I'm a Principal Researcher in Reinforcement Learning at Microsoft Research NYC.
Every Friday (4 PM ET) I cohost The Learning Salon (with John Krakauer).
In 2023 I am delighted to serve as the 10th cohort of mentors at the New Inc, the incubator for creative science, the new museum.
I study how we build models of the world and use them in memory, exploration, & planning.
I build and test brain & behavior inspired algorithms for learning & planning, e.g., AI for gaming with Xbox. My approach combines reinforcement learning, neural networks, large language models, & machine learning with behavioral experiments, fMRI, & electrophysiology.
My training is in cognitive computational neuroscience (via computer science and philosophy). I've previously worked at Columbia University, Electrophysiology, Memory, and Navigation Lab, did my postdoc at Princeton (where I collaborated with Ken Norman, Matt Botvinick, Jon Cohen, Nathaniel Daw), my PhD was in psychology (Berlin, Germany, Bernstein Center for Computational Neuroscience), BSc in software engineering (Tehran, Iran), & MSc in Philosophy of Science (Utrecht, Netherlands).
Publications & preprints
Hosein Hasanbeig, Hiteshi Sharma, Leo Betthauser, Felipe Vieira Frujeri,
Ida Momennejad (2023) ALLURE: Auditing and improving LLM-based evaluation of text using iterative in-context-learning.
[ arXiv ]
Safoora Yousefi*, Hosein Hasanbeig*, Leo Betthauser*, Akanksha Saran, Raphael Milliere, Ida Momennejad (2023) Neuroscience-inspired analysis of latent representations in large language models before and after in-context-learning.
Taylor Webb*, Shanka Subra Mondal*, Yi Wan, Brian Krabach, Ida Momennejad (2023) A prefrontal cortex-inspired architecture for planning in Large Language Models.
Ida Momennejad*, Hosein Hasanbeig*, Felipe Vieira Frujeri*, Hiteshi Sharma, Robert Ness, Nebojsa Jojic, Hamid Palangi, Jonathan Larson (2023) Evaluating cognitive maps and planning in Large Language Models with CogEval.
[ arXiv ] NeurIPS 2023.
Ida Momennejad (2023) Multiscale Predictive Representations Connect Memory, Space, and Planning in Brains and Machines. To appear in Space, Time, and Memory, Edited by Lynn Nadel and Sara Aronovitz, forthcoming in Oxford University Press.
Imitating Human Behavior with Diffusion Models (2023) Pearce, Rashid, Kanervisto, Bigness, Sun, Georgescu, Mac, Zheng Tan, Momennejad, Hofmann, Devlin. ICLR 2023 [ link ]
Ali Rahimi-Kalahroudi, Janarthanan Rajendran, Ida Momennejad, Harm van Seijen, Sarath Chandar (2023) Replay Buffer With Local Forgetting for Adaptive Deep Model-Based Reinforcement Learning
[ arXiv ]
Stephanie Milani, Arthur Juliani, Ida Momennejad, Raluca Georgescu, Jaroslaw Rzpecki, Alison Shaw, Gavin Costello, Fei Fang, Sam Devlin, Katja Hofmann (2023) Navigates Like Me: Understanding How People Evaluate Human-Like AI in Video Games. ACM CHI (Conference on Human Factors in Computing Systems) 2023. [ arXiv ]
Ida Momennejad (2022) A Rubric for Human-like Agents and NeuroAI.
