Norman Lab

Norman Lab

@ptoncompmemlab.bsky.social

Princeton Computational Memory Lab https://compmem.princeton.edu

2.3K followers100 following35 posts

Norman Lab reposted

Rolando Masís-Obando
@xrmasiso.bsky.social

What if we could tell you how well you’ll remember your next visit to your local coffee shop? ☕️ In our new Nature Human Behaviour paper, we show that the 𝗾𝘂𝗮𝗹𝗶𝘁𝘆 𝗼𝗳 𝗮 𝘀𝗽𝗮𝘁𝗶𝗮𝗹 𝗿𝗲𝗽𝗿𝗲𝘀𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻 can be measured with neuroimaging – and 𝘁𝗵𝗮𝘁 𝘀𝗰𝗼𝗿𝗲 𝗽𝗿𝗲𝗱𝗶𝗰𝘁𝘀 𝗵𝗼𝘄 𝘄𝗲𝗹𝗹 𝗻𝗲𝘄 𝗲𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲𝘀 𝘄𝗶𝗹𝗹 𝘀𝘁𝗶𝗰𝗸.

Nature Human Behaviour@nathumbehav.nature.com

New study by @xrmasiso.bsky.social et al shows that spatial contexts with more reliable brain representations better support memory for future experiences within them, revealing how stable neural maps help the brain organize and recall life events.

nature.comSpatial contexts with reliable neural representations support reinstatement of subsequently placed objects - Nature Human BehaviourThis study shows that spatial contexts with more reliable brain representations better support memory for future experiences within them, revealing how stable neural maps help the brain organize and recall life events.
Norman Lab
@ptoncompmemlab.bsky.social

New paper led by @codydong.bsky.social now out in @cp-trendscognsci.bsky.social, exploring the relationship between memory-augmented LLMs and human episodic memory – see Cody’s post below for a short thread and a non-paywalled paper link! #NeuroAI doi.org/10.1016/j.ti...

doi.orgRedirecting
Cody Dong@codydong.bsky.social

My first, first author paper, comparing the properties of memory-augmented large language models and human episodic memory, out in @cp-trendscognsci.bsky.social! authors.elsevier.com/a/1lV174sIRv... Here’s a quick 🧵(1/n)

authors.elsevier.comauthors.elsevier.com
Norman Lab
@ptoncompmemlab.bsky.social

New paper led by @jayneuro.bsky.social: Repetition of musical themes in Eternal Sunshine of the Spotless Mind reactivates memories of earlier scenes, and this neural reactivation correlates with subsequent memory for those scenes! Check out Jamal's thread below 👇

Jamal A. Williams@jayneuro.bsky.social

Music is an incredibly powerful retrieval cue. What is the neural basis of music-evoked memory reactivation? And how does this reactivation relate to later memory for the retrieved events? In our new study, we used Eternal Sunshine of the Spotless Mind to find out. www.biorxiv.org/content/10.1...

biorxiv.orgMusic-evoked reactivation during continuous perception is associated with enhanced subsequent recall of naturalistic eventsMusic is a potent cue for recalling personal experiences, yet the neural basis of music-evoked memory remains elusive. We address this question by using the full-length film Eternal Sunshine of the Spotless Mind to examine how repeated musical themes reactivate previously encoded events in cortex and shape next-day recall. Participants in an fMRI study viewed either the original film (with repeated musical themes) or a no-music version. By comparing neural activity patterns between these groups, we found that music-evoked reactivation of neural patterns linked to earlier scenes in the default mode network was associated with improved subsequent recall. This relationship was specific to the music condition and persisted when we controlled for a proxy measure of initial encoding strength (spatial intersubject correlation), suggesting that music-evoked reactivation may play a role in making event memories stick that is distinct from what happens at initial encoding. ### Competing Interest Statement The authors have declared no competing interest. National Institutes of Health, https://ror.org/01cwqze88, F99 NS118740, R01 MH112357
Norman Lab
@ptoncompmemlab.bsky.social

Thrilled that this collaborative project with the @maureenritchey.bsky.social lab, co-led by @gushennings.bsky.social and Paula Brooks, is out in preprint form!

