Paul Scotti (currently at ICLR!)
@humanscotti
Followers
1,439
Following
959
Media
42
Statuses
439
Head of Neuroimaging @StabilityAI , leading the @MedARC_AI Neuroimaging & AI Lab | Collaborating w/ @PrincetonNeuro @ptoncompmemlab
Princeton, NJ
Joined July 2009
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
América
• 544457 Tweets
Switch
• 240024 Tweets
Knicks
• 102178 Tweets
Brunson
• 74009 Tweets
ドカ食い
• 56804 Tweets
महाराणा प्रताप
• 51975 Tweets
Chivas
• 50709 Tweets
もちづきさん
• 49620 Tweets
MAURO AL 9009
• 48231 Tweets
Pacers
• 48157 Tweets
WINTER AT POLO ART OF ESCAPE
• 45545 Tweets
#jjk259
• 42307 Tweets
Cavani
• 37159 Tweets
#LingOrmCh3FanGreet
• 37102 Tweets
Choso
• 35488 Tweets
LINGORM TUKTUK GOGO
• 32163 Tweets
Millonarios
• 30488 Tweets
宇野昌磨
• 30416 Tweets
現役引退
• 27333 Tweets
Yuji
• 26822 Tweets
アイスの日
• 25144 Tweets
Gege
• 23241 Tweets
Josh Hart
• 20054 Tweets
Oilers
• 19005 Tweets
Calleri
• 16174 Tweets
Vancouver
• 15025 Tweets
血糖値スパイク
• 14608 Tweets
タキオン
• 14119 Tweets
#MaharanaPratap
• 13882 Tweets
Pachuca
• 13683 Tweets
Ceci
• 13679 Tweets
#Canucks
• 13496 Tweets
ヴィルシーナ
• 10977 Tweets
女性向け
• 10565 Tweets
結婚詐欺
• 10472 Tweets
Pinned Tweet
🧠👁️Our MindEye2 preprint is out!
We reconstruct seen images from fMRI brain activity using only 1 hour of training data.
This is possible by first pretraining a shared-subject model using other people's data, then fine-tuning on a held-out subject with only 1 hour of data.🧵
15
77
370
Announcing 🧠👁️ MindEye!
Our state-of-the-art fMRI-to-image approach that retrieves & reconstructs images from brain activity, is out as a preprint!
MindEye takes human brain activity as input and outputs reconstructed images like these.
Project page:
16
124
555
Happy to share that I am now Head of Neuroimaging at
@StabilityAI
!
My role entails leading the
@MedARC_AI
Neuroimaging & AI Lab:
This lab at MedARC changes the usual way research labs work...
🧵 1/n
11
23
191
@ashleyruba
That's awesome and very impressive! i want to add: some never get published, for reasons out of their control (bc of unclear data, mental health issues, bad work environment) and they and their work matter too. Unfortunately staying in academia can be a lot harder in these cases.
2
1
105
Very happy to share that in April I will begin a postdoc at
@PrincetonNeuro
with
@ptoncompmemlab
to work on projects involving real-time fMRI! (Have already been familiarizing myself with rt-cloud which simplifies the complexities of setting up real-time, preprint below)
New preprint! RT-Cloud: A Cloud-based Software Framework to Simplify and Standardize Real-Time fMRI 🧵
1
11
28
8
0
73
Now out in JEP:General! Visual working memory items drift apart (aka repulse/differentiate) due to active, not passive, maintenance.
(half year late to share bc I was waiting for publication pdf, but realized I'm not allowed to share non-paywalled pdf🤷)
1
9
69
1st project I started in my PhD, now in AP&P!
Summary: People memorized lots of colored real-world objects, unaware that many objects were same color. When tested on object colors, found swap & shift errors in relation to most commonly studied color.
1/
1
11
61
BrainLM is a foundation fMRI model which can be fine-tuned for all sorts of specialized downstream tasks.
We will have a public discussion on this paper this Fri 1pm ET on Zoom for
@MedARC_AI
NeuroAI journal club, feel free to join!
