Sam Nastase @samnastase Twitter profile

Pinned Tweet

Sam Nastase

2 months

Happy to see this work led by @zaidzada_ now published in @NeuroCellPress ! We use LLM embeddings to capture word-by-word linguistic content transmitted from the speaker's brain to the listener's brain in real-time, face-to-face conversations:

4

66

239

Last Seen Profiles

@bisceoff

@deb_santell

@Book_club_mumma

@knybeer

@utongnikaiser

@tsutomu0104

@samif9999

@yskyapu

@cukienaknikmati

@Oyebanji1414751

@kyonkyon1966

@muhamad92053877

@pandy_token

@emmabroadhurst1

@GlontiTea

@rulesbycam

@BinorRaja

@CburySquad

@BinorRaja

@stwmaniax

@KyberPlay

@xKaii92

@earthtojun

@Hariant86676880

@Nickatina_show

@BeeAhBrem

@RodwellGibbons

@KeliBiancaa

@mckennarobbins7

@JuniorFF00

@jiji_eth

@RicellyVit

@DuCrab

@Bono_zzZZ

@ShammySho

@kdr2L4eIlWgUKfE

Sam Nastase

@samnastase

4 years

@Reza_Zadeh can't wait for one of these robots to dance right through my front door and arrest me

8

9

531

Sam Nastase

@samnastase

4 years

@ldrinkh20

1

3

558

Sam Nastase

@samnastase

4 years

fMRI people be like "I know a spot" and then run 50,000 statistical tests just to show you where it is (I'm sorry)

10

46

494

Sam Nastase

@samnastase

2 years

How does the brain construct meaning from natural language? New preprint led by @sreejan_kumar @tedsumers with @TakaYamakoshi and @hawkrobe ! We ask how the internal computations of a transformer-based language model map onto cortical language processing:

10

96

445

Sam Nastase

@samnastase

4 years

Excited to (finally!) share our preprint describing the "Narratives" collection—compiling 7 years worth of naturalistic story-listening fMRI datasets from @HassonLab and @ptoncompmemlab at @PrincetonNeuro :

9

152

438

Sam Nastase

@samnastase

4 years

@paulkrugman paul krugman has not personally experienced very much anti-muslim sentiment and is willing to misunderstand some data to prove it

1

12

272

Sam Nastase

@samnastase

1 year

I have never been more excited to tell everyone about a paper! New preprint led by @zaidzada_ with @HassonLab : "A shared linguistic space for transmitting our thoughts from brain to brain in natural conversations"

7

82

270

Sam Nastase

@samnastase

5 years

In case you missed my talk at #SfN19 : We're rolling out the "Narratives" collection with fMRI data for 300+ subjects listening to 20+ naturalistic spoken stories (750+ functional scans from @HassonLab @ptoncompmemlab ). Data now available on @OpenNeuroOrg :

8

96

237

Sam Nastase

@samnastase

4 years

We have a short perspective piece out in the @NeuroImage_EiC special issue on Naturalistic Imaging where we try to articulate what a more "ecological" human neuroscience might look like—with a brief history of the term "ecological validity"

2

75

222

Sam Nastase

@samnastase

3 years

Happy to share that the "Narratives" data descriptor is now out @ScientificData ! Paper: Data: Website: Wanted to take this opportunity to highlight some projects already taking advantage of Narratives data:

2

67

189

Sam Nastase

@samnastase

1 year

Devastating to watch colleagues I respect be completely overcome with collective psychosis and tribalism right now. The US government is bankrolling war crimes against the people of Gaza on a world-historic scale, and all you can do is chastise people for speaking up?

9

23

173

Sam Nastase

@samnastase

4 years

If Elon Musk saying dumb shit about neuroscience pisses you off, wait til you hear about how he accumulates obscene wealth by exploiting the labor of working people.

Elon Musk

@elonmusk

4 years

A prior track record of exceptional achievement in engineering *is* required, but *no* prior experience working on the brain or human physiology is required. We will teach you what’s known about the brain, which is not much tbh.