Philosophical Transactions of the Royal Society B
Eigen Memory Trees (2022) Mark Rucker, Jordan T. Ash, John Langford, Paul Mineiro, Ida Momennejad
In revision [ arXiv ]
Social Network Structure Shapes Innovation: Experience-sharing in RL with SAPIENS (2022) Eleni Nisioti, Mateo Mahaut, Pierre-Yves Oudeyer, Ida Momennejad, Clément Moulin-Frier
Interaction-Grounded Learning with Action-inclusive Feedback
(2022) Tengyang Xie, Akanksha Saran, Dylan J. Foster, Lekan Molu, Ida Momennejad, Nan Jiang, Paul Mineiro, John Langford
NeurIPS 2022 [ arXiv preprint]
Towards Evaluating Adaptivity of Model-Based Reinforcement Learning Methods (2022) Yi Wan, Ali Rahimi-Kalahroudi, Janarthanan Rajendran, Ida Momennejad, Sarath Chandar, Harm van Seijen [ arXiv ]
NeuroNav: A Library for Neuroscience Inspired Learning (2022) Julliani, Barnett, Davis, Sereno, Momennejad, RLDM
PARSR: Priority Adjusted Replay for Successor Representation (2022)
Barnett, Momennejad, RLDM 2022
Describeland: One-Shot Learning from a Demonstration with Hierarchical Latent Language (2022) Nathaniel Weir, Xingdi Yuan, Marc-Alexandre Cote, Matthew Hausknecht, Romain Laroche, Ida Momennejad, Harm Van Seijen, Benjamin Van Durme. [ In revision ]
How Humans Perceive Human-like Behavior in Video Game Navigation (2022) Zuniga*, Milani*, Leroy*, Rzepecki, Georgescu, Momennejad, Bignell, Sun, Shaw, Costello, Jacob, Devlin, Hofmann.
CHI Late Breaking Work, 2022. [ PDF ]
Momennejad I (2021) Collective Minds: Social Network Topology Shapes Collective Cognition. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 377 (1843): 20200315.
[ PhilTrans , PDF ]
Pudhiyidath A, Morton NW, Viveros Duran R, Schapiro AC, Momennejad I, Hinojosa-Rowland DM, Molitor RJ, Preston AR (2022-accepted) Representations of temporal community structure in hippocampus and precuneus predict inductive reasoning decisions.
The Journal of Cognitive Neuroscience [ bioRxiv ]
Vlasceanu M, Dudik M, Momennejad I (2021) Network Structure, Gender Diversity, and Interdisciplinarity Predict the Centrality of AI Organizations.
[ psyRxiv PDF ]
Russek EM, Momennejad I, Botvinick M, Gershman S, Daw N (2021) Neural evidence for the successor representation in choice evaluation.
[ bioRxiv PDF ]
Devlin S*, Georgescu R*, Momennejad I*, Rzepecki J*, Zuniga E*, Costello G, Leroy G, Shaw A, Hofmann K (2021) Navigation Turing Test (NTT): Learning to Evaluate Human-Like Navigation. ICML 2021. [ arXiv PDF ]
Momennejad I (2020) Learning Structures: Predictive Representations, Replay, and Generalization. Current Opinions in Behavioral sciences.
Brunec I, Momennejad I (2019) Predictive Representations in Hippocampal and Prefrontal Hierarchies.
[ bioRxiv preprint ]
Momennejad I, Norman KA, Cohen JD, Singh S, Lewis RL (2019). Rational use of Episodic and Working Memory: A Normative Account of Prospective Memory. BioRxiv, 580324 https://doi.org/10.1101/580324.
[ bioRxiv preprint, SfN poster ]
Momennejad I, Howard M (2018) Predicting the future with multi-scale successor representations.
[ bioRxiv preprint, status: in revision ]
Momennejad I, Otto RA, Daw N, Norman KA (2018) Offline replay supports planning in human reinforcement learning. eLife 2018;7:e32548.
Russek E*, Momennejad I*, Botvinick MM, Gershman SJ, Daw N (2017) Predictive representations can link model-based reinforcement learning to model-free mechanisms. Plos Comp Biol.
[ Journal link, PDF, bioRxiv preprint, CODE ]
Momennejad I*, Russek E*, Cheong JH, Botvinick MM, Daw N, Gershman SJ (2017) The successor representation in human reinforcement learning: evidence from retrospective revaluation. Nature Human Behaviour, 1. [ Nat Hum Beh paper , PDF, free preprint , my python tutorials comparing SR & SR-Dyna, blogpost from Deepmind ]
Coman A, Momennejad I, Drach R, Geana A (2016) Mnemonic convergence in social networks: The emergent properties of cognition at a collective level. PNAS. doi: 10.1073/pnas.1525569113.