How to join:
Introducing BrainLM 🧠🤖the first foundation model for
#fMRI
analysis trained on 6,700 hours of brain activity data! Fine-tune for specialized tasks or leverage zero-shot inference capabilities!
@WuTsaiYale
@YaleCompsci
@YaleCBB
@YaleMed
6
107
375
1
11
52
@praxishabitus
... neither the inspector nor both versions of the woman are wearing the masks correctly (over their nose and mouth)?
1
2
45
So grateful that I was able to work in such amazingly supportive and intellectually rewarding labs during my PhD -- I'll always fondly remember these grad school years. Thanks
@juliedgolomb
and
@leberandy
! ❤️
Congratulations to newly minted Dr Paul Scotti, PhD!
@humanscotti
So very proud to have had you as a member of
@GolombLab
and
@leberatory
at
@OhioStatePsych
, and can’t wait to see your cool research continue as a postdoc with
@kennethanorman
!
0
3
67
1
0
44
MindEye really shines in its retrieval capabilities. It can retrieve the exact original image even among highly similar candidates.
This shows that the brain embeddings it creates retain fine-grained, image-specific information.
1
3
32
Preprint now available for our work on improving inverted encoding models (used for reconstructing stimuli from brain activity)! Just in time for
#VSS2021
see poster C91 today at 4pm EST
@juliedgolomb
@JiagengChen2
2
9
30
New paper w/ Dr Ashleigh Maxcey! We simultaneously measured forgetting via directed forgetting (subj tries to forget item) and recognition-induced forgetting (subj recalls a related item such that target item gets forgotten). Which is a stronger effect?
1
3
25
Starting the year off with good news, my first first-author paper is published! Recognition-induced forgetting of schematically related pictures. w/ Laura Janakiefski and Dr. Ashleigh Maxcey (non-paywalled pdf at )
1
0
23
MindEye achieves state-of-the-art performance across both low- and high-level image metrics (Table 1 in the preprint). Here are a few side-by-side comparisons with other papers.
3
2
24
Happy to share MindEye got accepted for spotlight poster at
#NeurIPS2023
! See you in New Orleans!
Updated camera ready paper is also now live on arxiv:
Has various revisions including new appendix figures and more comprehensive references to other work.
Announcement!📢
We're excited to share that our first paper since our public launch, MindEye, has been accepted as a *spotlight* at
#NeurIPS2023
!
The updated version of our paper with additional details/exps is now up:
Website:
12
44
198
4
3
24
1.8 billion regressions, testing fMRI brain predictivity using hundreds of pretrained models 👀
What models did best? Transformers or CNNs? Contrastive or non-contrastive?
Seems those factors didn't matter -- it's image diversity during training that mattered
1/ Excited to share a new NeuroAI preprint: “What can 1.8 billion regressions tell us about the pressures shaping high-level visual representation in brains and machines?” Led by Colin Conwell, with
@_jacobprince_
,
@cvnlab
, and
@grez72
.
9
135
407
0
3
24
We scale this up to large-scale image databases like LAION-5B!
MindEye can look through the billions of images in LAION-5B and find the image it thinks was originally presented.
This means MindEye can output realistic images from brain activity without any generative model.
1
1
23
Presenting
#CNS2021
poster (D63) this morning on improvements to inverted encoding models. Work done w/
@juliedgolomb
and
@JiagengChen2
. Stop by and say hi!
Link to poster:
0
6
23
Congratulations!! amazingly well done job to the best wifey ❤️ (this cake has a hidden T among the Ls)
Huge congratulations to newly minted Ph.D., Dr. Yoolim Hong, who did a fantastic job defending her dissertation today!!! 🎉🎓🍾🥂📄😁👏🏻👏🏼👏🏾
@HongYoolim
@leberandy
4
0
42
1
0
23
#Neurohack19
has been an amazing experience, learned so much and met some truly great people. For the hacking, our group merged
@pycortex
w
@neurosynth
and an anatomical atlas to functionally map the brain. Check it out and the other great projects live-tweeted by
@kirstie_j
!
0
4
22
Very thankful for all the interest in our work at
#NeurIPS2023
! Great to chat with so many incredible people!