1K

912

20K

4

12

157

Sam Nastase

@samnastase

5 years

Our tutorial review on intersubject correlation (ISC) analyses with @HassonLab , Valeria Gazzola, and Christian Keysers is now up on @biorxivpreprint : . Check out the accompanying Jupyter Notebook tutorial showcasing @UseBrainIAK :

0

67

156

Sam Nastase

@samnastase

6 years

These women are close friends and colleagues, and their perseverance has been a constant source of inspiration for me. Outing these predators and attacking the power structures that support them is a greater contribution to our field than any experiment or manuscript.

Sasha Brietzke

@sbrietz

6 years

I stand in righteous fury beside fellow women in science in filing a lawsuit against Dartmouth. We want our voices to be heard. We demand to be taken seriously. We demand change. Thread (1/10) #metooSTEM #MeToo

66

726

2K

1

20

141

Sam Nastase

@samnastase

4 years

New preprint w/ @ArielYGoldstein @HassonLab et al using GPT2 encoding/decoding models to demonstrate that neural activity during naturalistic story-listening encodes: (1) the context-specific meaning of words in a narrative (2) up to hundreds of milliseconds before word onset.

Hasson Lab

@HassonLab

4 years

New preprint from the Hasson Lab with @ArielYGoldstein and many collaborators at NYU and Google Research showing deep connections between the way the human brain and deep language models like GPT2 process language in natural settings:

9

70

240

1

33

107

Sam Nastase

@samnastase

4 years

Our paper on connectivity SRM w/ @YunFeiLiu3 @hannahillman @ptoncompmemlab @HassonLab is now out in the @NeuroImage_EiC special issue on naturalistic imaging. We use cSRM to estimate a single shared response space across 10 story-listening fMRI datasets.

1

40

105

Sam Nastase

@samnastase

3 years

@Neuro_Skeptic

3

7

95

Sam Nastase

@samnastase

2 years

Our experiments in using real-time fMRI neurofeedback to alter the interpretation of an ambiguous, naturalistic narrative—a colossal amount of work by @acmennen with @ptoncompmemlab —now published in NeuroImage: Reports

Norman Lab

@ptoncompmemlab

3 years

New preprint led by @acmennen , with @samnastase @yeshurunyaara @hassonlab : Real-time neurofeedback to alter interpretations of a naturalistic narrative 🧵

1

28

83

1

23

96

Sam Nastase

@samnastase

5 years

Traditional publication is a bit anticlimactic in the age of preprints, but the final version of our introductory review on intersubject correlation (ISC) methods is available (open access) in the "tools of the trade" series at SCAN:

Measuring shared responses across subjects using intersubject correlation

Abstract. Our capacity to jointly represent information about the world underpins our social experience. By leveraging one individual’s brain activity to m

academic.oup.com

Sam Nastase

@samnastase

5 years

Our tutorial review on intersubject correlation (ISC) analyses with @HassonLab , Valeria Gazzola, and Christian Keysers is now up on @biorxivpreprint : . Check out the accompanying Jupyter Notebook tutorial showcasing @UseBrainIAK :

0

67

156

1

25

89

Sam Nastase

@samnastase

5 years

Our perspective piece w/ @HassonLab is out in @NeuroCellPress ! We try to provide some intuitions on how modern artificial neural networks learn and the implications for neuroscientists, drawing parallels with evolutionary theory and ecological psychology:

Direct Fit to Nature: An Evolutionary Perspective on Biological and Artificial Neural Networks

To guide adaptive behavior and support predictions in real-life contexts, the brain may rely on opaque, over-parameterized models capable of directly fitting to the multidimensional world, while...

www.cell.com

Hasson Lab

@HassonLab

5 years

Our new paper in @NeuroCellPress argues against the classical cognitive opposition to deep neural networks as useful scientific models of the mind. We argue that both the brain and deep neural networks belong to a particular family of "direct-fit" models:

10

209

509

3

27

76

Sam Nastase

@samnastase

5 years

This is an exceptionally bad take. All the sympathy for Dave's family, friends, students; mental illness is awful. But under no circumstances should blame be directed at the plaintiffs.

