Sign in

Amit Goldenberg

@amit-goldenberg.bsky.social
3.8K followers 1.3K following 65 posts

I study emotions, collectives, technology and their interaction. Harvard Business School | Harvard Psych (affiliated) | D3

PostsRepliesMedia
Amit Goldenberg @amit-goldenberg.bsky.social · 04/09/2026
Any modification would be a good start. More broadly, almost any definition has a threshold that is evaluated along a continuum. Avoiding all of them means defining nothing.
000
Amit Goldenberg @amit-goldenberg.bsky.social · 01/09/2026
Thanks for reading the paper and for reacting. I agree that setting a criterion for emotions is challenging, but I think that in this case it's doable. Is the model modifying its speed of processing in response to a stimulus? Its energy consumption? If so, I'll be happy to say it has emotions.
100
Amit Goldenberg @amit-goldenberg.bsky.social · 24/08/2026
LLMs don't have emotions. New correspondence in Nature Human Behavior link - rdcu.be/fBIbp
0142
Amit Goldenberg @amit-goldenberg.bsky.social · 15/05/2026
Yup
010
Amit Goldenberg @amit-goldenberg.bsky.social · 14/05/2026
One of the biggest insights from my research over the past few years: Groups (not just the individuals within them, but groups as distributed entities) do a remarkable range of things to manage emotion. See new paper in American Psychologist with Yajun Cao psycnet.apa.org/doiLanding?d...
psycnet.apa.org
APA PsycNet
271
Amit Goldenberg @amit-goldenberg.bsky.social · 16/04/2026
New paper out in @nathumbehav.nature.com led by Kyle Lafollette & Doroteja Rubez. Challenging the mechanism for the implicit association test (IAT) rdcu.be/fdKtb
rdcu.be
Challenging the mechanism for the implicit association test
Nature Human Behaviour - Using racing diffusion models to analyse 115,601 implicit association tests across 39 topics, LaFollette et al. found that response caution explained more variance in...
070
Amit Goldenberg @amit-goldenberg.bsky.social · 06/08/2025
We scrolls about 100 meters a day of social media content, much of it highly emotional. In a new paper led by Jonas Schöne (JPSP) we show that people tend to remember stronger emotional expressions, which leads them to estimate feeds as more emotional. psycnet.apa.org/record/2026-...
psycnet.apa.org
APA PsycNet
0123
Amit Goldenberg @amit-goldenberg.bsky.social · 01/08/2025
🔖 NEW PAPER ALERT 🔖 Out in PNAS – we used equation discovery algorithms to improve models of human reinforcement learning. Led by the wonderful Kyle Lafollette with an amazing team Yanni Yuval @roeyschurr.bsky.social and @d-melnikoff.bsky.social www.pnas.org/doi/epdf/10....
pnas.org
PNAS
Proceedings of the National Academy of Sciences (PNAS), a peer reviewed journal of the National Academy of Sciences (NAS) - an authoritative source of high-impact, original research that broadly spans...
04714
Amit Goldenberg @amit-goldenberg.bsky.social · 01/07/2025
New paper in Nature Human Behavior (led by Matan Rubin), where we manipulated the perceived identity of a real-time empathic response to be or AI. We observed a drop in perceived empathy when label was AI, and this effect was mainly driven by emotion. rdcu.be/et9ad
rdcu.be
Comparing the value of perceived human versus AI-generated empathy
Nature Human Behaviour - AI mimics empathy well, but does its empathy feel different? The authors show human empathy has unique value: human-attributed responses are rated as more empathic than...
0133
Reposted by Amit Goldenberg
Jay Van Bavel, PhD @jayvanbavel.bsky.social · 03/06/2025
The combination of #artificialintelligence and #socialmedia poses a threat to democracy. Our new paper explains how AI swarms can fabricate grassroots consensus, fragment shared reality, engage in mass harassment, interfer with elections, and erode institutional trust: osf.io/preprints/os...
910959
Amit Goldenberg @amit-goldenberg.bsky.social · 01/04/2025
Totally agree. So much information.
