more?
Emma and Ezra look at two peer-reviewed studies from the same research group. A year-long survey of more than two thousand adults found a small link between social chatbot use and later emotional isolation, though only on one of its two loneliness measures. A preregistered two-week experiment with first-semester university students found that texting a randomly paired fellow student left students less lonely than texting a supportive chatbot, which was not detectably different from journaling. The survey can't show cause, and neither study tested the law or was designed around minors.
Along the way: why "not detectably different" isn't "nothing," and one question to ask of any claim about chatbots and loneliness.
This episode doesn't take a side on the New York law or any other policy; it covers what the research found.
This podcast uses AI-generated host voices. Emma Reed and Ezra James are not real people. The research, scripts, and fact-checking are directed by a human producer.
What to take away
For practitioners and policymakers
The research
Primary paper. Folk, D., & Dunn, E. (2026). How does turning to AI for companionship predict loneliness and vice versa? Psychological Science, 37(4), 276–286. https://doi.org/10.1177/09567976261427747 (open access). Found: in an exploratory survey of more than two thousand adults over a year, people who increased their chatbot use reported more emotional isolation four months later, and lonelier people used chatbots more, but the links were small and appeared on one of two loneliness measures. Limit: observational and not preregistered, so it can't show cause. Corresponding author: Dunigan Folk.
Companion paper. Li, R.-N., Folk, D., Singh, A., Ungar, L., & Dunn, E. (2026). Is a random human peer better than a highly supportive chatbot in reducing loneliness over time? Journal of Experimental Social Psychology, 125, Article 104911. https://doi.org/10.1016/j.jesp.2026.104911 (open access). Found: in a preregistered two-week experiment with first-semester students, those paired with a human peer ended up less lonely than both the journaling group and the chatbot group, and the chatbot group was not detectably different from journaling. Limit: short, one chatbot, one university, and the human pairs had met briefly in person first. Corresponding author: Ruo-Ning Li.
Both papers come from the same research group, so they are related, not independent.
In the news
The fine print
Credits
Hosts: Emma Reed and Ezra James (AI-generated voices, made with ElevenLabs). Outro music: created with Eleven Music. Website: frompeertoear.com. Contact and corrections: [email protected]. If we got something wrong, tell us and we'll fix it here.
Starting January 1, 2027, many adults on expansion Medicaid will have to document work, volunteering, or school to keep their coverage, and the federal rules for SNAP work requirements are being rewritten too. So what did researchers find when work requirements were actually tried? In Episode 2, Emma and Ezra look at two peer-reviewed studies.
The first followed Virginia SNAP participants when the state brought its work rule back in 2013, using a birthday cutoff at age fifty as a natural experiment: many more people left the program, and the study could not detect a gain in employment. The second looked at Georgia, the first state to attach work requirements to a Medicaid expansion: no clear change in coverage or employment against neighboring states, and coverage that lagged South Dakota, which expanded without a requirement. Along the way: why "no detectable gain" isn't "no gain," how turning fifty can work as an experiment, and three neutral questions to ask of any proposal.
This episode doesn't tell you what to think about these rules; it covers what the research found.
What to take away
For practitioners and policymakers
The research
In the news
The fine print
Credits
Hosts: Emma Reed and Ezra James (AI-generated voices, made with ElevenLabs). Outro music created with Eleven Music. Website: frompeertoear.com. Contact and corrections: [email protected]. If we got something wrong, tell us and we'll fix it.
This podcast uses AI-generated host voices. Emma Reed and Ezra James are not real people. The research, scripts, and fact-checking are directed by a human producer.
Overdose deaths in the U.S. have fallen three years in a row, and everybody has a theory about why. In our first episode, Emma and Ezra look at what peer-reviewed research can and can't tell us about one of those theories: criminal penalties for drug possession.
Colorado made possessing small amounts of fentanyl a felony in 2022. Oregon went the other direction in 2021 and decriminalized possessing small amounts of every drug. Two studies, two directions, one narrower question: when a state changes the penalty for possession, do overdose deaths change? Neither study found a detectable change, and both are honest about what they can't show. Along the way: why "no detectable change" isn't the same as "no effect," how researchers build a statistical twin of a state, and three questions worth asking of any proposal.
Colorado voters approved Proposition 132 on November 3, 2026. This episode doesn't tell you how to vote or what policy to adopt; it covers what the research found.
What to take away
For practitioners and policymakers
The research
In the news
The fine print
Credits
Hosts: Emma Reed and Ezra James (AI-generated voices, made with ElevenLabs). Outro music created with Eleven Music. Website: frompeertoear.com. Contact and corrections: [email protected]. If we got something wrong, tell us and we'll fix it.
This podcast uses AI-generated host voices. Emma Reed and Ezra James are not real people. The research, scripts, and fact-checking are directed by a human producer.
From the publisher's feed