iiinova — where your digital self keeps evolving

Dirk Wulff @dirkwulff.bsky.social · 01/10/2026

📣 Delighted to share that I have joined WU (Vienna University of Economics and Business) as Professor of Artificial Intelligence and Behavioral Science. Announcements for at least one postdoc-level position will follow soon. Please reach out if you are interested.

68016

Dirk Wulff @dirkwulff.bsky.social · 30/09/2026

Women still exit psychological research at a disproportionate rate. Tracking 80k researchers, our new article maps the gender attrition gap and shows that differences in performance, networks, and institutions explain only part of it. psycnet.apa.org/doi/10.1037/... arxiv.org/abs/2510.13273

0187

Dirk Wulff @dirkwulff.bsky.social · 29/09/2026

🚨 New preprint Excited to share new work led by Raluca Rilla, showing how cheap, open agents threaten online data collection and how open-text analysis beyond conventional traps can help mitigate this threat. 🔗 arxiv.org/abs/2609.31054 With @Anne-Marie Nussberger and @ruimata.bsky.social .

044

Dirk Wulff @dirkwulff.bsky.social · 03/08/2026

🚨 Out now in TICS How can language models help cognitive science? @ruimata.bsky.social & I outline 5 uses: mapping research fields, formalizing theories, cleaning up constructs & measures, predicting behavior across tasks, and capturing environmental variation. 🔗 doi.org/10.1016/j.ti...

03812

Dirk Wulff @dirkwulff.bsky.social · 01/07/2026

🚨 Now out in PNAS In decision research, "talk is cheap." We show it isn't. Using LLMs to analyze participants' free-text explanations of their choices, we find verbal reports are a rich, scalable window into how people actually decide. 🔗 www.pnas.org/doi/10.1073/... Led by Kamil Fulawka

04215

Dirk Wulff @dirkwulff.bsky.social · 19/06/2026

Finally, we show that all of this means that the psychological profile of LLM responses can be steered through the design of the measurement instruments, with profiles of very capable models varying dramatically as a function of whether forward or reversed instruments were used.

110

Dirk Wulff @dirkwulff.bsky.social · 19/06/2026

Across 29 self-report and task instruments measuring personality and risk-taking, we show that reliability (internal consistency) is driven by response bias and therefore is fully dependent on the instruments' levels of response orthogonality.

100

Dirk Wulff @dirkwulff.bsky.social · 19/06/2026

We show that response bias in LLMs is partially but not fully reduced for larger and proprietary versus open models, with all models exceeding the level of response bias in humans.

100

Dirk Wulff @dirkwulff.bsky.social · 19/06/2026

For a big five inventory, we show that our sample of LLM models (N = 56) scatters along an axis indicative of response biases-driven responding, opposite to the pattern found in humans.

100

Dirk Wulff @dirkwulff.bsky.social · 19/06/2026

We base our analysis on a formal psychometric framework that unifies the idea of reverse-keying across self-report and task instruments under the notion of response orthogonality.

100

Dirk Wulff @dirkwulff.bsky.social · 19/06/2026

🚨 New preprint Do LLMs have stable dispositions measurable using instruments designed for humans? Led by @jelenameyer.bsky.social, our new preprint shows that LLM responses are driven by response bias, rendering their psychological profiles method artifacts. @dgarcia.eu @arc-mpib.bsky.social

1216

Dirk Wulff @dirkwulff.bsky.social · 10/06/2026

📣 Reporting checklist for LLMs in behavioral and social science New article presenting a consensus-based reporting checklist (GUIDE-LLM) for the use of LLMs in the behavioral and social sciences to foster transparency, reproducibility, and ethical use. 🔗 www.nature.com/articles/s41...

02110

Dirk Wulff @dirkwulff.bsky.social · 20/05/2026

🚨 New preprint 🚨 Excited to share a new preprint led by @zakashussain.bsky.social. We review how large language models compare to humans in behavior and representations. Our conclusion: LLM cognition is not simply human-like or non-human-like. It is jagged. 🔗 Preprint: osf.io/preprints/ps...