030
Amit Goldenberg @amit-goldenberg.bsky.social · 01/04/2025
So to conclude, we can definitely predict something from the way people walk. Arousal (calm to excited)? Much more likely! Valence – I would need better proof. But this is a good start. arxiv.org/pdf/2503.04190
arxiv.org
010
Amit Goldenberg @amit-goldenberg.bsky.social · 01/04/2025
The mean absolute error of the model after training was 1.11, which means the model was off by about one point on average. This is not bad, or even surprisingly good, especially given that valence results were restricted to a pretty narrow range, but can we predict valence? Not sure.
110
Amit Goldenberg @amit-goldenberg.bsky.social · 01/04/2025
As for the results – it's unclear whether the music impacted emotions in the expected direction (not reported in the paper). One thing is clear: the valence scale – which ranged from very negative -1 to very positive 9 – had valence results mostly between 3 and 9, so no strong negative valence.
100
Amit Goldenberg @amit-goldenberg.bsky.social · 01/04/2025
You can already imagine one major issue with the design, which is that different types of music may lead people to walk differently regardless of their emotional state. So are we predicting what music people heard? Or what they felt?
100
Amit Goldenberg @amit-goldenberg.bsky.social · 01/04/2025
The paper asked 20 people to walk while listening to music that was previously validated to elicit different emotions. For example, to elicit high-arousal negative emotions (akin to anger or fear), participants heard music from Dracula or high-wire stunts and lighting was red/yellow shining.
100
Amit Goldenberg @amit-goldenberg.bsky.social · 01/04/2025
Can you predict whether people are happy or sad by the way they walk? If we could, that would be pretty fun. Anyway, this paper claims you can predict both valence (positive negative) and arousal (calm and excited) from footsteps. But, can you? See thread🧵 arxiv.org/pdf/2503.04190
arxiv.org
220
Reposted by Amit Goldenberg
Deborah Shulman @debshulman.bsky.social · 28/03/2025
Can learning from past peace processes help promote peace in ongoing conflict? Check out our new paper (with Michal Reifen-Tagar, Noa Omri, and Eran Halperin @picr-lab.bsky.social ) journals.sagepub.com/doi/abs/10.1... 1/
171
Amit Goldenberg @amit-goldenberg.bsky.social · 28/03/2025
wow! Thanks for sharing!!
000
Amit Goldenberg @amit-goldenberg.bsky.social · 24/03/2025
These results are consistent with other companion studies that suggest the story of using AI as a companion is a story of access—people who don't have access to human friends are more likely to use AI as a substitute.
140
Amit Goldenberg @amit-goldenberg.bsky.social · 24/03/2025
What was more interesting to me is the correlation between use time and loneliness. Higher use time was positively associated with loneliness. This is very likely driven by the fact that lonelier people are just more likely to spend more time with the chat.
1114
Amit Goldenberg @amit-goldenberg.bsky.social · 24/03/2025
Having worked on these topics for years, I also find it highly unlikely that spending 5.32 minutes on average per day talking to ChatGPT would lead to a reduction in loneliness. Loneliness is really hard to change.
110
Amit Goldenberg @amit-goldenberg.bsky.social · 24/03/2025
Across all conditions (but with no control group that didn’t use ChatGPT), the authors report a reduction in loneliness over time. This, as the authors admit, basically means nothing because there was no control condition and this could just be the effect of time.
100
Amit Goldenberg @amit-goldenberg.bsky.social · 24/03/2025
The RCT (where people were asked to use ChatGPT for 5 minutes a day, at least 4 days a week, for a total of 4 weeks) is disappointing. First, there was no control group, but rather a SUPER complicated (3x3) design of different types of use.
100
Amit Goldenberg @amit-goldenberg.bsky.social · 24/03/2025
Does daily use of ChatGPT reduce loneliness? I was very excited to see that OpenAI just published its first paper on GPT use and well-being with an experimental design. The outcome is quite disappointing . 🧵 www.media.mit.edu/publications...
