Emory Richardson
LightNews LightNews
banner
emoryrchrdsn.bsky.social
Emory Richardson
@emoryrchrdsn.bsky.social
1.6K followers 3.3K following 160 posts
cognitive scientist.

intuitive theories, collaboration, cumulative culture, networks, philosophy of science/mind/bio, language. Also kettlebells.

past: @UMich @Yale @UChicago @stjohnscollege.

https://rchrdsnemory.github.io/site/
Posts Replies Media Videos
Pinned
emoryrchrdsn.bsky.social
Emory Richardson @emoryrchrdsn.bsky.social · Aug 4
New blog post inspired by my 5yo's most recent library book: asking an LLM for help is kind of like asking Amelia Bedelia for help!

AKA, why the range of tasks where using AI is worth your time is probably a lot narrower than you think.

rchrdsnemory.github.io/site/blog/20...
Reposted by Emory Richardson
usrbinr.bsky.social
so what is the best way to share our shiny apps? @posit.co connect? or shinapps.io?

#rstats
January 26, 2026 at 1:28 AM
Reposted by Emory Richardson
annoyingreposter.bsky.social
A fascinating (for me) paper about BNE in multi-agent systems:
arxiv.org/abs/2506.08292
From Debate to Equilibrium: Belief-Driven Multi-Agent LLM Reasoning via Bayesian Nash Equilibrium
Multi-agent frameworks can substantially boost the reasoning power of large language models (LLMs), but they typically incur heavy computational costs and lack convergence guarantees. To overcome thes...
arxiv.org
January 26, 2026 at 8:22 AM
Reposted by Emory Richardson
markriedl.bsky.social
Musk’s ability to alter the worldview of people now expands beyond just users of Grok.
brucedlittle.bsky.social Bruce Little @brucedlittle.bsky.social · 3d
Time to dump ChatGPT!

The latest model of ChatGPT has begun to cite Elon Musk’s Grokipedia as a source on a wide range of queries, including on Iranian conglomerates and Holocaust deniers, raising concerns about misinformation on the platform.
Latest ChatGPT model uses Elon Musk’s Grokipedia as source, tests reveal
Guardian found OpenAI’s platform cited Grokipedia on topics including Iran and Holocaust deniers
www.theguardian.com
January 25, 2026 at 3:21 PM
Reposted by Emory Richardson
jdlong.cerebralmastication.com
There ya go. That’s the message.
January 25, 2026 at 1:35 AM
Reposted by Emory Richardson
perfors.net
Sending this around for people in different time zones. I'm keen for it to get as much reach as possible because I think there's a great need for something like this out there! See thread below it for details, and please share widely! :)
perfors.net Andrew Perfors @perfors.net · 6d
I just created a series of seven deep-dive videos about AI, which I've posted to youtube and now here. 😊

Targeted to laypeople, they explore how LLMs work, what they can do, and what impacts they have on learning, well-being, disinformation, the workplace, the economy, and the environment.
Part 1: How do LLMs work?
YouTube video by Andrew Perfors
www.youtube.com
January 22, 2026 at 8:51 PM
Reposted by Emory Richardson
mattansb.msbstats.info
Finished teaching my new Advance Stats for Psych graduate course today with a heavy emphasis on both DAGs and shifting away from coefficient interpretation and towards models as prediction machines.

Both went great!

The latter was extremely helpful for logistic regression (for obvious reasons 😵‍💫)!
dingdingpeng.the100.ci Julia M. Rohrer @dingdingpeng.the100.ci · Aug 25
Ever stared at a table of regression coefficients & wondered what you're doing with your life?

Very excited to share this gentle introduction to another way of making sense of statistical models (w @vincentab.bsky.social)
Preprint: doi.org/10.31234/osf...
Website: j-rohrer.github.io/marginal-psy...
Models as Prediction Machines: How to Convert Confusing Coefficients into Clear Quantities Abstract Psychological researchers usually make sense of regression models by interpreting coefficient estimates directly. This works well enough for simple linear models, but is more challenging for more complex models with, for example, categorical variables, interactions, non-linearities, and hierarchical structures. Here, we introduce an alternative approach to making sense of statistical models. The central idea is to abstract away from the mechanics of estimation, and to treat models as “counterfactual prediction machines,” which are subsequently queried to estimate quantities and conduct tests that matter substantively. This workflow is model-agnostic; it can be applied in a consistent fashion to draw causal or descriptive inference from a wide range of models. We illustrate how to implement this workflow with the marginaleffects package, which supports over 100 different classes of models in R and Python, and present two worked examples. These examples show how the workflow can be applied across designs (e.g., observational study, randomized experiment) to answer different research questions (e.g., associations, causal effects, effect heterogeneity) while facing various challenges (e.g., controlling for confounders in a flexible manner, modelling ordinal outcomes, and interpreting non-linear models). Figure illustrating model predictions. On the X-axis the predictor, annual gross income in Euro. On the Y-axis the outcome, predicted life satisfaction. A solid line marks the curve of predictions on which individual data points are marked as model-implied outcomes at incomes of interest. Comparing two such predictions gives us a comparison. We can also fit a tangent to the line of predictions, which illustrates the slope at any given point of the curve. A figure illustrating various ways to include age as a predictor in a model. On the x-axis age (predictor), on the y-axis the outcome (model-implied importance of friends, including confidence intervals). Illustrated are 1. age as a categorical predictor, resultings in the predictions bouncing around a lot with wide confidence intervals 2. age as a linear predictor, which forces a straight line through the data points that has a very tight confidence band and 3. age splines, which lies somewhere in between as it smoothly follows the data but has more uncertainty than the straight line.
January 19, 2026 at 6:28 PM
Reposted by Emory Richardson
dingdingpeng.the100.ci
Ever stared at a table of regression coefficients & wondered what you're doing with your life?

