iiinova — where your digital self keeps evolving

David Epstein @davidnotdave.bsky.social · 31/07/2026

My red text is on the right-hand side of the screencap. If you understand factor analysis and/or the circumplex model of affect, please explain it to me as if I were a toddler. And please don't be mean.

Screencap of "Independence and Bipolarity in the Structure of Current Affect" (1998), but Lisa Feldman Barrett and James Russell. My question is about how valence and arousal can be independent of each other while being represented as the circumference of a circle (whereby changes in one will restrict the range of the other). Shouldn't they be the full interior of a square?

000

David Epstein @davidnotdave.bsky.social · 16/07/2026

Here's a study protocol (no results, just protocol), newly published in a peer-reviewed journal. Looking at the dates, I don't know why. If you want your protocol publicly available, just put it on a preprint server.

To appear in:
Contemporary Clinical Trials Communications
Received Date: 2 June 2025
Revised Date: 5 June 2026
Accepted Date: 26 June 2026

Results
This study is ongoing and data collection is expected to be concluded in July 2025, with plans to publish the study results by November 2025.

100

David Epstein @davidnotdave.bsky.social · 07/05/2026

Not sure how this figures into the ol' h-index.

Screencap from John Oliver's May 4 "Last Week Tonight," showing a paper about tianeptine that was written by Kirsten Smith, Jeff Rogers, Justin Strickland, and me.

2304

David Epstein @davidnotdave.bsky.social · 29/03/2026

So, don't look for the Helpers, is what I hear you saying?

Screenshot of an 1970s SNL commercial parody called "Placenta Helper," which famously never aired.

110

David Epstein @davidnotdave.bsky.social · 28/03/2026

Made my own sign for No Kings. No use of generative AI here. I did it the way the Amish would do it: Photoshop.

A graph of approval/disapproval ratings for 47 as their shape widens over time like a sideways letter V (disapproval up from 41% to 62%, approval down from 47% to 36%). They form a set of shark's teeth, inside the mouth of a shark in pursuit of him. Header, above the shark: "Help him 'tank.'"

020

David Epstein @davidnotdave.bsky.social · 18/10/2025

Signs made for tomorrow. Because every season is spooky season!

Side one of sign: Stephen King's "Misery" book cover Photoshopped to say "No Kings, Less Misery."

Side two of sign: Stephen King's "It" book cover Photoshopped to say "Kings Ain't It."

032

David Epstein @davidnotdave.bsky.social · 07/08/2025

I’m proud of the way this paper turned out. We said: --Stop revictimizing people. Don’t say, “You were sexually abused, so you’re more prone to iatrogenic addiction, so we won't prescribe opioids for your pain.” --Monitor whole communities. link.springer.com/article/10.1007/s11469-025-01527-w

Text excerpts from "Latent Classes of Adverse Childhood Experiences (ACEs) and Adult Substance-use Problems and Psychosocial Outcomes: Complex and Heterogeneous Associations." (International Journal of Mental Health and Addiction)

1113

David Epstein @davidnotdave.bsky.social · 28/06/2025

Unsuccessful trial, good preprint. I fervently hope the final sentence makes it through peer review.

Micro interventions based on real-time monitoring of affective instability: a proof-of- concept study.
Merlijn Olthof et al.

Excerpt 1 of 3:
Design: We used a quasi-randomized within-person experimental design for testing the micro-intervention effects during increased EWMSD [exponentially weighted moving standard deviation of affect ratings] vs. control periods (with EWMSD below the threshold). At the start of the study, participants were assigned at random to one of two phases: the experimental phase (micro-intervention is applied when there is increased EWMSD) and the control phase (micro-intervention is applied semi-randomly).
[David’s comment: This is the appropriate control for Forer effects, and no one else ever uses it!]

Excerpt 2 of 3:
Interestingly, from the data viewing sessions, we also learned that there was a large heterogeneity in how persons experienced - and reacted to - the intervention and burst measurements. While some persons were really enthusiastic about the intervention, others found it annoying or confronting because they found that they had no time to engage in the intervention. Some persons never actually performed the activity they selected, some nearly always performed it in the next 30 minutes, but others used it as 'a reminder'…
[David’s comment: This is the kind of frankness we need more of.]

Excerpt 3 of 3:
These results can inform future research, which, who knows, may at some point be able test the hypothesis that we were after.
[David’s comment: OK, marry me.]

050

David Epstein @davidnotdave.bsky.social · 24/05/2025

Pinker's assertion about trigger warnings (that they "can do more harm than good") is based on research that doesn't generalize to the intended situations, as I told the author five years ago.