083

Dirk Wulff @dirkwulff.bsky.social · 13/05/2026

🚨 Additional tickets 🚨 📣 Behavioral Clones Workshop - Join us online! One-day workshop on behavioral clones — 🤖 AI systems that model human behavior — following the Machine+Behavior conference. 📅 May 20, 9 am - 6:15 pm (CET) 🔗 Register: www.eventbrite.ch/e/behavioral...

130

Dirk Wulff @dirkwulff.bsky.social · 11/05/2026

📣 Postdoc position Build 🤖 LLM-based behavioral clones for economic games together with Pınar Uğurlar and me. The position (up to 3 years) is based in Pınar's lab at Özyeğin in Istanbul, with regular visits to my lab. Please share / reach out if interested. 🔗 drive.google.com/file/d/1ftqi...

074

Dirk Wulff @dirkwulff.bsky.social · 05/05/2026

📣 Behavioral Clones Workshop - Join us online! A one-day workshop on behavioral clones — 🤖 AI systems that model human decision-making and social behavior — following the Machine Behavior conference. Join online via Zoom (no remote Q&A). Register at: www.eventbrite.ch/e/behavioral...

064

Dirk Wulff @dirkwulff.bsky.social · 22/04/2026

🚨 New preprint 🚨 Excited to share this work led by @kriegmair.bsky.social, establishing 🤖 machine individuality by decomposing 75M LLM ratings into shared variance, bias, idiosyncrasy, and noise, revealing coherent, cross-dimensional fingerprints unique to each model. 🔗 arxiv.org/pdf/2604.16755

051

Dirk Wulff @dirkwulff.bsky.social · 05/03/2026

🚨 Updated preprint 🚨 Excited to share this updated preprint, in which @ruimata.bsky.social and I discuss five ways LLMs can help address longstanding challenges in cognitive science and psychology, examining both opportunities and pitfalls. 🔗 arxiv.org/abs/2511.00206

1103

Dirk Wulff @dirkwulff.bsky.social · 06/02/2026

🚨 New publication 🚨 Excited to share this new forum piece led by @kevinetiede.bsky.social We propose broadening risk communication to cover neglected dimensions outlined in our risk-information taxonomy, requiring greater reliance on simulated experiences. Open access: doi.org/10.1016/j.ti...

084

Dirk Wulff @dirkwulff.bsky.social · 08/01/2026

We further demonstrate an important moderating role of embedding extraction method and confirm results suggesting that final-layer embeddings underrepresent psycholinguistic information.

000

Dirk Wulff @dirkwulff.bsky.social · 08/01/2026

We show that accessibility cascades from lexical to semantic information similarly across all transformer language models, but is still realized differently in encoder and decoder models.

100

Dirk Wulff @dirkwulff.bsky.social · 08/01/2026

🚨 New preprint 🚨 Excited to share new work led by @tikhomirova.bsky.social, presenting a large-scale evaluation of the accessibility of psycholinguistic information in transformer language models. Preprint: arxiv.org/abs/2601.03798

175

Dirk Wulff @dirkwulff.bsky.social · 19/12/2025

🚨 New preprint 🚨 We present the decisions-from-experience database (DfE-DB), including data from 168 studies. The data are currently shared with the original authors and made public upon publication. 🔗 Database: github.com/dwulff/dfe-db 🔗 Preprint: www.biorxiv.org/content/10.6...

0239

Dirk Wulff @dirkwulff.bsky.social · 10/12/2025

🔬 How has psychology contributed to achieving the 🇺🇳 SDGs? Our new preprint maps 230,000 psychology articles to the SDGs using our text2sdg R package (www.text2sdg.io) and analyzes historical trends in topics, national and gendered contributions, and citation patterns. 🔗 arxiv.org/abs/2512.08628

0117

Dirk Wulff @dirkwulff.bsky.social · 07/11/2025

How can LLMs advance the cognitive sciences? @ruimata.bsky.social and I put together a review on how LLMs can help us address five long-standing problems in cognitive science (e.g., disciplinary silos or lack of generalizability). 🔗 arxiv.org/2511.00206

0130

Dirk Wulff @dirkwulff.bsky.social · 04/11/2025

🚨 New preprint - Advancing Cognitive Science with LLMs Excited to share this new preprint together with Rui Mata. We review how 🤖 LLMs can help address long-standing problems in cognitive science, highlighting opportunities and pitfalls. 🔗 Preprint: arxiv.org/abs/2511.00206

0134

Dirk Wulff @dirkwulff.bsky.social · 29/10/2025

🚨 New preprint 🚨 Are reinforcement learning models complete accounts of decisions from experience if they ignore explicit memory? In this new preprint, we show that people indeed form robust explicit memory representations that flexibly guide later decisions. 🔗 Preprint: doi.org/10.1101/2025...

