iiinova — where your digital self keeps evolving

William J. Brady @williambrady.bsky.social · 04/09/2026

Registration is open for the 4th annual Computational Psychology preconference at @spspnews.bsky.social Annual conference in Philly 🎊 We are on for a full day on Thursday, Feb 11. 3 keynote themes + a debate that couldn't be more timely 🧵👇

1144

William J. Brady @williambrady.bsky.social · 29/06/2026

In this new paper w/ @joshcjackson.bsky.social , @nickornstein.bsky.social , @felix-chenwei.bsky.social , & Bolun Sun, we argue generative AI also distorts social learning, but because it optimizes for something different, it distorts in a different direction.

171

William J. Brady @williambrady.bsky.social · 27/05/2026

Result 2: Engagement-based feeds raised perceived partisan animosity (esp. after the election) AND reduced norm perception accuracy. Unexpected: for norm-perception people underestimated how acceptable others found toxic political content.

191

William J. Brady @williambrady.bsky.social · 27/05/2026

Another interesting tidbit: because we designed our own engagement algo, we could see separate influence of personalized vs community engagement signals. Turns out community signals are much more responsible for IME amplification (see SI)!

2102

William J. Brady @williambrady.bsky.social · 27/05/2026

Result 1: engagement-based feeds amplified IME & toxic content vs. a chronological feed. The biggest jumps were in moral outrage (up to +79%) and political content (up to +57%) after the election. Our diversified extremity algorithm reduced this exposure (by aiming to reduce extreme users)

1142

William J. Brady @williambrady.bsky.social · 27/05/2026

We also built & tested an intervention: a "diversified extremity" algorithm. Instead of amplifying outgroup posts (which can backfire), it reduces the outsized influence of extreme users—who post most of the toxic political content—to make feeds more socially representative.

1141

William J. Brady @williambrady.bsky.social · 27/05/2026

✨New paper out @nature.com ✨ For 8 weeks around the 2024 US election, we randomly assigned 2,000 people to use social media algos we built ourselves. Do engagement-based algorithms amplify intergroup, moral & emotional (IME) content—and does that distort how we see political norms? 🧵🔗 👇

513762

William J. Brady @williambrady.bsky.social · 28/02/2026

Also! Causal evidence from A/B testing dataset: victim framing causes more click throughs

020

William J. Brady @williambrady.bsky.social · 28/02/2026

@killianmcloughlin.bsky.social social media news especially likely to use villian / victim framing; low quality news especially; more likely to evoke outrage and draw engagement when doing so #spsp

1173

William J. Brady @williambrady.bsky.social · 26/02/2026

New work from @drsanaz.bsky.social : using smart phone sensing data, one of the biggest predictors of people higher on authoritarian measure was Facebook and social media use #spsp

1172

William J. Brady @williambrady.bsky.social · 24/02/2026

We're excited about the upcoming Computational Psychology preconference at @spspnews.bsky.social this Thursday. See our action-packed full day agenda below! Featuring 3 keynote talk themes with related early-career speakers, data blitz session, panel discussion. Don't miss it! #SPSP

1219

William J. Brady @williambrady.bsky.social · 11/11/2025

Key result #3: Motive inferences were malleable: we developed an intervention that corrected peoples’ motive inferences - increasing people’s inference of behavioral motive to out-partisan caused them to be more willing to have a political conversation even in context of outrage

110

William J. Brady @williambrady.bsky.social · 11/11/2025

Key result #1: People largely reported their in-partisans’ (and their own) motives for outrage was to raise awareness or inspire action (behavioral motive), but thought political opponents motives were to shame or troll (contra-hedonic motive), which was a vast overestimation.

120

William J. Brady @williambrady.bsky.social · 11/11/2025

In online experiments and a field study on Reddit, we asked users to report their motives for posting outrage and then had observers infer the motives.

110

William J. Brady @williambrady.bsky.social · 11/11/2025

✨New preprint! Why do people express outrage online? In 4 studies we develop a taxonomy of online outrage motives, test what motives people report, what they infer for in- vs. out-partisans, and how motive inferences shape downstream intergroup consequences. Led by @felix-chenwei.bsky.social 🧵👇

14018

William J. Brady @williambrady.bsky.social · 29/09/2025

The computational psych preconference is back @spspnews.bsky.social for a full day! This year's lineup: 👉theory-driven modeling: Hyowon Gweon 👉data-driven discovery: @clemensstachl.bsky.social 👉application: me 👉 panel: @steveread.bsky.social Sandra Matz, @markthornton.bsky.social Wil Cunningham

1248

William J. Brady @williambrady.bsky.social · 07/05/2025

@desmond-ong.bsky.social LLMs show context-insensitivity in empathetic responding which leads to comments rated as more inappropriate - can be harmful in therapeutic setting

081

William J. Brady @williambrady.bsky.social · 20/02/2025

Jonas Dalege: minority groups can only influence if they have strong connections and the majority group doesn't have a strong belief network #spsp2025 #comppsych

062

William J. Brady @williambrady.bsky.social · 20/02/2025

@cinoolee.bsky.social: compiled 1000+ sm posts about racial discrimination and matched set on non-race topic...both AI and humans are sig more likely to flag posts about race for removal (36% more likely!) Triggered by identity threat + tone-policing #spsp2025 #comppsych

0147

William J. Brady @williambrady.bsky.social · 20/02/2025

@gandalfnicolas.bsky.social: social groups are represented in LLMs in similar ways compared to human data, warmth and competence are key but only explain about 50% of variance (need more dimensions). #spsp2025 #comppsych