Very excited to share this gentle introduction to another way of making sense of statistical models (w @vincentab.bsky.social)
Preprint: doi.org/10.31234/osf...
Website: j-rohrer.github.io/marginal-psy...
Models as Prediction Machines: How to Convert Confusing Coefficients into Clear Quantities Abstract Psychological researchers usually make sense of regression models by interpreting coefficient estimates directly. This works well enough for simple linear models, but is more challenging for more complex models with, for example, categorical variables, interactions, non-linearities, and hierarchical structures. Here, we introduce an alternative approach to making sense of statistical models. The central idea is to abstract away from the mechanics of estimation, and to treat models as “counterfactual prediction machines,” which are subsequently queried to estimate quantities and conduct tests that matter substantively. This workflow is model-agnostic; it can be applied in a consistent fashion to draw causal or descriptive inference from a wide range of models. We illustrate how to implement this workflow with the marginaleffects package, which supports over 100 different classes of models in R and Python, and present two worked examples. These examples show how the workflow can be applied across designs (e.g., observational study, randomized experiment) to answer different research questions (e.g., associations, causal effects, effect heterogeneity) while facing various challenges (e.g., controlling for confounders in a flexible manner, modelling ordinal outcomes, and interpreting non-linear models). Figure illustrating model predictions. On the X-axis the predictor, annual gross income in Euro. On the Y-axis the outcome, predicted life satisfaction. A solid line marks the curve of predictions on which individual data points are marked as model-implied outcomes at incomes of interest. Comparing two such predictions gives us a comparison. We can also fit a tangent to the line of predictions, which illustrates the slope at any given point of the curve. A figure illustrating various ways to include age as a predictor in a model. On the x-axis age (predictor), on the y-axis the outcome (model-implied importance of friends, including confidence intervals). Illustrated are 1. age as a categorical predictor, resultings in the predictions bouncing around a lot with wide confidence intervals 2. age as a linear predictor, which forces a straight line through the data points that has a very tight confidence band and 3. age splines, which lies somewhere in between as it smoothly follows the data but has more uncertainty than the straight line.
August 25, 2025 at 11:49 AM
Reposted by Emory Richardson
kitwhitfield.bsky.social
Had an intriguing editorial discussion recently: how did people in the past talk about 'minutes' when they didn't have watches or standardised times? How does that affect your thinking?

Come down an Elizabethan/Jacobean rabbit hole with me.

1/
January 17, 2026 at 10:34 AM
Reposted by Emory Richardson
jamellebouie.net
over the course of 1871, congress held seven months of hearings on ku klux klan and other white vigilante violence in the south, they took detailed testimony from hundreds of black men and women attesting to klan terror. (1/?)
January 14, 2026 at 2:06 AM
Reposted by Emory Richardson
adamjkucharski.bsky.social
Want to get the data out of a PDF figure? As in, the actual data – not a rough trace-along-the-lines version?

I made an app you might like: adamkucharski.github.io/pdf2plot/

It all started a few years ago... 🧵
January 13, 2026 at 9:12 PM
Reposted by Emory Richardson
tomerullman.bsky.social
a project I really like, now officially out!

"Shape Guides Visual Pretense"

by Qian and me

paper link: direct.mit.edu/opmi/article...

I'll walk through a quick version here

To get a sense of it, first consider:

Would it make more sense to pretend that this block is a car, or a strawberry?
January 6, 2026 at 2:34 PM
Reposted by Emory Richardson
libbyheeren.bsky.social
The very first DS Lab ep on fun #Positron settings is on YouTube! Come 🤣 at how I can't figure out "rainbow parens" are "bracket pair colorization" in VSC, despite @rorylawless.com pinging me with the answer in the chat 😂 OH AND SO MANY RESOURCE LINKS!

#rstats #databs #python

youtu.be/QIYyeuZ_ISY
Exploring Positron settings | Isabel Zimmerman & Davis Vaughan | Data Science Lab
YouTube video by Posit PBC
youtu.be
January 7, 2026 at 2:51 AM
Reposted by Emory Richardson
gabrielwaterhouse.bsky.social
I am very excited to announce that over the holidays, my first ever paper (w/ @samiyousif.bsky.social) was published in Cognitive Science! Here, we describe a new illusion of *number*: The Crowd Size Illusion!

onlinelibrary.wiley.com/doi/10.1111/...
January 5, 2026 at 5:04 PM
Reposted by Emory Richardson
tadegquillien.bsky.social
A fascinating new paper by Amanda Royka and colleagues explores why monkeys fail false belief tasks.