Gus Hennings@gushennings.bsky.social

Hello Bsky world, I am excited to announce our latest preprint “Eye movements reveal the cognitive dynamics supporting successful memory suppression” in collaboration with Paula Brooks, @ptoncompmemlab.bsky.social, and @maureenritchey.bsky.social! 🎉👀 osf.io/preprints/ps... /thread

osf.ioOSF
Norman Lab
@ptoncompmemlab.bsky.social

@qlu.bsky.social is starting his lab at City U of Hong Kong! This is a truly amazing opportunity for trainees interested in computational cognitive neuroscience and neuroAI

Qihong (Q) Lu@qlu.bsky.social

I’m thrilled to announce that I will start as a presidential assistant professor in Neuroscience at the City U of Hong Kong in Jan 2026! I have RA, PhD, and postdoc positions available! Come work with me on neural network models + experiments on human memory! RT appreciated! (1/5)

Norman Lab reposted

Chris Baldassano
@chrisbaldassano.bsky.social

Thank you to Ingrid Wickelgren and the team at Quanta for putting together this great piece, describing work by my lab and others on the neural representations of events

Quanta Magazine@quantamagazine.org

New neuroscience research is en route to unlocking a universal human code for recording experiences as memory. What’s their secret weapon? Airport scenes in movies. www.quantamagazine.org/how-event-sc...

quantamagazine.orgHow ‘Event Scripts’ Structure Our Personal Memories | Quanta MagazineBy screening films in a brain scanner, neuroscientists discovered a rich library of neural scripts — from a trip through an airport to a marriage proposal — that form scaffolds for memories of our exp...
Norman Lab
@ptoncompmemlab.bsky.social

Our paper on how neural codes track prior events in a narrative and predict subsequent memory for details, led by @collinsilvy.bsky.social, is now out in Communications Psychology! rdcu.be/d93Vc #neuroskyence #psychscisky

rdcu.beNeural codes track prior events in a narrative and predict subsequent memory for detailsCommunications Psychology - Context-dependent temporal structures are represented in multiple ways in parallel in the human brain during processing of temporally extended events. The neural...

Norman Lab reposted

Sebastian Michelmann
@s-michelmann.bsky.social

Excited to share our preprint "Fast-timescale hippocampal processes bridge between slowly unfurling neocortical states during memory search" 🧠✨ We leverage iEEG to elucidate the fast neural mechanisms by which long multimodal narratives are unfurled in continuous memory-search tinyurl.com/wjkr3dvf

tinyurl.comFast-timescale hippocampal processes bridge between slowly unfurling neocortical states during memory searchPrior behavioral work showed that event structure plays a key role in our ability to mentally search through memories of continuous naturalistic experience. We hypothesized that, neurally, this memory...
Norman Lab
@ptoncompmemlab.bsky.social

Training in the Method of Loci promotes the development of widespread conjunctive representations in neocortex that bind items and loci -- very excited to be part of this project led by @huangjiawen.bsky.social and @chrisbaldassano.bsky.social

Jiawen Huang@huangjiawen.bsky.social

New preprint with Akshay Manglik, Nick Dutra, Hannah Tarder-Stoll, @tchamberlain.bsky.social, Robert Ajemian, @qiongzhang.bsky.social, @ptoncompmemlab.bsky.social, @chrisbaldassano.bsky.social looking at conjunctive representation during Method of Loci! www.biorxiv.org/content/10.1...

biorxiv.orgBinding items to contexts through conjunctive neural representations with the Method of LociSchematic prior knowledge can provide a powerful scaffold for episodic memories, yet the neural mechanisms underlying this scaffolding process are still poorly understood. A crucial step of the scaffo...