With co-presenters
@iScienceLuvr
and
@apeoffire
0
2
22
Incredible work from
@HuzeYann
team dominating the Algonauts 2023 challenge to predict fMRI activity from images.
Largest factor for their success seems to be how they used the previous 32 seen images (~2 min of past images) as context to predict the fMRI of current image!
0/. Brain is not ViT. we scored 70.8 in the Algonauts 2023 visual brain competition, w/o ensemble we can do 66.8 score. other teams (including me in the past) are struggling with 60.
“memory” is the secret.
paper:
code&webpage:
4
50
196
0
6
22
We are very fortunate to have learned from prior pioneering work done by people like
@OzceFurkan
,
@gallantlab
,
@NishimotoShinji
,
@ZijiaoC
,
@HelenJuanZhou
,
@zijin76
,
@amykooz
,
@mertrory
,
@M_ostlyHarmless
,
@alex_ander
and others. Check their papers if you found this interesting!
1
1
20
Excited to share that I am leading the fMRI project for MedARC! Reconstructing visual perception in real-time and building foundation fMRI models. This research will be open such that anyone can join our efforts. Excited to leverage the power of open science for medical research
Announcing the launch of the MedARC!
MedARC is a novel, open, & collaborative approach to medical AI research.
It was created to develop large-scale AI models for medicine & build interdisciplinary teams to address clinical needs. (1/10)
12
128
500
3
2
20
MindEye was openly developed through volunteer contributions in the
@MedARC_AI
Discord server. We are actively working on more related projects, so please join us!
MedARC homepage:
Lab Notion:
3
0
19
Want to keep up-to-date with NeuroAI literature? I'm hosting a Trends in NeuroAI reading group on Zoom!
Tmrw 11am ET we're discussing
@Meta
MEG-to-image paper.
2 weeks ago Junbeom presented his 4d fMRI Swin transformers paper.
@MedARC_AI
@StabilityAI
Today we're sharing new research that brings us one step closer to real-time decoding of image perception from brain activity.
Using MEG, this AI system can decode the unfolding of visual representations in the brain with an unprecedented temporal resolution.
More details ⬇️
64
501
2K
1
3
19
@EMostaque
@chrypnotoad
@StabilityAI
We've got some really exciting experiments running in the coming months using fMRI and ECoG, and then plans for multimodality + hopefully applying to new BCI tech😄
2
4
20
.
@OzceFurkan
released a preprint last week showing the best performance so far in doing fMRI - to - Image reconstructions
He will be giving a public talk about his reconstruction approach tomorrow at 11am EDT in the MedARC discord!
1
2
18
Great work assessing these methods--we should all be aware that the current NSD reconstruction papers are limited to staying within a specific image distribution (specifically images like those in MS-COCO)
But I want to clarify that "semantic leak" paints the wrong idea--
Recent studies have shown photorealistic reconstructions from fMRI data using CLIP/diffusion models and the NSD dataset (Left). We evaluated the methods on the Deeprecon dataset (Shen+ 2017/19) with added annotations, but found the results not so impressive ( Right).
4
52
127
1
4
18
For recons we train a 940M param MLP to map fMRI voxels to CLIP image space, then use a diffusion prior trained from scratch to align modalities.
This means we can reconstruct using any pre-trained generative model, without fine-tuning, that accepts CLIP image latents as inputs.
1
1
16
Happy to share a preprint with
@aj_collegio
and
@GWUAttentionCo1
: Takeaway finding: task-irrelevant high-level semantic information constrains object-based attentional guidance more than low-level Gestalt (boundary) information.
1
2
15
Discussing the winning Algonauts 2023 approach this Friday 9am EDT on Zoom for
@MedARC_AI
NeuroAI journal club
Lead author
@HuzeYann
will join us to answer questions!
How to join:
0/. Brain is not ViT. we scored 70.8 in the Algonauts 2023 visual brain competition, w/o ensemble we can do 66.8 score. other teams (including me in the past) are struggling with 60.