072

William J. Brady @williambrady.bsky.social · 20/02/2025

Sanaz Talaifar: smart phone passive sensing study: Liberals have less fun than conservatives (they are on social media lot...). Also, people vastly overestimate daily routine differences among lib vs conservatives #spsp2025 #comppsych

132

William J. Brady @williambrady.bsky.social · 20/02/2025

@curtispuryear.bsky.social a study of a decade of twitter posts shows and new measure of moralization shows that moralization has substantially increased over time (d = .45!). Driven by both self selection and within-user increase over time! #comppsych #spsp2025 (also stay tuned for this paper😎)

1198

William J. Brady @williambrady.bsky.social · 20/02/2025

Niyat Kachhiyapatel: LLM analysis of events in people's lives: people feel most uncertain during negative, emotional inducing events #comppsych #spsp2025

071

William J. Brady @williambrady.bsky.social · 20/02/2025

@madalina.bsky.social: sample of 63 countries test several interventions to increase climate policy support. Writing letter to future gen is best. Doom & gloom messaging great for increasing sharing but not for increasing support! #spsp2025 #comppsych

0193

William J. Brady @williambrady.bsky.social · 20/02/2025

Xavier Roberts-Gaal & @fierycushman.bsky.social: environmental variability predicts goal learning and stability predicts action learning in a really cool transmission experiment #spsp2025 #comppsych

0112

William J. Brady @williambrady.bsky.social · 20/02/2025

Gabriel Fajardo & @freemanjb.bsky.social have created a new connectionist model of trait formation (impression from faces). High dimension, dynamic models may buy us better model outcomes and situational variability #spsp2025 #comppsych

1146

William J. Brady @williambrady.bsky.social · 14/02/2025

Key result #3: A bridging algorithm that boosted bipartisan-liked content partially countered these misperceptions—yet it risked underrepresenting genuine disagreement. This reveals a complex tension between reducing polarization and maintaining accurate norm perception.

140

William J. Brady @williambrady.bsky.social · 14/02/2025

Key result #2: We also gauged participants perceptions of social norms after interacting with various feeds. Users in the engagement-based feds overestimated the descriptive and prescriptive norms for posting IME content ➡️ mediated greater posting intentions of blame content

130

William J. Brady @williambrady.bsky.social · 14/02/2025

Key result # 1: From natural user behavior as input, both passive and active engagement algorithms amplified IME posts disproportionately higher than their baserate in the network➡️ overrepresentation of IME content in feeds

130

William J. Brady @williambrady.bsky.social · 14/02/2025

Method: Across 3 experiments (N = 6,107) users scrolled through content in mock social media environment. We measured passive (dwell time) and active (like/share) engagement to train algos, and manipulated content based on categories drawn from social learning theory

120

William J. Brady @williambrady.bsky.social · 14/02/2025

👀New preprint! In 3 prereg experiments we study how engagement-based algorithms amplify ingroup, moral and emotional (IME) content in ways that disrupt social norm learning (and test one solution!) w/ @joshcjackson.bsky.social and my amazing lab managers @merielcd.bsky.social & Silvan Baier 🧵👇

35221

William J. Brady @williambrady.bsky.social · 05/02/2025

We think our model is an efficient way to explain public opinion in response to viral transgressions (see image below). Link to paper: osf.io/preprints/os...

231

William J. Brady @williambrady.bsky.social · 05/02/2025

✨New preprint✨How does repeated exposure to transgressions online shape moral judgment? Results show two competing processes: 💠We get desensitized = ⬇️ wrong 💠Transgression seems more infamous =⬆️wrong Relative strength of each may predict outrage to viral transgressions w/ @danieleffron.bsky.social

22916

William J. Brady @williambrady.bsky.social · 28/11/2024

Findings, cont: outrage-evoking misinfo seems to be driven by non-epistemic motives. It is shared more without reading, and when news articles evoked outrage in our experiments it caused people to share more misinfo, even tho it did not impact ppl’s truth discernment!

15310

William J. Brady @williambrady.bsky.social · 28/11/2024

Findings: misinfo consistently evokes more outrage than trustworthy news, and when it does it is shared more. This is specific to outrage - not just a general association with negative emotion in general

16812

William J. Brady @williambrady.bsky.social · 28/11/2024

Method: 1M facebook links + 34k users worth of Twitter data, 2016 and 2020 + behavioral experiment (N = 1450). Multiple classification schemes for misinfo, studies that control for users being in different networks

1464

William J. Brady @williambrady.bsky.social · 28/11/2024

New paper out in @ScienceMagazine! In 8 studies (multiple platforms, methods, time periods) we find: misinformation evokes more outrage than trustworthy news, when it does it's shared more + ppl are less likely to read before sharing. w/ @killianmcl1 @Klonick @mollycrockett 🧵👇

12535201149

William J. Brady @williambrady.bsky.social · 04/09/2024

Want to get paid ($40)💰 for browsing Bluesky as you normally would? Explore new feeds designed by a research team at Northwestern University and report back on your experiences. Help shape the future of social media ranking algorithms 👀 🔗 Sign Up: tinyurl.com/nufeedstudy 💌 FAQ: bit.ly/bluesky-faq

515150

William J. Brady @williambrady.bsky.social · 24/10/2023

I recently joined the Technology & Social Behavior program at Northwestern -an amazing cross-disciplinary group of computational social scientists! Interested in a PhD in this area? See below for an information session this Thursday registration: t.co/q0W8xKka0C

040