Norman Lab reposted

Sebastian Michelmann
@s-michelmann.bsky.social

So happy that our paper on event segmentation in large language models is now out in Behavior Research Methods! tinyurl.com/2j76882b With @mtoneva.bsky.social, @ptoncompmemlab.bsky.social, and Manoj Kumar, we show that LLMs can segment narrative text into meaningful events similarly to humans.

tinyurl.comLarge language models can segment narrative events similarly to humans - Behavior Research MethodsHumans perceive discrete events such as “restaurant visits” and “train rides” in their continuous experience. One important prerequisite for studying human event perception is the ability of researche...
Norman Lab
@ptoncompmemlab.bsky.social

Thrilled that this epic study led by @coralineiordan.bsky.social is now out in PNAS!

Coraline Rinn Iordan@coralineiordan.bsky.social

New paper story time (now out in PNAS)! We developed a method that caused people to learn new categories of visual objects, not by teaching them what the categories were, but by changing how their brains worked when they looked at individual objects in those categories. www.pnas.org/doi/10.1073/...

pnas.orgSculpting new visual categories into the human brain | PNASLearning requires changing the brain. This typically occurs through experience, study, or instruction. We report an alternate route for humans to a...

Norman Lab reposted

Rolando Masís-Obando
@xrmasiso.bsky.social

🔔𝐍𝐄𝐖 𝐏𝐑𝐄𝐏𝐑𝐈𝐍𝐓 𝐀𝐋𝐄𝐑𝐓🔔 Beyond excited to present our new work showcasing 𝐡𝐨𝐰 𝐰𝐞 𝐜𝐚𝐧 𝐩𝐫𝐞𝐝𝐢𝐜𝐭 𝐡𝐨𝐰 𝐰𝐞𝐥𝐥 𝐲𝐨𝐮 𝐰𝐢𝐥𝐥 𝐫𝐞𝐢𝐧𝐬𝐭𝐚𝐭𝐞 𝐚 𝐧𝐞𝐰 𝐦𝐞𝐦𝐨𝐫𝐲 𝐛𝐞𝐟𝐨𝐫𝐞 𝐭𝐡𝐚𝐭 𝐦𝐞𝐦𝐨𝐫𝐲 𝐢𝐬 𝐟𝐨𝐫𝐦𝐞𝐝! Wait what? Exciting collab w/ @ptoncompmemlab.bsky.social & @chrisbaldassano.bsky.social Link: www.biorxiv.org/content/10.1... (1/11)

View full thread
View full thread
Norman Lab
@ptoncompmemlab.bsky.social

We also hypothesized that neural differentiation would be correlated with the amount the predicted item came to mind during prediction errors, more so in the REM group than the Wake and non-REM sleep-only groups. This pattern was reliable (at an uncorrected threshold) in bilateral DG. (7/9)

View full thread
Norman Lab
@ptoncompmemlab.bsky.social

Using the same task as Kim et al. (2017), we measured how the representations of A and B changed across a period of consolidation with fMRI. We manipulated the presence or absence of REM sleep in a daytime nap during that consolidation period (we also included a quiet wake control group). (5/9)

Norman Lab
@ptoncompmemlab.bsky.social

We found more differentiation in the group with REM sleep than the Wake and non-REM sleep-only groups in the right CA2/3/DG (significant at an uncorrected threshold). An exploratory analysis found that the effect was concentrated in the right DG. (6/9)

View full thread
Norman Lab
@ptoncompmemlab.bsky.social

Here, we sought to test the preregistered hypothesis that learning during REM sleep helps differentiate the neural representations of related memories, by expanding on a prior fMRI study by Kim et al. (2017) showing that prediction errors lead to neural differentiation in the hippocampus… (3/9)

Norman Lab
@ptoncompmemlab.bsky.social

Kim et al. found that, when an item predicted in a particular context (e.g., A predicts B) failed to appear and was later restudied in a different context, the representations of A and B became less similar in the CA2/3/DG region of the hippocampus (Kim et al., 2017). (4/9)

Norman Lab
@ptoncompmemlab.bsky.social

New paper led by Kailong Peng: Real-time fMRI neurofeedback that promotes coactivation of cortical representations drives integration in the hippocampus #neuroskyence #psychscisky royalsocietypublishing.org/doi/10.1098/...