“memory” is the secret.
paper:
code&webpage:
4
50
196
0
3
15
Training a neural network to predict brain responses to images isn't substituting one black box for another
Work showing you can dissect the network to uncover and visualize fine-grained (and voxel-specific) functional selectivity, going beyond coarse categories
1/ 🧠🔍 New research alert! We've trained a neural network to predict brain responses to images, and then “dissected” the network to examine the selectivity of spatial properties across high-level visual areas. Discover more about our work:
3
56
205
1
0
15
To ensure objects are in right spatial locations, colors mostly correct, etc. we use img2img with a separate autoencoder that maps brain activity to Stable Diffusion’s variational autoencoder.
This produces blurry reconstructions which we use to initialize the diffusion process.
1
0
14
Open to anyone: Zoom tomorrow 9am EDT to discuss this paper that reconstructs video from fMRI!
First author
@ZijiaoC
will join us to answer questions!
Part of the
@MedARC_AI
fMRI journal club series
🧵🧠 We're witnessing incredible scientific progress in image & text reconstruction from fMRI nowadays. But what about reconstructing video from fMRI? Allow me to introduce our recent preprint: Mind-Video
38
204
611
1
4
15
To those who couldn't attend last month's OnNeuro meeting, where we talked about expectation's role on perceptual decision-making, the recording is now available on YT. Thanks to
@summerfieldlab
and
@flodlan
for writing the paper!
0
4
13
Looking forward to reading through these chapters on all the latest prominent work in visual memory. Grateful I could make a modest contribution to the induced forgetting chapter w/
@ashleigh_maxcey
. Great job to
@WilmaBainbridge
and
@timothyfbrady
for editing & organizing!
Learn about the latest & greatest research on visual memory in our new book! So many thanks to
@timothyfbrady
on inviting me on this editorial adventure. The stars here are really our fantastic chapter contributors: 1/4
2
24
96
2
1
15
Love the idea behind this. Why wait for conferences to share your work? Why wait to be invited to present somewhere? Why force your talk to fit within the required talk duration?
We are excited to announce the Distributed Working Memory Series (
#DWMS
). The DWMS offers a platform for researchers on working memory to present their work to the working-memory community, from everywhere, at any time, without invitation. Thread: 1/8.
2
70
91
0
2
12
Amazing work using ECoG data from epilepsy patients during natural conversation shows how LLM embeddings can be used to capture word-by-word neural alignment!
Linguistic representations from speaker’s brain (before they actually start speaking) re-emerge in the listener’s brain
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"
6
81
264
1
1
14
100% Transparent Volunteer-Driven Science
I was very pleasantly surprised by all the benefits of keeping MindEye development public & open-source and inviting crowd-sourced collaboration; been eye-opening to adopt this open research approach w/ amazing people like
@iScienceLuvr
This project was openly developed via volunteer contributions in the
@MedARC_AI
Discord server and GitHub.
Open-source/decentralized research initiatives have been successful in AI (
@AiEleuther
,
@laion_ai
,
@openbioml
,
@ml_collective
) & our project further demonstrates that!
2
6
39
1
1
13
The lab is remote, open-source, and open to the public to join. There's tons of potential here to rethink what a research lab can be, and to benefit from crowd-sourced intelligence.
@MedARC_AI
and initiatives like
@AiEleuther
@laion_ai
@ml_collective
all show how this can work.
1
2
13
New paper out in
@NatureHumBehav
by
@aj_collegio
,
@joecool890
, myself, and Dr. Sarah Shomstein! Even if objects are of identical retinal size, if you know that one object is physically smaller, then you allocate more attention to the smaller object.
Sizzling-hot new paper in
@NatureHumBehav
from Sarah Shomstein’s lab
@GWUAttentionCo1
, on which
@humanscotti
serves as a co-author. Fancy outlet for your first publication, Paul!
0
1
3
0
2
13
wee ooo wee ooo wee ooo
@humanscotti
closing out the final talk session of the day! Come hear about how competition between between similar visual working memory items underlies repulsion.