03612

Dirk Wulff @dirkwulff.bsky.social · 23/10/2025

🚨 New publication: How to improve conceptual clarity in psychological science? Thrilled to see this article with @ruimata.bsky.social out. We discuss how LLMs can be leveraged to map, clarify, and generate psychological measures and constructs. Open access article: doi.org/10.1177/0963...

04419

Dirk Wulff @dirkwulff.bsky.social · 19/10/2025

Thanks. Yeah, it's probably a consequence of the search query we used.

110

Dirk Wulff @dirkwulff.bsky.social · 17/10/2025

🚨 New preprint and online tool 🚨 Thrilled to share new work, mapping the 🗺️ landscape of behavioral reinforcement learning using an 🤖 LLM-powered bibliometric approach. We built an online tool so you can explore the landscape yourself. Online tool: mpib.berlin/vFVqU Preprint: osf.io/6c2va_v1

1153

Dirk Wulff @dirkwulff.bsky.social · 16/10/2025

🚨 New preprint 🚨 Analyzing the academic trajectories of 78,216 psychology researchers, we demonstrate a persistent gender attrition gap, with women psychologists dropping out of academia at consistently higher rates than men psychologists. Preprint: arxiv.org/pdf/2510.13273

14119

Dirk Wulff @dirkwulff.bsky.social · 15/10/2025

🚨 Doctoral positions 🚨 The IMPRS LIFE is looking to fill up to three PhD positions. If you’re interested in working at @arc-mpib.bsky.social with me on topics related to large language models, semantic networks, and aging, please reach out! Apply by Nov 30: www.imprs-life.mpg.de/11840/how-to...

063

Dirk Wulff @dirkwulff.bsky.social · 09/10/2025

🚨 New article 🚨 Excited to share this article, presenting a full tutorial on mouse tracking and other movement tracking techniques in psychological research, with examples from our R package mousetrap. Paper: link.springer.com/article/10.3... R package: pascalkieslich.github.io/mousetrap/in...

075

Dirk Wulff @dirkwulff.bsky.social · 28/08/2025

🚨 New preprint 🚨 What are the reasons underlying human choice in the face of risk? Excited to share new work with Kamil Fulawka and Ralph Hertwig, establishing #LLMs as a scalable solution to uncovering the reasons behind people's choices from free text reports. Preprint: osf.io/preprints/ps...

071

Dirk Wulff @dirkwulff.bsky.social · 13/08/2025

🚨 New article 🚨 How to measure semantic #networks of individuals reliably? Excited to share this article, led by @aeschbach.bsky.social and with @ruimata.bsky.social. Based on simulations, we provide recommendations on designing and interpreting behavioral studies. Article: lnkd.in/dhWiwEBU

080

Dirk Wulff @dirkwulff.bsky.social · 23/07/2025

🚨 New publication 🚨 Excited to see this published in Findings of #ACL2025, led by @zakashussain.bsky.social. We critically evaluate claims that #LLMs are "just next token predictors" or "just machines" and call for a more measured discussion on LLM cognition. aclanthology.org/2025.finding...

062

Dirk Wulff @dirkwulff.bsky.social · 23/06/2025

Happy to have concluded another iteration of our 5-day open LLMs course, together with @zakashussain.bsky.social. If you are interested in LLMs for behavioral and social sciences, check out our... Tutorial: doi.org/10.3758/s134... Open materials: github.com/zak-Hussain/...