Nicholas A. Christakis

@NAChristakis

5 years

Plaintiffs & their lawyers in Dartmouth harassment case should be deeply ashamed of their scorched earth tactics, falsely accusing good & innocent man, Prof David Bucci, of complicity,contributing to his suicide. But there's lots of blame to go around 1/

13

46

138

2

7

68

Sam Nastase

@samnastase

4 years

Narratives tap into the brain's built-in machinery for organizing and communicating our knowledge of events in the world—short piece w/ @roelwillemsRU , @branka_mvojevic out in @TrendsNeuro arguing for the importance of narratives in human neuroscience:

Trends in Neurosciences

@TrendsNeuro

4 years

“Narratives for Neuroscience” - @TrendsNeuro ‘Forum’ by @RoelWillemsRU , @SamNastase & @Branka_Mvojevic 50-day free access link in the thread

1

11

32

2

22

70

Sam Nastase

@samnastase

5 years

New preprint w/ @HassonLab @ptoncompmemlab ! We use connectivity SRM to estimate a single shared response space across 10 story-listening datasets (different subjects and stimuli)—dramatically increases dimensionality of shared information across subjects:

0

23

67

Sam Nastase

@samnastase

2 years

New preprint led by @mafeilong ! Years of work developing the next generation of hyperalignment—the Individualized Neural Tuning (INT) model—culminates in this wonderful animation: hyperaligning the "kitty/puppy" neural representational space 🤩

Feilong Ma

@mafeilong

2 years

The INT model is based on a new hyperalignment variant, "warp hyperalignment", which affords modeling both functional and topographic idiosyncrasies ("what" and "where") based on a functional template (hyperalignment common space).

1

4

21

1

13

63

Sam Nastase

@samnastase

5 years

@nandorvila anarcho-gandalf

0

53

Sam Nastase

@samnastase

6 years

New from @haxbylab : Connectivity-based hyperalignment—sophisticated aggregation of fine-grained functional data across subjects—dramatically outperforms anatomical normalization and yields a shared feature space without a shared stimulus

1

24

50

Sam Nastase

@samnastase

5 years

New preprint with @HassonLab . Serious anxiety that we're going to piss off each of these communities—evolutionary biology, AI, neuroscience, psychology—for different reasons. That said, please channel your antagonism into an #openreview on @biorxivpreprint or @PubPeer !

Hasson Lab

@HassonLab

5 years

Excited to release a preprint of our perspective piece on the parallels between evolution, artificial neural networks, and biological neural networks—and the ramifications for experimental neuroscience:

4

59

159

2

6

49

Sam Nastase

@samnastase

1 year

Open review of a nice preprint by @vigliocco_g and colleagues at @ecologicalbrain () on their vision for a more ecological human neuroscience:

Ecological Brain: Reframing the Study of Human Behaviour and Cognition

The last decade has seen substantial advances in the capacity to record behaviour and neural activity in humans in real-world settings, to simulate real-world situations in laboratory settings and to...

osf.io

2

15

49

Sam Nastase

@samnastase

6 years

Whoa, it's a great feeling to see publications based entirely on one of your shared datasets. Especially when they tackle questions you were kicking around but never fully developed. Shout out to @OpenNeuroOrg and @datalad for making it possible:

Frontiers - Neuroscience

@FrontNeurosci

6 years

Goal-Directed Processing of Naturalistic Stimuli Modulates Large-Scale Functional Connectivity: Zhenfu Wen, Tianyou Yu, Xinbin Yang, Yuanqing Li

0

6

1

13

46

Sam Nastase

@samnastase

3 months

Happy to see this work led by @sreejan_kumar and @tedsumers finally out @NatureComms after... ~24 months in our completely normal and well-functioning publication system 🙃 (NB: we updated the title, apologies to your reference manager)

Sreejan Kumar

@sreejan_kumar

3 months

Excited to share our work now published in Nature Communications - shared functional specialization in tansformer language models and the human brain! With co-first author @tedsumers , senior author @samnastase and others ( @hawkrobe , @TakaYamakoshi ). [1/n]

3

60

231

2

3

44

Sam Nastase

@samnastase

6 years

Really excited about this new preprint from @mafeilong and @haxbylab —hyperalignment resolves topographic idiosyncrasies across individuals and improves the reliability of individual differences in fine-grained cortical functional architecture:

0

19

46

Sam Nastase

@samnastase

5 years

starting to get the hang of this social distancing thing

1

2

44

Sam Nastase

@samnastase

1 year

I don't know who needs to hear this, but even if enemy combatants were in fact entrenched in a hospital, that would not even remotely justify leveling the hospital with 1000+ patients and medical staff inside.