@HongYoolim
@leberandy
@juliedgolomb
#OPAM2019
0
0
10
0
0
13
This work wouldn’t have been possible without
@abanerjee99
(co-first author),
@iScienceLuvr
(MedARC CEO),
@ptoncompmemlab
,
@StabilityAI
(supplied necessary computational resources), and of course the Natural Scenes Dataset (e.g.,
@Emsquem
,
@cvnlab
, Naselaris lab)
2
0
12
Please help manage me and others in a great lab and conduct the coolest research 👍
@GolombLab
is looking for our next lab manager! Job ad below. Know any awesome undergrads who want additional research experience in cog neuro before applying to PhD programs? Send them our way!
1
20
16
0
1
10
My mom sent me a link to this article today because she was concerned for me. Sigh
I know that fMRI research has serious replicability problems that need to be addressed, but news headlines like this are deeply harmful to our field, both in terms of public understanding of the work we do and in the way that other neuroscientists see us.
35
87
504
1
0
9
Both of my incredibly savvy advisors are now on Twitter!
@leberandy
(lab:
@leberatory
) and
@juliedgolomb
(lab:
@GolombLab
)!
0
0
10
Wrote a blog post with
@iScienceLuvr
on MindEye for the
@StabilityAI
research hub 🥳
Thanks to Stability AI &
@EMostaque
for their generosity in supporting neuroAI open research
Presenting
@MedARC_AI
's inaugural research paper, MindEye: an innovative method to retrieve & reconstruct images from fMRI brain scans.
Led by
@iScienceLuvr
, MedARC is Stability AI's fully supported medical research hub. Stay tuned for more amazing work!
4
30
100
2
3
9
Took part in this last year, I highly recommend it! You'll learn so many things about neuroimaging and best research practices
We are now accepting applications to participate in NeuroHackademy 2020:
0
81
124
0
0
9
0
0
9
sometimes it's best for your sanity to just let the model finish training and check on it at the end
1
0
9
If a subject's train data did not include any images of pizza, the model will not do as well generalizing to pizza during test, and that's normal and expected!
If we truly want recons across any image distribution, this is a deeper issue about needing larger, broader datasets
1
0
8
@chazfirestone
Maybe people just love reading about psychology so much that the store wanted to highlight the psychology (sub)section from the general science section!
0
0
7
Thanks so much for the support!! Please check out if you want to help contribute to this idea of virtual webinars 👍
.
@OnNeuro
is a great project by
@humanscotti
to host and aggregate virtual talks in neuroscience
2
44
102
0
0
7
More proof of the potential research applications of mapping brains to pre-trained multimodal models
After mapping fMRI responses to CLIP space, authors train a model across-subjects to map to a Shared Decodable Concept space to localize interpretable semantic concepts
[1/8] Alona Fyshe will present some of our recent NeuroAI work in her keynote at
@CanadianAI
today. We approach this from various directions, but today I will highlight our most recently submitted work on identifying shared decodable concepts in the brain.
1
16
45
1
1
7
@phc_phd
hah my sentence "you can implement our python package in one line of code" got captioned as "one line of coke"
1
0
7
Hosting an online project pitch day for
@MedARC_AI
next Mon 11am ET if anyone has NeuroAI project ideas and interested in doing open research together!
Also we have a journal club this Fri 11am ET where
@ReeseKneeland
is presenting on brain-optimized inference for fMRI recons!
1
1
8
The goal in reconstructing seen images from brain activity isn't to return the original image. Recons reflect our individual mental representations, meaning they are theoretically & clinically informative.
New work shows recons for visual illusions preserve the illusory percept.
Visual illusions suggest that what we see is not always a direct reflection of the world, but rather a constructed representation. Is it possible to generate images mirroring illusory experiences directly from brain activity?
Check our preprint!
2
20
79
0
3
7
Flying in today for NeurIPS (first time at an ML conference!)
Looking forward to meeting new folks, send me a DM if you'd like to chat
I will be at
#NeurIPS2023
next week!
We will be presenting
@MedARC_AI
's spotlight paper "Reconstructing the Mind's Eye" (also my first NeurIPS paper!), be sure to come on Thurs and check it out!
Also happy to chat about medical AI, diffusion/generative models, LLMs, etc.