Tweets from June 8, 2020.

Payton Jones: “You may want to reconsider your use of trigger warnings. Our new paper, just appearing in Clinical Psychological Science, suggests they may do more harm than good.”

Me, in a series of five replies: “This is a nicely conducted experiment, but it's a very limited operationalization of the broad question it means to address— limited both by the stimuli used and the responses assessed….The "markedly distressing" cues (text from Dosteovsky, Ballard, etc.) were tame indeed. I don't see anything in the Results to indicate that they even markedly bothered anyone! So, not a very stringent test of whether trigger warnings help or harm. Re responses assessed: If nothing else, use of TWs conveys courtesy and respect. Presentation of highly disturbing material without TWs can convey an indifference or erasure that might be inherently distressing to people with trauma histories. Your study didn't assess that aspect of people's responses to your stimuli and TWs--a major omission in any consideration of TWs' cost:benefit or risk:benefit ratio. (But, again, you would have needed nastier stimuli to assess this.) In sum: an interesting demonstration and a very worthy endeavor, but I wouldn't use it to conclude that if I plan to show graphic films of sexual assault to a roomful of survivors, the best thing I can do for them is omit a warning. This study won't generalize to that.”

2996

David Epstein @davidnotdave.bsky.social · 02/05/2025

New weekend of Tesla Takedown protests, new two-sided sign. (I'm reading 𝘍𝘶𝘯𝘯𝘺 𝘉𝘦𝘤𝘢𝘶𝘴𝘦 𝘐𝘵'𝘴 𝘛𝘳𝘶𝘦, Christine Wenc's excellent book about the history of The Onion.)

Side one of sign, from an old issue of the Onion's "American Voices" feature: "Phrases like 'massive redistribution of wealth' have historically brought rivers of blood. That said, it's time for a massive redistribution of wealth."

Side two of sign: "Austerity? You go first, E.M. I'll join in if it looks good."

041

David Epstein @davidnotdave.bsky.social · 26/04/2025

New sign made for tomorrow's Tesla Takedown. As a scientist, I've had it drummed into me--you're always supposed to show data.

Graph of Tesla stock freefalling: "You can't keep a good man down. But THIS schmuck? You can."

050

David Epstein @davidnotdave.bsky.social · 05/04/2025

My two-sided sign for today's "Hands Off" protest outside the Social Security Administration building in Catonsville, MD. The Baltimore Banner put the crowd at "over 300," but I think it was 𝘰𝘷𝘦𝘳 over that. And Catonsville is but a suburb.

Side one of sign: "Let's get a little more juice in that third rail."

Side two of sign: "Uncap FICA."

1142

David Epstein @davidnotdave.bsky.social · 01/04/2025

Going cold, turkey. I saw this paper on April 1, but it was not published on April 1.

Abstract:
Craving is a central phenomenon in cocaine use disorder (CUD), significantly influencing relapse risk and treatment outcomes. The STANDUP® method, an innovative telemedicine-based intervention, integrates structured behavioral modification, cognitive restructuring and digital therapeutics to target reward, relief and obsessive craving subtypes.

The six-month structured routine program includes daily behavioral exercises that focus on:

• Breaking addictive cycles through structured daily planning (Mappa del Giorno).
• Cold showers as a physiological and psychological tool for increasing self-discipline and stress tolerance.

000

David Epstein @davidnotdave.bsky.social · 30/03/2025

This title made me happy out of all proportion to the stimulus. I want to thank it. Right now it's only a preprint. Together, psychological-science Bluesky, we can commit to editor/reviewer stances that might bring it all the way to publication. And that would be kinda great.

Alcohol use disorder symptoms are kinda detectable in daily life (but it depends on how they are operationalized)

Kevin King
Dani Kang
Megan Schultz
Christine Lee
Jonas Dora
Cassandra Boness
Ashley Watts

1131

David Epstein @davidnotdave.bsky.social · 15/03/2025

On both sides of the road in Owings Mills, MD, there were ~200 of us making it less socially comfortable to do business with Tesla. It matters. Purchases are elections that happen every day. Nearly nonstop response of thumbs-up honking from drivers-by, including 18-wheelers. Salt of the earth.

1130

David Epstein @davidnotdave.bsky.social · 07/12/2024

You've created a situation whereby I have to post one of my old tweets.

Screencaps of tweets from 2022, in which someone wrote:

"i feel like you could shrink joss whedon's face and no one would notice. like 15-20% shrinkage and you could do it undetected. he might not even notice it in the mirror"

So I obliged.