A natural explanation would be that monkeys wrongly assume that other agents share their own knowledge.

Royka et al. find that this is NOT the case...
www.sciencedirect.com/science/arti...
Exploring the evolutionary roots of theory of mind: Primate errors on false belief tasks reveal representational limits
Human adults flexibly reason about others' unobservable mental states, a capacity known as Theory of Mind (ToM). Unfortunately, the roots of this capa…
www.sciencedirect.com
January 2, 2026 at 5:21 PM
Reposted by Emory Richardson
paulgp.com
Yo what the fuck short form videos are bad

Short-Form Videos Degrade Our Capacity to Retain Intentions

arxiv.org/pdf/2302.03714

Ht: HowTown
December 31, 2025 at 2:40 AM
emoryrchrdsn.bsky.social
This guy’s not on bsky but the experiments are really interesting, and fit my sense that LLMs will reliably give good answers about things that only highly specialized nerds have ever written anything about, and do better the larger that community of nerds is and the more unique their jargon is. BUT
December 12, 2025 at 4:17 AM
Reposted by Emory Richardson
neuranna.bsky.social
The last chapter of my PhD (expanded) is finally out as a preprint!

“Semantic reasoning takes place largely outside the language network” 🧠🧐

www.biorxiv.org/content/10.6...

What is semantic reasoning? Read on! 🧵👇
Semantic reasoning takes place largely outside the language network
The brain's language network is often implicated in the representation and manipulation of abstract semantic knowledge. However, this view is inconsistent with a large body of evidence suggesting that...
www.biorxiv.org
December 11, 2025 at 6:34 PM
Reposted by Emory Richardson
youcaughtscott.com
I’M LOSING MY SHIT THIS IS SO FUNNY
December 11, 2025 at 3:13 PM
Reposted by Emory Richardson
emilymbender.bsky.social
Had a really disconcerting interaction with what I'm pretty sure was an artificial customer service agent today.

Short thread >>
December 9, 2025 at 12:55 AM
emoryrchrdsn.bsky.social
It looks like some of the AI people (Ilya Sutskever) have heard Alison Gopnik pointing out that humans aren’t actually an AGI either, but this is the passage I think of every time I hear them say that.
December 4, 2025 at 4:39 PM
Reposted by Emory Richardson
stewartcoles.bsky.social
🚨New pub alert!🚨 Now available open-access in @journal-of-comm.bsky.social, we (w/ @dkreiss.bsky.social, @danlane.bsky.social, & @shannimcg.bsky.social) critique political communication's "Identity Turn" and offer instead a foundation for studying #polcomm from a *group* perspective. 🧵
Abstract: Recent political communication scholarship finds that groups and identities play a central role in the crises faced by political and media systems globally, particularly in democracies. Yet an individualist orientation in the literature has resulted in key theoretical and conceptual limitations, preventing a broader group-centric theoretical framework from emerging. We synthesize disparate bodies of theory on groups, politics, and communication to offer three basic propositions underlying a group theory of political communication. First, it is the group—not the individual—that is the fundamental organizing unit of social and political life. Second, groups are constituted through communication, which is central to how they define their politics. Third, groups and politics are reciprocally influencing forces through political communication, oriented around power. We offer a framework for studying the role of groups in political communication at the micro, meso, and macro levels, providing a concrete agenda for the study of groups in political communication.
December 3, 2025 at 6:04 PM
Reposted by Emory Richardson
brunelcce.bsky.social
1/ 🎙️ New paper by Dr Aiyana Willard argues it’s time to retire HADD = “we’re hardwired to see agents EVERYWHERE” and replace it with something that *actually* fits the evidence 🪦➡️📚

@brunelpsy.bsky.social

🔗 Read it 👇
Belief as explanation: a motivation-based theory of agency and anthropomorphism in religious belief
Supernatural agent beliefs are ubiquitous across cultures, yet many theories aimed at explaining this fact have not held up to scrutiny. The most famous of these, the Hyperactive Agency Detection D...
www.tandfonline.com
November 28, 2025 at 12:37 PM
Reposted by Emory Richardson
coltoncasto.bsky.social
What does it mean to understand language? We argue that the brain’s core language system is limited, and that *deeply* understanding language requires EXPORTING info to other brain regions.
w/ @neuranna.bsky.social @evfedorenko.bsky.social @nancykanwisher.bsky.social
arxiv.org/abs/2511.19757
1/n🧵👇
What does it mean to understand language?
Language understanding entails not just extracting the surface-level meaning of the linguistic input, but constructing rich mental models of the situation it describes. Here we propose that because pr...
arxiv.org
November 26, 2025 at 4:26 PM
Reposted by Emory Richardson
mivich.bsky.social
My new book, The Division of Rationalized Labor, is now shipping! A brief summary of the argument to follow…
Front cover of The Division of Rationalized Labor. The cover includes four pictures: pen and paper, microscope, factory tower, police badge. Modern-looking yellow lines and graphs are superimposed.
November 26, 2025 at 5:45 PM