3
6
77
0
2
7
I am extremely grateful for this opportunity. Huge thank you to
@iScienceLuvr
(CEO of MedARC) for his leadership in creating MedARC and
@StabilityAI
for being willing to invest in open neuroscience research.
1
1
7
Wow this is absurd. The majority of authors weren't able to provide raw data even to the editor of the journal their paper was submitted to
My editorial paper entitled, "No raw data, no science: another possible source of the reproducibility crisis". When I feel the results are too beautiful, I requested raw data before sending it out for review. 40 out of 41 did not send me the right data.
56
2K
4K
2
1
7
I am happy to say I'm not even leaving Princeton since I can work remotely for Stability; I'll become a visiting research scientist in
@ptoncompmemlab
to continue our existing projects while establishing new collaborations between MedARC and Princeton.
1
0
7
@acdeeplearning1
@arankomatsuzaki
in theory it should work off the bat! since similar patterns of activation for visual imagery occur for perception of same stimuli. Also the dataset we used will be releasing a mental imagery variant soon so we can immediately test this when it gets released
0
0
6
@MattiaRigotti
@joeyzyzyzy
I don't understand, did Neuralink say that they were the first to demonstrate this? I thought its a cool demo of their current bci implant
1
0
5
It seems weird to train a model on ImageNet images and then expect it to generalize during testing using images of artificial geometric shapes, esp. when the comparison methods used pretrained frozen models that assume realistic image outputs (e.g., Stable/Versatile Diffusion)
1
0
6
Don't miss out on this! Incredible that it'll be free and available to everyone this year.
NeuroHackademy 2020 is going online! And because of the online/remote format, we are opening it up to anyone who wants to participate. Please sign up here:
11
199
376
1
0
4
Here's a video walkthrough of the VSS poster if curious--note that it's only 6 minutes long so for more info/questions please get in touch :)
0
0
5
That moment when you present a poster later in the week using an approach the speaker has just invalidated
Interesting stuff from Tim Brady saying that we need to use perceptually linear space when we study memory. The meta-analysis bit is especially impressing
#VSS2018
1
5
16
0
0
5
To any cognitive researchers, I am holding monthly online journal clubs to discuss current work in the field -- think of it as a way to get outside of your academic bubble. Check out the site and maybe join us next time:
0
1
4
@schulzb589
would love to test this if we had data for it -- we hope to extend this to mental imagery and at that point it'd be very cool to test the difference in reconstructions for aphantasia population
0
0
4
so cool to see a video talking about our paper! you did a really great job explaining and framing everything!
@humanscotti
,
@HongYoolim
,
@leberandy
, and
@juliedgolomb
in the
@GolombLab
recently published a super cool paper about how representations in working memory actively repel each other. Wild stuff. Check out my video description:
1
2
6
0
0
4
VWM for item orientation is biased by mean of other items (which were also tested). Bias towards mean and precision increased as range of orientations narrowed. Supports hierarchical encoding framework: narrower orientation range makes ensemble info easier to 1/5
1
0
4
My whole life I looked up to Zimbardo... how could you do this Phil :(
The Stanford Prison Experiment was massively influential. We just learned it was a fraud. via
@voxdotcom
0
0
1
0
0
4
Congrats on finally getting this published! Important work to improve the conceptualization and diagnosis of misophonia
🆕📊🎉 Howdy
#AcademicTwitter
! Made an account to share my new
#Misophonia
research (with
@leberandy
and
@z_saygin
), published today in
#JournalOfClinicalPsychology
#JCLP
{1/7}
4
2
25
0
0
4
Thanks so much to Stephen Adamo for giving his talk at this month's OnNeuro meeting today! He talks about how he has used basic research on undergraduates to better inform how professional radiologists search for tumors.
0
1
4
Check out the entries! So many interesting illusions, I particularly enjoyed the "Once Upon A Time" entry.
0
0
4
@HuzeYann
Ben Kubit in Norman Lab is actually doing a study like this! He plays catchy short tunes many times to participants and then tries to decode impromptu earworms via fMRI
1
0
4
Also a massive thank you to my postdoc advisor
@kennethanorman
for taking me into his lab, allowing me the freedom to pursue these crazy projects, and for being such a caring, hardworking, and insightful mentor throughout our time together.