3

2

45

Sam Nastase

@samnastase

5 years

@mbeisen @BernieSanders This is disingenuous as hell and I would have expected better from you, @mbeisen .

4

0

40

Sam Nastase

@samnastase

4 years

2020: scientists "discover" that men can in fact be bisexual, eat hot chip, and lie

3

2

42

Sam Nastase

@samnastase

4 years

This is not historically out-of-character for a country built on white supremacy, and is not unlike the cowboy shit we do all around the world. If you think this isn't "us"—you're not paying very close attention.

1

0

40

Sam Nastase

@samnastase

6 years

Goosebumps after @kirstie_j 's #OHBM2018 #Brainhack talk on building open communities in science: "When I get up in the morning, I think what are the three things I need to fight today? 1. patriarchy, 2. white supremacy, 3. capitalism."

1

10

39

Sam Nastase

@samnastase

4 years

Very proud to be part of Jim's ( @haxbylab ) decade of work on hyperalignment: Here are some ways you can provide concrete support to Black communities and antifascists struggling against white supremacy and state-sanctioned violence:

Ida Gobbini

@IGobbini

4 years

This time with the link to the paper! Congratulations to Jim Haxby, @swaroopgj @samnastase @mafeilong for a great review on hyperalignment, just published in @eLife @aina_puce @spornslab @constable_todd @torwager @jeremyRmanning

0

25

84

0

10

39

Sam Nastase

@samnastase

6 years

My first (small-scale!) data publication: . Running experiments with the mindset of publishing the data and sharing every gruesome detail really pushes you to be conscientious about data collection and analysis. Data (and stimuli) at

3

9

37

Sam Nastase

@samnastase

4 years

trying to read theory after melting my brain on social media

0

37

Sam Nastase

@samnastase

5 years

Power to these women for setting the terms of a resolution with delinquent @dartmouth . Money and reputation are the only language institutions like @dartmouth speak.

1

36

Sam Nastase

@samnastase

5 years

btw all anatomical images are de-faced with pydeface () in keeping with best practices—eat your heart out @nytimes

Sam Nastase

@samnastase

5 years

In case you missed my talk at #SfN19 : We're rolling out the "Narratives" collection with fMRI data for 300+ subjects listening to 20+ naturalistic spoken stories (750+ functional scans from @HassonLab @ptoncompmemlab ). Data now available on @OpenNeuroOrg :

8

96

237

2

1

34

Sam Nastase

@samnastase

6 years

@JeremyPhilosoph @batssiki @PaulRoundy1 @yael_niv @MeTooSTEM @McLNeuro @AlertStem More seriously, seeing men taking up space in a thread like this with complacent, reactionary bullshit is repulsive. Your personal anecdotes (and the wildly problematic study cited above) are not a valuable contribution and your attitude is part of the problem.

2

0

33

Sam Nastase

@samnastase

1 year

I have unconditional sympathy for people losing loved ones; civilian casualties are universally atrocious—but just because you're experiencing loss or trauma during wartime does not give you the moral authority to pave the way for genocide.

1

32

Sam Nastase

@samnastase

5 years

Fearless leader @MatteoVdOC is here for Brainhack Princeton and we're channeling @yarikoptic !

0

32

Sam Nastase

@samnastase

4 years

New preprint from @elb7897 @luke_slipski @mafeilong @haxbylab introducing "hybrid hyperalignment"—combines both response time series data and functional connectivity data to estimate a single best-of-both-worlds hyperalignment model:

Erica Busch

@ericalbusch

4 years

check out our new preprint! Hybrid Hyperalignment: A single high-dimensional model of shared information embedded in cortical patterns of response and functional connectivity.

1

23

72

1

8

31

Sam Nastase

@samnastase

5 years

"There is no abusive 'genius' who could not be replaced by someone who isn't shitty" — @phyzang at #NeuroHack19 , credit to @kateleth

Elizabeth DuPre

@emdupre_

5 years

Ending on a slightly different, really (really) important note : academia is very hierarchical, opening us up to abuses of power. We as a community need to collectively work to create a positive culture 💪

0

1

9

1

9

31

Sam Nastase

@samnastase

7 years

New preprint from @caravanuden and @haxbylab combining hyperalignment and forward encoding models to predict detailed semantic tuning across subjects:

1

12

32

Sam Nastase

@samnastase

1 year

I have no patience for "both sides" or "it’s complicated"—it's not complicated! History rarely serves up such a clear-cut case. If you think you would've been one of the good guys when we were decimating indigenous Americans, or carpet-bombing Vietnam—this was your chance.