010

David Epstein @davidnotdave.bsky.social · 06/12/2024

I had an invitation lined up for my virtual attendance to #ACNP2024, but we left it too long, and ACNP has shut the door. Attendees, please post prolifically about the goings-on. I ask with puppydog eyes. If puppydog eyes could talk.

From the ACNP website: "NEW in 2024: Member invited guest invitations will be limited to 400. The College has reached the limit of 400 member guest invitations. Member guest invitations are no longer available for the 2024 Annual Meeting."

020

David Epstein @davidnotdave.bsky.social · 01/12/2024

I haven't posted enough for this RoasterBot to work well, but it kinda worked well anyway.

Screenshot of output from an AI app called the Bluesky Roaster (blueskyroast.com), based on my Bluesky content.

"David Epstein: The Neuroscience Nerd with a Side of Sarcasm."

"David's tweets are like an academic paper that took a detour through a comedy club—insightful, often dense, but always with a punchline waiting at the end. Just when you think you're lost in the jargon, he'll throw in a zinger about convenience samples that feels like a punch to the gut (in a good way)."

110

David Epstein @davidnotdave.bsky.social · 15/11/2024

I'd like to see more of this in science: say what decisions you made, in what order, and why. Btw, it's not about preregistration. Dishonest people can "preregister" analyses they've already done. Honest people can be honest without preregistering. It's the honesty and clarity I applaud.

At the time of preregistration, we liked this idea because it is a more stringent criterion compared to statistical significance. However, applying this criterion in practice made us realize that this is not the most suitable strategy for replication: 1) it is arguably overly stringent, requiring near-perfect alignment between the studies, which is unrealistic given natural sampling variability; 2) it can lead to an apparent lack of replication even if estimates are close if Cls are narrower in the replication study. For that reason, below we discuss our results both in light of this preregistered criterion and the overall strength and direction of effects.
Megan E. Schultz, M.S.
Jonas Dora, Ph.D.
Kevin M. King Ph.D.

4304

David Epstein @davidnotdave.bsky.social · 31/10/2024

I think this is the first year we've ever created a pumpkin situation here. Jack o'liberty. Treats were issued a moment ago to a smol RBG.

Illuminated white pumpkin on our steps, HARRIS-WALZ '24 carved in.

020

David Epstein @davidnotdave.bsky.social · 16/06/2024

And Christina didn't see a thing. (I have no good explanation for why I did this.)

Andrew Wyeth's Christina Photoshopped onto the slope of the grass in Kurt's photo

110

David Epstein @davidnotdave.bsky.social · 14/06/2024

Our new #kratom paper explains frankly why we're putting these data into press now. Sometimes you publish because you don't want to soft-pedal or file-drawer counterpoints to your other work. www.frontiersin.org/journals/pharmacology/articles/10.3389/fphar.2024.1412397/

...the following report, which we originally planned to publish in 2023. We had initially framed it as an update on a rapidly evolving situation of kratom product diversification...but we now offer it for a more enduring purpose: to help ensure that peer-reviewed publications about kratom do not reflect the polarization seen in popular discourse. If specific groups of investigators consistently emphasize only positive or only negative aspects of kratom use, the credibility of their contributions may suffer. Our recent publications about kratom largely illustrated how kratom use appears to benefit many consumers. Most of our participants had found products and dosage patterns that appeared...satisfactory to them and not detrimental to their health or daily functioning. We stand by those findings, but we have always noted that they are not the whole story....Here, we give voice to consumers who believe that their relationship with kratom was not unambiguously positive or beneficial...

041

David Epstein @davidnotdave.bsky.social · 18/04/2024

"Did you come up with a snappy title for our paper?" "I sure did!" "Great! And you checked to make sure it hasn't been used before?" "Oh, this is guaranteed fresh."

Screenshot of paper titles from a Google Scholar search:
From lab to life: Evaluating the reliability and validity of psychophysiological data from wearable devices in laboratory and ambulatory settings.
From Lab to Life: Bridging Gaps in Motion Capture to Increase Public Usability through Integrated Hardware and Software Solutions.
From Lab to Life: Phoneme Recognition Systems Making a Difference in Assistive Technologies.
From lab to life: Decoding the predictive power of present bias on adolescent self-control behaviors.
From lab to life: how wearable devices can improve health equity.
From Lab to Life: Challenges and Perspectives of fNIRS for Hemodynamic Neurofeedback in Real-World Environments.
From lab to life: Fine-grained behavior monitoring in the elderly's home.
Serial versus concurrent multitasking: From lab to life.
Translations in gerontology: From lab to life: Psychophysiology and speed of response.