072

Dirk Wulff @dirkwulff.bsky.social · 20/06/2025

🚨 New preprint 🚨 How to improve risk communication? In this new piece, spearheaded by ‪@kevinetiede.bsky.social‬, we argue that traditional risk communication often fails to convey experiential dimensions of risk, and that experiential simulations can help fill this gap. doi.org/10.31234/osf...

073

Dirk Wulff @dirkwulff.bsky.social · 29/05/2025

🚨 Postdoc position in India 🚨 Work with Kavitha Ranganathan and me to explore the potential of LLMs for improving risk communication. The position (Senior Research Fellow) includes a research stay and affiliation with@arc-mpib.bsky.social at @mpib-berlin.bsky.social.

132

Dirk Wulff @dirkwulff.bsky.social · 15/05/2025

🚨 New preprint 🚨 Excited to share this work with Dorothée Bentz, where we map the landscape of contamination OCD triggers and analyze individual differences in trigger importance and relatedness. Preprint: osf.io/preprints/ps...

021

Dirk Wulff @dirkwulff.bsky.social · 30/04/2025

How to improve conceptual clarity in psychology? Excited to share a new preprint with @ruimata.bsky.social, discussing approaches based on large language models as possible solutions. Preprint: osf.io/preprints/ps...

0145

Dirk Wulff @dirkwulff.bsky.social · 11/03/2025

How can we reduce conceptual clutter in the psychological sciences? @ruimata.bsky.social and I propose a solution based on a fine-tuned 🤖 LLM (bit.ly/mpnet-pers) and test it for 🎭 personality psychology. The paper is finally out in @natrevpsych.bsky.social: go.nature.com/4bEaaja

15229

Dirk Wulff @dirkwulff.bsky.social · 20/02/2025

Do LLMs think? Excited to share our updated preprint critically discussing two "Justaic" stances claiming that LLMs lack cognition because they are "just" next-token predictors or "just" machines. Led by @zakashussain.bsky.social and with @ruimata.bsky.social . osf.io/preprints/os...

3163

Dirk Wulff @dirkwulff.bsky.social · 19/12/2024

Yes. We need to rely more on open LLMs: osf.io/preprints/os...

040

Dirk Wulff @dirkwulff.bsky.social · 10/12/2024

🚨 New preprint 🚨 Excited to share this preprint led by Zak Hussain (with @ruimata.bsky.social and Ben Newell), comparing the semantic contents of embedding models from text, brain, and behavior data using the psychNorms metabase: github.com/zak-hussain/... Preprint: arxiv.org/pdf/2412.04936

0158

Dirk Wulff @dirkwulff.bsky.social · 02/12/2024

This is an instructive illustration of crucial differences between humans and LLMs. However, we should keep in mind that just as with humans, it matters a lot how we ask LLMs questions. Also, it is worth noting that similar familiarity/fluency processes have long been demonstrated in humans, too.

191

Dirk Wulff @dirkwulff.bsky.social · 22/11/2024

🚨 New article 🚨 Can experienced-based simulations help communicate financial risks? Replicating and extending past work, this article with Tomás Lejarraga and Kavitha Ranganathan concludes that the benefits of experienced-based simulations are limited. Article: doi.org/10.1017/jdm....

083

Dirk Wulff @dirkwulff.bsky.social · 18/11/2024

🚨 New article 🚨 Analyzing over 1,000,000 sampling decisions, @spektor.bsky.social and I show how people are adaptive explorers who reduce multiple types of uncertainty and leverage prior expectations and meta-knowledge about their decision environment. Open access article: lnkd.in/dWdSvQSq

0199

Dirk Wulff @dirkwulff.bsky.social · 07/10/2024

🚨 Job alert 🚨 One more week to apply for two positions (PhD and Postdoc) at @arc-mpib.bsky.social on using LLMs to improve generalizability in the behavioral and social sciences. The positions are fully funded for three years. Apply by October 13! Thank you for sharing!

043

Dirk Wulff @dirkwulff.bsky.social · 14/09/2024

🚨 Job alert 🚨 I am looking for a PhD and Postdoc to join @arc-mpib.bsky.social and work with me on using LLMs to improve the generalizability in the behavioral and social sciences. The positions are fully funded for three years. Apply by October 13! Thank you for sharing!

056