1

6

30

Sam Nastase

@samnastase

5 years

Most important slide from my favorite talk at #OHBM2019 — @kirstie_j on the deepest challenges for building a more inclusive, equitable science. "Open science must break down: patriarchy, white supremacy, capitalism"

2

5

31

Sam Nastase

@samnastase

5 years

Dinner for one: standing over my sink eating fried eggs off a paper towel.

5

0

30

Sam Nastase

@samnastase

1 year

Israel has had all the power in this situation for decades. Israel has openly identified as a colonial project since the 40s and has been committing war crimes and perpetrating genocide against the people of Palestine—from this position of power!—for decades.

1

30

Sam Nastase

@samnastase

4 years

PSA: Academics, please do not use your platform to boost cute videos of cops kneeling, dancing, or hugging protestors. We get enough of that shit on cable news—it undermines the movement. The pro-police propaganda machine is already very effective, they don't need your help.

2

31

Sam Nastase

@samnastase

5 years

@Reuters

0

28

Sam Nastase

@samnastase

6 years

@JeremyPhilosoph @batssiki @PaulRoundy1 @yael_niv @MeTooSTEM @McLNeuro @AlertStem Jeremy Willis has confidently declared that "no one is biased against women STEM anymore" in response to this thread—just wanted to bring this to your attention!

3

0

28

Sam Nastase

@samnastase

1 year

The silence—and less-than-silence in many cases—from universities and faculty that claim to support DEI while we watch this genocide unfold in real-time is deafening. What are you doing to help your Palestinian students feel safe?

1

4

30

Sam Nastase

@samnastase

4 years

Curating, versioning, and validating this growing dataset to the point that other people (I hope!) can feel confident using the data is probably the most difficult scientific problem I've worked on so far. I hope all the "research parasites" out there will take full advantage!

1

0

28

Sam Nastase

@samnastase

1 year

The Israeli government is unilaterally cutting water and electricity to a city of 2 million people, half of which are children. Israeli state media has bragged about dropping 6,000 bombs on Gaza. 500+ Palestinian children have been killed in the past 6 days. Who has the power?

1

27

Sam Nastase

@samnastase

5 years

Very interesting work from @Guo_Jiahui_ , @IGobbini et al. Subject-specific cortical topographies for functional localizers (e.g., FFA) are embedded in responses to naturalistic movies, and hyperalignment can recover them.

Ida Gobbini

@IGobbini

5 years

Our new paper on estimating the face perception system using hyperalignment is out.

0

9

34

0

6

28

Sam Nastase

@samnastase

4 years

Open review of @biorxivpreprint by @dixy0 and Biswal () using PCA to discover multiple shared fMRI response components during naturalistic movie-watching on @PubPeer : ; and on my website:

0

6

29

Sam Nastase

@samnastase

5 years

Curious about new developments in #hyperalignment ? Come check out our @HassonLab poster (M506) at #OHBM2019 this afternoon—we're using shared connectivity to find a common response space across disjoint datasets (different subjects and different stimuli).

0

8

29

Sam Nastase

@samnastase

1 year

Here's the centerpiece: We find that the same linguistic content that emerges in the speaker's brain before word articulation, later re-emerges in the listener's brain after word articulation.

2

4

28

Sam Nastase

@samnastase

1 year

(A) they are fundamentally racist; (B) they materially / psychologically benefit from keeping the colonial project running and their conscience clean; or (C) they had no principled critique of colonialism to begin with and are overwhelmed by pro-colonial Western propaganda.

2

4

25

Sam Nastase

@samnastase

4 years

Nice to see one of your open datasets appear in a preprint! Open review of work by @MarcCoutanche et al () on their interesting new "representational connectivity" metric via @PubPeer : ; or on my website:

1

5

27

Sam Nastase

@samnastase

6 years

Highly recommend this tool—you'd be extremely hard-pressed to put together a preprocessing pipeline as sophisticated and robust as fMRIPrep (and many of us have spent some years trying). Running an fMRIPrep tutorial for folks in @HassonLab and @ptoncompmemlab next week!

bioRxiv

@biorxivpreprint

6 years

FMRIPrep: a robust preprocessing pipeline for functional MRI #bioRxiv

0

12

7

0

6

27

Sam Nastase

@samnastase

4 years

This is some truly mask-off shit. Here we have the NYPD commissioner clumsily trying to manufacture consent for cops to avoid personal accountability when violently attacking protestors.