[ Journal link , Podcast ]
Wisniewsky D, Reverberi C, Momennejad I, Kahnt T, Haynes J-D (2016) The role of parietal cortex in the representation of task-reward-association. The Journal of Neuroscience. [ Journal link ]
Haynes J-D, Wisniewsky D, Gorgen K, Momennejad I, Reverberi C (2015) fMRI decoding of intentions: compositionality, hierarchy, and prospective memory. Brain-Computer Interface (BCI), 3rd International Winter Conference. doi: 10.1109/IWW-BCI.2015.7073031. [ IEEE link ]
MIND 2019 "Cognitive Maps" computational summer school (Methods In Neuroscience at Dartmouth)
- Lecture, tutorial, and hackathon team.
MIND 2018 computational summer school (Methods In Neuroscience at Dartmouth)
- Lecture, tutorial, and hackathon team.
MIND 2017 computational summer school (Methods In Neuroscience at Dartmouth)
- Two Lectures, tutorial, and hackathon team.
Introduction to cognitive neuroscience - with Mattew Botvinick at Princeton University
Sample Talks and Workshops (April-September 2019)
April 21, 1-5 PM: Computational Justice Workshop (hands-on agent-based simulation in Python), Heart of the Machine series, Pioneerworks, Brooklyn, NY
May 13-14: Context and Episodic Memory (CEMS), Philadelphia, PA
May 16-18: Control Processes 2019, Brown University, Providence, RI
May 28-29: Talk and visit, Computational Neuroscience Center, University of Washington, Seatle, WA
July 10, 1-5PM: Structure for Efficient Reinforcement Learning (SERL) Workshop
4th Multidisciplinary Conference on Reinforcement Learning and Decision Making (RLDM), Montreal, Canada
August 8-17: "Cognitive Maps" at MIND19 summer school for computational methods in neuroscience at Dartmouth
September 10-13: What is biological Computation? Santa Fe Institute
Momennejad I, Allahyari M (2017) Refiguring: in Conversation. Futureproof, Haverford College, Haverford, PA, pp 16-31.
Momennejad I (2011) Thinking in movements, rehearsing freedom. In: Wagner L and Gomes-Carrillo de Castro Eds., Sounds, Space, Body, a process. Ernst Schering Foundation and Association of Neuroesthetics: Berlin, pp:115-125.
Momennejad I (2011) Seeing with eyes closed: the neuro-epistemology of perceptual reality. In: Agudio E and Franke I Eds., Seeing with eyes closed. Association of Neuroesthetics: Berlin, pp:15-21.
Momennejad I and Franke I (2011) Two practices of seeing with eyes closed: contemporary art and science in dialogue. In: Agudio E and Franke I Eds., Seeing with eyes closed. Association of Neuroesthetics: Berlin, pp:9-13.
Media: Talks, outreach, press
A video of talk at the Stanford Psychology Colloquium "Predictive representations in memory and planning", Stanford University, 2017.
A video of talk on our elife paper"Offline Replay Supports Planning" at CCN 2017 (Cognitive Computational Neuroscience conference), Columbia Univesity, New York.
Video of MIND2017 workshop, "Thinking in graphs (Day 2): Social temporal networks and collective memory", Dartmouth college, 2017.
Video of MIND2017 workshop, "Thinking in graphs (Day 1): Cognitive maps & memory replay", Dartmouth college, 2017.
Video of MIND2018 workshop, Dartmouth college, 2018.
A video of keynote talk "Predictive maps: What does it mean to have human-like memory and agency?" (min 16 onward), at Facets Conference 2017, The Future of AI, Goethe Institut, New York.
Deep mind blog post featuring a new article on predictive representations as well as my paper on the successor representation.
Spinney L (7 March 2017) "How facebook, fake news and friends are warping your memory. Research on collective recall takes on new importance in a post-fact world". Nature feature on our work on collective memory