020

David Epstein @davidnotdave.bsky.social · 16/03/2024

This thing is making a reappearance on Twitter, and no one appears to know that the answer is zero. Good luck claiming your earned-income tax credit on those gains, cowboy.

I bought a cow for $800. I sold it for $1000. I bought it again for $1100.1
sold it again for $1300. How much did I earn?

020

David Epstein @davidnotdave.bsky.social · 08/03/2024

You could say that scientists have a compulsion to fold things, but that would be unfair. Scientists have a compulsion to fold things a specified number of times. (Screenshot: quasirandom selection from papers I've recently read.)

text excerpts:

"The importance of the results from Study I are threefold."

"The argument against class criterion-oriented psychometrics is thus two-fold."

"the number of pharmacies distributing buprenorphine was 43.1-fold greater than OTPs distributing methadone in 2010. However, this ratio decreased to 33.3-fold in 2021."

"MOR agonist that is 100-fold more potent than morphine"

"Aims were threefold. In study one..."

"The aim of the simulation study was threefold."

"two-fold higher than crystal MDMA"

110

David Epstein @davidnotdave.bsky.social · 19/02/2024

"Orange you glad we didn't maintain a data column labeled 'methamphetamine'?"

Screenshot from a published protocol called "Personalized Deep Learning for Substance Use in Hawaii: Protocol for a Passive Sensing and EMA Study."

Excerpt: "As a precaution, the interface on the app will not be labeled as 'substance consumption' and 'substance use' but rather as 'banana consumption' and 'banana use.' Furthermore, our data will be stored and labeled as 'fruit' rather than 'substance use.'"

020

David Epstein @davidnotdave.bsky.social · 16/02/2024

And furthermore, I might be willing to die on the same hill as these fellas...

John Hoenig and Dennis Heisey: "We feel that researchers often do not need a rigorous understanding of confidence intervals to use them to good advantage. Although we cannot demonstrate it formally, we suspect that imperfectly understood confidence intervals are more useful and less dangerous than imperfectly understood p values and hypothesis tests. For example, it is surely prevalent that researchers interpret confidence intervals as if they were Bayesian credibility regions; to what extent does this lead to serious practical problems? .... If informally motivated confidence intervals lead to better science than rigorously motivated hypothesis testing, then perhaps the rigor normally presented to students destined to be applied researchers can be sacrificed."

000

David Epstein @davidnotdave.bsky.social · 16/02/2024

Ah, but that brings us to whether a CI is a cat's-eye or a cliff. I'm a cat's-eye guy myself.

100

David Epstein @davidnotdave.bsky.social · 15/02/2024

I kinda startled myself with this little experiment in power calculations. You can have zero ability to find the effect you're looking for, but more than zero "power," all of which reflects your likelihood of overestimating the effect.

If your sample size is 50, then the smallest correlation that will be statistically significant (at .05, two-tailed) is .28.

An observed r of .27 in that sample cannot be statistically significant (at .05, two-tailed).

And yet…

A power analysis would tell you that you have 48% power to detect that correlation with n = 50.

The power analysis actually means you have a 48% probability of getting a p value <.05 with n = 50. When that happens, it can only mean that the correlation in your sample is spuriously large: bigger than .27.

It's impossible, with n=50, to detect the correlation at its true size (r=.27) and have it be statistically significant.

Presumably, this problem is less extreme at conventional power levels (80% or more), but the underlying fact seems to remain.

251

David Epstein @davidnotdave.bsky.social · 05/12/2023

I really liked the discussions of equity, privacy, and consent in this big-data study group at #ACNP2023.

The Science and Ethics of Measuring and Modeling Individual and Group Behavior.

Cheryl Corcoran, Chair. Holly Moore, Co-Chair. Elizabeth Stafford, Moderator. Laura Cabrera, Justin Baker, Tingting Liu, Satrajit Ghosh, and Sara Berger, Participants.

020

David Epstein @davidnotdave.bsky.social · 21/10/2023

This is why I'm glad there's a section for confidential comments to the editor.

I don’t usually say “reject.” I usually say, “Just present the findings more tentatively.” And I’m sure the authors will want to do so. They’ll say something like, “Let’s publish this as a preliminary finding. It will inspire future research to resolve the ambiguities.” But it’s not a case of “no harm done.” People love genetics and they love to overread it. There will be a press release and a tweet and a TikTok dance, all trumpeting how genes determine...

020