183
Reposted by Amit Goldenberg
Michelle Greene @mgreenephd.bsky.social · 21/03/2025
🚨New job alert! My lab is hiring a lab manager. Perfect position for a graduating senior who is interested in full-time research before applying to graduate school. #Neuroskyence #PsychSciSky #VisionScience #MLSky Apply here: www.higheredjobs.com/details.cfm?...
higheredjobs.com
Full-time Research Assistant / Lab Manager - HigherEdJobs
Jobs in higher education. Faculty and administrative positions at colleges and universities. Updated daily. Free to job seekers.
13323
Amit Goldenberg @amit-goldenberg.bsky.social · 06/02/2025
Can emotion regulation interventions spread from treated to non-treated people in a group? We examine this question in a paper published in @naturecomms.bsky.social, led by Michael Pinus and Yajun Cao. www.nature.com/articles/s41...
nature.com
Emotion regulation contagion drives reduction in negative intergroup emotions - Nature Communications
Here the authors show that an emotion regulation intervention for reducing negative intergroup emotions in conflicts can decrease negative emotions not only in treated participants but also in nontrea...
1104
Amit Goldenberg @amit-goldenberg.bsky.social · 20/01/2025
The best examples of emotional contagion is infants starting to cry when they hear another person crying (knowing nothing about the context). But could it simply be an aversion to noise? And is contagion stronger for neg emotions than for pos emotions? onlinelibrary.wiley.com/doi/10.1111/...
onlinelibrary.wiley.com
Contagious Crying Revisited: A Cross‐Cultural Investigation Into Infant Emotion Contagion Using Infrared Thermal Imaging
Contagious crying in infants has been considered an early marker of their sensitivity to others’ emotions, a form of emotional contagion, and an early basis for empathy. However, it remains unclear w...
1134
Amit Goldenberg @amit-goldenberg.bsky.social · 10/01/2025
Very cool paper on using ML to predict habit formation.
040
Amit Goldenberg @amit-goldenberg.bsky.social · 09/01/2025
My lab and I are looking to take an applied class on Graph Machine Learning, similar to this Stanford class: web.stanford.edu/class/cs224w/. Any suggestions?
042
Amit Goldenberg @amit-goldenberg.bsky.social · 05/01/2025
People's misrepresentation of collective political beliefs as more extreme than they actually are is an issue associated with various polarization-related outcomes (including acrophily). It seems that disclosing this bias doesn't fully solve the problem. www.sciencedirect.com/science/arti...
sciencedirect.com
Disclosing sample bias fails to fully correct judgments of partisan extremity
How do we infer the beliefs of an entire group (e.g., Democrats) after being exposed to the beliefs of only a handful of group members? What if we kno…
060
Amit Goldenberg @amit-goldenberg.bsky.social · 18/12/2024
Come join our multi-country study on social preferences for political extremes!!
050
Amit Goldenberg @amit-goldenberg.bsky.social · 12/12/2024
If you are a 4 year old that loves salad.
000
Amit Goldenberg @amit-goldenberg.bsky.social · 12/12/2024
When asked to describe street food in the US and why it's so bad compared to other countries, I like to point out that I live in a place where Goldfish crackers are considered a legitimate salad topping.
1201
Amit Goldenberg @amit-goldenberg.bsky.social · 05/12/2024
oh wow. very interesting.
010
Amit Goldenberg @amit-goldenberg.bsky.social · 05/12/2024
Now, here is a caveat: this paper came out in 1968, during a time of significant polarization in the U.S. due to the Vietnam War and the Civil Rights Movement. It's possible that attraction to extremes is especially salient during such tumultuous times.
130
Amit Goldenberg @amit-goldenberg.bsky.social · 05/12/2024
When we present our work on attraction to extremity, people often ask whether this is a new phenomenon. We found this great paper from 1968 showing that people assign more sincerity, competence, trustworthiness, and likability to extremes on their own side. www.sciencedirect.com/science/arti...
1173
Amit Goldenberg @amit-goldenberg.bsky.social · 27/11/2024
Very cool!
120
Reposted by Amit Goldenberg
Uri Hertz @urihertz.bsky.social · 26/11/2024
New paper! Led by Elena Kozakevitch-Arbel, with Simone Shamay-Tsoory, we used multi-dimensional reinforcement learning approach to examine whether people are sensitive to different dimensions of social scenarios during empathic interactions www.nature.com/articles/s44... 🧵
nature.com
Adaptive empathic response selection is sensitive to multiple dimensions of social interaction - Communications Psychology
When providing emotional support and deciding on an empathic reaction, responders were sensitive to changes in the person requiring empathy, the emotional state of that person, and the cause of their ...