royalsocietypublishing.orgInducing representational change in the hippocampus through real-time neurofeedback | Philosophical Transactions of the Royal Society B: Biological SciencesWhen you perceive or remember something, other related things come to mind, affecting how these competing items are subsequently perceived and remembered. Such behavioural consequences are believed to...
View full thread
Norman Lab
@ptoncompmemlab.bsky.social

The model predicts that, when differentiation occurs as a result of this unsupervised learning mechanism, it will be rapid and asymmetric, and it will give rise to anticorrelated representations in the region of the brain that is the source of the differentiation. (4/5)

Norman Lab
@ptoncompmemlab.bsky.social

Classic supervised learning models posit that, when two stimuli predict similar outcomes, their representations integrate. However, these models have recently been challenged by studies showing that pairing stimuli with a shared associate can sometimes cause differentiation. (2/5)

Norman Lab
@ptoncompmemlab.bsky.social

We provide an unsupervised NN model that can explain these and other related findings by implementing the Nonmonotonic Plasticity Hypothesis (NMPH), whereby moderate coactivity weakens weights, leading to differentiation, and strong coactivity strengthens weights, leading to integration (3/5)

Norman Lab
@ptoncompmemlab.bsky.social

New preprint from @qlu.bsky.social and Ali Hummos, showing how episodic memory supports the learning of structured task representations!

Qihong (Q) Lu@qlu.bsky.social

How can we efficiently find task representations online that work well for the ongoing task while leveraging relations with other tasks? Our simulations show that episodic memory can help!  URL: shorturl.at/tT246 Here’s a summary: 1/n

shorturl.atEpisodic memory supports the acquisition of structured task representationsGeneralization to new tasks requires learning of task representations that accurately reflect the similarity structure of the task space. Here, we argue that episodic memory (EM) plays an essential role in this process by stabilizing task representations, thereby supporting the accumulation of structured knowledge. We demonstrate this using a neural network model that infers task representations that minimize the current task's objective function; crucially, the model can retrieve previously encoded task representations from EM and use these to initialize the task inference process. With EM, the model succeeds in learning the underlying task structure; without EM, task representations drift and the network fails to learn the structure. We further show that EM errors can support structure learning by promoting the activation of similar task representations in tasks with similar sensory inputs. Overall, this model provides a novel account of how EM supports the acquisition of structured task representations. ### Competing Interest Statement The authors have declared no competing interest.

Norman Lab reposted

Paul Scotti
@humanscotti.bsky.social

🧠👁️ Our MindEye2 preprint is out! We reconstruct seen images from fMRI activity using only 1 hour of training data. This is possible by first pretraining a shared-subject model using other people's data, and then fine-tuning on a held-out subject with only 1 hr of data. arxiv.org/abs/2403.11207

Norman Lab
@ptoncompmemlab.bsky.social

Delighted to share this new work led by @collinsilvy.bsky.social, with Ross Kempner and Sunita Srivatasan ! #neuroskyence #psychscisky

Silvy Collin@collinsilvy.bsky.social

I’m excited to announce our new preprint -- Neural codes track prior events in a narrative and predict subsequent memory for details! URL: https://biorxiv.org/cgi/content/short/2024.02.13.580092v1 (1/8)

Norman Lab
@ptoncompmemlab.bsky.social

Now out in Cognitive Science! We use GPT-2 to show that Bayesian surprise, which tracks changes in the content and confidence of predictions, accounts for human event boundaries in a narrative better than surprisal, which only tracks the predicted probability of individual words tinyurl.com/kd9dpub8

Norman Lab reposted

Paul Scotti
@humanscotti.bsky.social

Our MindEye fMRI-to-Image paper got accepted as spotlight for #NeurIPS2023! See you in New Orleans, will be my first time attending NeurIPS :) Updated camera ready paper is also now live on arxiv: arxiv.org/abs/2305.182... Includes new expts, appendix figures, more references to other work 🧠📈

You are up to date.