1
1
4
@alex_ander
@kencan7749
@ykamit
Potentially another case would be framing identification as retrieval (to clarify the original stimulus not needing to be in the set, and tying to the same terminology used in computer vision)
bc id argue retrieval approaches serve unique goals beyond reconstruction..
1
0
4
Cool to see this work using THINGS dataset to obtain image reconstructions from MEG using similar approach as fMRI-to-image papers
Today we're sharing new research that brings us one step closer to real-time decoding of image perception from brain activity.
Using MEG, this AI system can decode the unfolding of visual representations in the brain with an unprecedented temporal resolution.
More details ⬇️
64
501
2K
0
0
4
Haha this seems like a fun idea! Seems like a great way to ensure lots of varied discussion points to a journal club. (Now add a hidden role who needs to get away with stating lies about the paper and turn it into a game of Werewolf)
Really cool/fun small group method for reading papers I’d never heard about: role-playing! Love the “archaeologist” role
(Would be great for a journal club too!)
8
38
147
0
1
4
Interesting preprint summarizing representational alignment across ML / neuro / psych and proposing a common language across disciplines to help cross-field communication
🧵🎉 Our new preprint is up, and we’d love your feedback! We're "Getting Aligned on Representational Alignment" - the degree to which internal representations of different (biological & artificial) information processing systems agree. 🧠🤖🔬🔍
#CognitiveScience
#Neuroscience
#AI
7
119
447
0
0
4
You can apply our IEM procedure to your neuroimaging data in just 1 line of Python code using our package!
1
0
4
Attention and WM fluctuate the same way, lapsing together, adding evidence to notion that both share a common cognitive resource. Not due to task-general fluctuations like alertness or how engaged subject was.
Really excited to have this work from my post-doc out! Check out the paper here I've also posted all the data and code for the task and analyses and
0
26
111
1
0
3
(This marks the start of my goal to read one academic paper every morning, and then post my summary and related questions on Twitter. Wish me luck!)
0
0
3
We're just one step away from captcha bot solvers. Soon we'll have to transcribe latex equations to prove we're human 😢
I got to try GPT-4's multimodal capabilities and it's quite impressive! A quick thread of examples...
Let's start out with solving a CAPTCHA, no big deal
70
496
3K
0
0
3
@rudzinskimaciej
@EMostaque
@chrypnotoad
@StabilityAI
yes we are collaborating on EEG-to-image work currently too
1
0
3
What's a swap error? Incorrectly reporting that an object's original color was the most common color. You don't need to understand the modeling in the paper to appreciate this, these visible bumps in the response histograms show the swapping:
2/
2
0
3
@jsaito25
in python its simple! im less familiar with R but i imagine R must also have some way to do this kind of indexing?
1
0
2
@harrison_ritz
@psychopy
I just made the switch to use psychopy instead of psychtoolbox and same experience! Should've switched earlier
0
0
3
I'm so very honored to have been invited to present for OnNeuro!
Yesterday, Paul Scotti (
@humanscotti
) discussed an article from
@parkinsoncm
,
@ThaliaWheatley
, and
@akleinb
, entitled "Similar neural responses predict friendship". Thanks to all who attended!
0
0
5
1
0
3
@JuliaJYan
@PhDForum
I've tried both Mendeley and EndNote and always have a handful of incorrect references at the end (formatting errors or missing info). I have to manually check every ref to trust my bibliography before submissions. So I would love to have an automated reference checking system!
1
0
3
RIF produced stronger effect than DF. Also found DF is susceptible to response confusion (I remember object was associated w/ Forget instruction, should I report "New"?) and that nearly half of subj may have disregarded the Forget cues. RIF>DF holds after controlling for these.
1
0
3
@HaoTingW713
That would make sense if the project is a couple years old maybe, but if the authors just submitted to a journal and still can't provide either organized or unorganized data (at risk of rejection otherwise), then how did they analyse the data in the first place?
1
0
3