Thomas G. Donlon

@NYPDPC

4 years

New Yorkers may begin to see our officers with black mourning bands across their shields—and across their hearts. We wear these in quiet commemoration of our 27 brothers and sisters we’ve lost to #COVID . Another way we honor our vow to #NeverForget .

14K

1K

5K

4

1

27

Sam Nastase

@samnastase

3 years

Exciting to see new papers growing out of the Narratives dataset! If you're interested in naturalistic neuroimaging and how the brain processes spoken language, see if our open dataset might help kickstart your project:

Jean-Rémi King

@JeanRemiKing

3 years

Our latest paper is out: GPT-2’s activations predict the degree of semantic comprehension in the human brain, by @c_caucheteux , @agramfort & @jrking0 The summary thread below 👇 1/8

7

85

368

1

3

25

Sam Nastase

@samnastase

4 years

Here's all the relevant links: Preprint: DataLad: OpenNeuro: Code: Website: Let me know if you encounter any irregularities, or if you have any questions.

GitHub - snastase/narratives: Narratives: fMRI data for evaluating models of naturalistic language...

Narratives: fMRI data for evaluating models of naturalistic language comprehension - snastase/narratives

github.com

1

3

25

Sam Nastase

@samnastase

4 years

We think this dataset has a lot of potential for reuse— spoken stories are very rich stimuli and can inform many hypotheses about language and brain function. We hope the dataset can serve both as a didactic tool and as a benchmark for testing language models against brain data.

1

3

25

Sam Nastase

@samnastase

5 years

More coming soon (via @datalad )—including quality control, preprocessed derivatives, time-stamped transcripts, and a data descriptor preprint... stay tuned! Slides: Code: Website:

GitHub - snastase/narratives: Narratives: fMRI data for evaluating models of naturalistic language...

Narratives: fMRI data for evaluating models of naturalistic language comprehension - snastase/narratives

github.com

1

2

25

Sam Nastase

@samnastase

6 years

Anybody at #OHBM2018 interested in representational similarity analysis or the neural basis of action understanding—come visit! (poster #2859 )

3

8

25

Sam Nastase

@samnastase

6 years

Open peer review of the fMRIPrep preprint () by @oesteban , @ChrisFiloG et al on @AcademicKarma :

FMRIPrep: a robust preprocessing pipeline for functional MRI

Preprocessing of functional MRI (fMRI) involves numerous steps to clean and standardize data before statistical analysis. Generally, researchers create ad hoc preprocessing workflows for each new...

www.biorxiv.org

1

7

25

Sam Nastase

@samnastase

2 years

Come chat with @sreejan_kumar and I about functionally specialized circuits in transformers and how they map onto human brain activity during natural language comprehension at the Friday poster session (P-2.32)! @CogCompNeuro #ccneuro22

Sreejan Kumar

@sreejan_kumar

2 years

I'll be at San Francisco this week presenting this work at @CogCompNeuro ! Please don't hesitate to reach out if you want to chat! #ccneuro22

1

2

29

0

7

25

Sam Nastase

@samnastase

5 years

Important implications for hyperalignment here. As function becomes decoupled from macroanatomy (e.g., in transmodal cortex), we'll need a framework for jointly modeling both shared function and topographic variability across individuals: @haxbylab

A computational model of shared fine-scale structure in the human connectome

Author summary Resting state fMRI has become a ubiquitous tool for measuring connectivity in normal and diseased brains. Current dominant models of connectivity are based on coarse-scale connectivity...

journals.plos.org

Bratislav Misic

@misicbata

5 years

Gradients of structure-function tethering across neocortex | our latest in @PNASNews with @bertha_vr @LauraESuarez24 @rossdavism @goliashf @CaseyPaquola @martijnheuv @BorisBernhardt @Nathan_Spreng @TheNeuro_MNI

1

76

138

0

3

24

Sam Nastase

@samnastase

5 years

Spent 3 hours wrecking and then haphazardly fixing my conda environment on a Sunday afternoon (back on my bullshit) 💅

2

0

22

Sam Nastase

@samnastase

6 years

Can't say I'm surprised— @dartmouth runs like a business and will use whatever means necessary to protect its reputation and ultimately its bottom line. They are literally turning the plaintiffs' initial optimism and faith in the administration against them. Reaching new lows.