13210
Amit Goldenberg @amit-goldenberg.bsky.social · 21/04/2024
Overall, these results begin to help us understand the nature of emotional support by LLMs and how it compares to humans. Very excited by future work on this.
000
Amit Goldenberg @amit-goldenberg.bsky.social · 21/04/2024
Here's the kicker: GPT's reappraisal quality was higher when it was semantically closer to the vignette. For humans it was the opposite—they were rated higher when they were able to generalize away from the vignette!! This reveals an important difference between GPT and humans.⬇️
120
Amit Goldenberg @amit-goldenberg.bsky.social · 21/04/2024
Using sentence embedding, we compared the distance between the reappraisals and vignettes. Human reappraisals (green) tended to be much more similar to the vignettes than those of GPT-4 (orange), which tended to be more general. ⬇️
100
Amit Goldenberg @amit-goldenberg.bsky.social · 21/04/2024
We were wondering whether the difference was driven by effort or skill, so we incentivized humans to write better reappraisals (up to 150% of their base pay if ranked in the top 25% of reappraisers). This led to an increase in time spent (15-22 seconds) but no change in quality. ⬇️
100
Amit Goldenberg @amit-goldenberg.bsky.social · 21/04/2024
Human raters evaluated reappraisals on four dimensions: effectiveness, empathy, novelty, and specificity. GPT-4 outperformed humans on all dimensions except for specificity ("This rethinking is specific to the following scenario"). GPT-4 ranked in the 85th percentile in quality. ⬇️
100
Amit Goldenberg @amit-goldenberg.bsky.social · 21/04/2024
We created 18 vignettes expressing negative emotions with a similar structure: "My friend forgot my birthday after saying we'd go to dinner together. I feel unwanted." We similarly trained both humans and GPT-4 on how to produce reappraisals (N human reappraisals = 4195) ⬇️
100
Amit Goldenberg @amit-goldenberg.bsky.social · 21/04/2024
First paper from our lab on LLMs and emotional support (led by @joannazli): Skill but not effort drive GPT overperformance over humans in cognitive reframing of negative scenarios. Short thread of the results 1/7   osf.io/preprints/ps... #Psychology #AcademicSky #SocialPsyc
osf.io
OSF
1143
Reposted by Amit Goldenberg
Mark Rubin @markrubin.bsky.social · 12/04/2024
“Should I be a caring man or a dominant man?” Postdoc at the University of Geneva with Vincenzo Iacoviello and Clara Kulich #SocialPsyc #Psychology #PhDSky #PsychJobs
easp.eu
Post-doc position in Social Psychology, University of Geneva
Deadline: 30 April, 2024 • At the University of Geneva, we have an open position for a Post-Doctoral Research Fellow. The successful candidate is expected to conduct their research within the social ...
164
Amit Goldenberg @amit-goldenberg.bsky.social · 15/02/2024
🚨new paper🚨Stereotypically, women are seen as more emotional than men. How is this translated into judgments of emotionality? In a paper by Megan Gorges we show that stereotypes turn into judgements when people aggregate multiple emotional expressions. #psychsciski #socialpsyc osf.io/preprints/os...
042
Reposted by Amit Goldenberg
April Bailey @ahbailey.bsky.social · 29/01/2024
New work in press at PSPB! With Aashna Poddar, Adina Williams, and Andrei Cimpian We found intersectional male-centric and White-centric biases based on a list of 4,000+ first names in large-scale (English) text using word embeddings Results overview ⬇️ osf.io/preprints/os...
1188
Reposted by Amit Goldenberg
Jay Van Bavel, PhD @jayvanbavel.bsky.social · 30/01/2024
Our paper explains how social identity shapes information processing and contributes to the belief and spread of #misinformation We argue that understanding the interplay between social identity and accuracy is crucial in addressing misinformation www.sciencedirect.com/science/arti...
24728