Sasha Brietzke

@sbrietz

6 years

Wow. I had low expectations, but Dartmouth's response to the #Dartmouth7 lawsuit is a case study of institutional defensiveness in the face of appropriate demands for positive change. Check it out for yourself in attached link. (Thread 1/5) #MeTooSTEM

4

46

95

0

6

22

Sam Nastase

@samnastase

5 years

Curating data in a way that actually streamlines re-use is hard work and a scientific contribution in its own right! Glad to add my (small) experience to this @nature commentary on building better infrastructure and recognition for data sharing.

2

7

23

Sam Nastase

@samnastase

4 years

If you're a high-profile academic and you're using this opportunity to platform "good cops" or your favorite governor, you need to log off.

0

22

Sam Nastase

@samnastase

1 year

It's like putting a pot on high heat and then crying foul when it boils over. I know a lot of people are in shock right now and this drives them to lash out with prejudice, but the important question to ask in a situation like this is: Who has the structural power?

1

21

Sam Nastase

@samnastase

1 year

Why are people still equivocating? Why are people trying to intimidate dissenters into backing one of the most advanced militaries in the world as it obliterates colonized people? Why are so many people committed to the wrong side of history? This is the best I can come up with:

1

21

Sam Nastase

@samnastase

1 year

Final version out in @PNASNews ! Colossal amount of work by @Guo_Jiahui_ @mafeilong 🙌 Large portion of reliable variance in brain activity when humans view dynamic, naturalistic face videos is *not* captured by state-of-the-art neural networks for face identification:

Guo Jiahui

@Guo_Jiahui_

1 year

Do deep convolutional neural networks (DCNN) trained for face recognition learn internal representations of faces similar to human cognitive and neural representations? See our paper in @PNASNews w/ @mafeilong @MatteoVdOC @samnastase @haxbylab @IGobbini 🧵

2

27

92

1

8

20

Sam Nastase

@samnastase

4 years

All data used in this project are publicly available via @OpenNeuroOrg and @datalad as part of the "Narratives" data collection. Narratives on OpenNeuro: Narratives on DataLad:

1

4

22

Sam Nastase

@samnastase

1 year

I genuinely hope that when the shock wears off, some people will have a change of heart and come around—but I think that's probably optimistic. Colonial violence begets violence! This is a fundamental lesson of history.

1

0

21

Sam Nastase

@samnastase

2 years

Check out @zaidzada_ 's poster on our recent (not-yet-published!) work using large language models to capture the linguistic content of brain-to-brain coupling during face-to-face conversations between ECoG patients! P-3.20 @CogCompNeuro #ccneuro22

Zaid Zada

@zaidzada_

2 years

Come check out my poster on modeling brain-to-brain communication via deep language models in natural conversations @CogCompNeuro Saturday night. #ccneuro22

3

21

88

1

5

23

Sam Nastase

@samnastase

1 year

For any of my colleagues who want to learn more, here are two short documentaries I recommend: Gaza Fights for Freedom (2019): The Viewing Booth (2022): Both films obviously come with many trigger warnings.

2

0

20

Sam Nastase

@samnastase

4 years

0

22

Sam Nastase

@samnastase

5 years

If you want someone to blame, you can blame the department and university leadership, as well as the abusers themselves. That kind of shit doesn't happen in a vacuum; it takes a village to enable and shelter abusers for over a decade.

1

21

Sam Nastase

@samnastase

6 years

Creating @datalad datasets and analyzing them with "datalad run"... I'm going mad with power!!! #versioncontrol #reproducibility @yarikoptic

0

3

22

Sam Nastase

@samnastase

6 years

New paper with @mafeilong , @swaroopgj , and @haxbylab ! Hyperalignment factors out idiosyncrasies in functional–anatomical correspondence, revealing individual differences in fine-grained functional architecture otherwise obscured by anatomical normalization

Feilong Ma

@mafeilong

6 years

Our new paper shows that individuals differ reliably in fine-grained cortical functional architecture, and resolving topographic idiosyncrasies using hyperalignment is key to reveal this. Thanks to @samnastase @swaroopgj @haxbylab and colleagues/reviewers!

1

21

46

0

8

21

Sam Nastase

@samnastase

5 years

You can find the rest of my open reviews at . Earnest, well-intentioned scientific criticism should not be received with animosity.

Nikolaus Kriegeskorte

@KriegeskorteLab

5 years

@pezlarson @TimKietzmann @KordingLab @INM7_ISN Check out @samnastase ’s recent open review of one of our papers or any or my open reviews on . Open reviews should be critical and constructive.

0

2

1

0

20

Sam Nastase

@samnastase

4 years

If you're fixated on the president's theatrics during this nationwide uprising against police terror, then the state propaganda (via corporate media) is working. Redirecting discontent toward an individual or figurehead is how institutions avoid having to make structural changes.

2

3

21

Sam Nastase

@samnastase

5 years

Those women went through hell and are fighting tooth and nail to do the restorative work that @dartmouth leadership didn't (and still doesn't!) have the moral clarity to do on its own.

1

19

Sam Nastase

@samnastase

1 year

If all you can do right now is use your platform to dehumanize Palestinians and attack anyone who speaks up on their behalf, you are quite literally manufacturing consent for genocide—and you'll have to carry that weight forever.

1

2

19

Sam Nastase

@samnastase

4 years

This release comprises 345 subjects, 891 functional scans, and 27 diverse stories totaling ~4.6 hours of unique stimuli (~43,000 words). This amounts to ~6.4 days of fMRI data with accompanying stimuli comprising ~1.4 million words—all organized according to the @BIDSstandard .

1

2

19

Sam Nastase

@samnastase

5 years

@neuroecology I think the AFNI bootcamp handouts are still the highest form of slide aesthetic

0

20

Sam Nastase

@samnastase

5 years

New commentary on the widespread cortical representation of observed actions (as revealed by dynamic, naturalistic stimuli)—even in putative "form" areas like ventral temporal cortex:

Ida Gobbini

@IGobbini

5 years

New opinion piece on the organization of the ventral temporal cortex with Jim Haxby and @samnastase

0

17

44

1

3

20

Sam Nastase

@samnastase

5 years

This work is an extension of connectivity hyperalignment from @haxbylab ( which also works for RS!). Potential benefits for both RS and naturalistic designs—namely, better accommodate individual differences in fine-grained functional neuroanatomy.

A computational model of shared fine-scale structure in the human connectome

Author summary Resting state fMRI has become a ubiquitous tool for measuring connectivity in normal and diseased brains. Current dominant models of connectivity are based on coarse-scale connectivity...

journals.plos.org

Simon Eickhoff

@INM7_ISN

5 years

Naturalistic stimuli (movies, stories) keep subjects engaged, reduce motion & variability. Excellent for fMRI! But RS is much easier to pool & reuse? Work by @samnastase / @HassonLab suggests hyperalignment & shared response models can overcome this

1

30

61

0

4

19

Sam Nastase

@samnastase

3 years

Thank you @ingridjohnsrude ! Really great group and lots of good questions—thanks for the invitation @aysha_motala !

Ingrid Johnsrude

@ingridjohnsrude

3 years

@samnastase gave a scholarly, data-rich, thought-provoking, utterly delightful talk on direct perception and naturalistic methods in cognitive neuroscience in our @BMI_WesternU #coffeetalk this morning. Whoever gets to hire this guy is very lucky!

1

14

1

0

19

Sam Nastase

@samnastase

3 years

Open review of @biorxivpreprint by @H_Lee_Masson @leylaisi () using encoding models to capture cortical responses to social interaction in two movie-watching fMRI datasets—on @PubPeer : ; and on my website:

1

3

19

Sam Nastase

@samnastase

4 months

Join Eshin's new lab at UCSD! Eshin is one of my science heroes—cannot recommend this opportunity highly enough 🚀

Eshin Jolly

@Eshjolly

4 months

Some very exciting news... after many, many, winter beards, I'm overjoyed to be joining sunny @UCSDPsychology as an Asst Prof in Jan 2025! Looking to recruit folks interested in doing some fun science, while building some cool (and useful) toys! Details:

37

39

296

0

19