Melissa Franch, PhD
@mfranch.bsky.social
Postdoc in the Hayden lab at Baylor College of Medicine studying neural computations of natural language & communication in humans. Sister to someone with autism. F32 NIDCD Fellow | Autism Research Institute funded | Simons F2F | she/her. melissafranch.com
Pinned
Melissa Franch, PhD
@mfranch.bsky.social
· Mar 20
I’m excited to share my newest work with @benhayden.bsky.social, and the work I’m most proud of to date, on characterizing semantic coding in single-neuron hippocampal activity in patients with autism during natural language comprehension!
www.biorxiv.org/content/10.6...
www.biorxiv.org/content/10.6...
Reposted by Melissa Franch, PhD
New work out today!
The result of a 15 year project to offer genuine linking hypotheses between linguistics and neuroscience: a revised research program, and its first result - a novel neural binding operation ('Meld') capturing core properties of natural language syntax 🧵
arxiv.org/abs/2609.14384
The result of a 15 year project to offer genuine linking hypotheses between linguistics and neuroscience: a revised research program, and its first result - a novel neural binding operation ('Meld') capturing core properties of natural language syntax 🧵
arxiv.org/abs/2609.14384
Formal Properties of Language as Constraints on Neural Dynamics
What must a neural system be capable of to implement language? Current research annotates stimuli with linguistic variables and tests which electrodes, voxels, or language-model layers predict neural ...
arxiv.org
September 15, 2026 at 4:04 AM
New work out today!
The result of a 15 year project to offer genuine linking hypotheses between linguistics and neuroscience: a revised research program, and its first result - a novel neural binding operation ('Meld') capturing core properties of natural language syntax 🧵
arxiv.org/abs/2609.14384
The result of a 15 year project to offer genuine linking hypotheses between linguistics and neuroscience: a revised research program, and its first result - a novel neural binding operation ('Meld') capturing core properties of natural language syntax 🧵
arxiv.org/abs/2609.14384
Reposted by Melissa Franch, PhD
Our review on superposition in AI & brains is published in
@nature.com #Machine #Intelligence 🌐
Neural networks encode concepts in superposition. We can recover them in 3 steps: identifiability, compressed sensing, interpretability.
W/ @david-klindt.bsky.social O'Neil Reizinger & Maurer 🌟
@nature.com #Machine #Intelligence 🌐
Neural networks encode concepts in superposition. We can recover them in 3 steps: identifiability, compressed sensing, interpretability.
W/ @david-klindt.bsky.social O'Neil Reizinger & Maurer 🌟
July 22, 2026 at 5:29 PM
Our review on superposition in AI & brains is published in
@nature.com #Machine #Intelligence 🌐
Neural networks encode concepts in superposition. We can recover them in 3 steps: identifiability, compressed sensing, interpretability.
W/ @david-klindt.bsky.social O'Neil Reizinger & Maurer 🌟
@nature.com #Machine #Intelligence 🌐
Neural networks encode concepts in superposition. We can recover them in 3 steps: identifiability, compressed sensing, interpretability.
W/ @david-klindt.bsky.social O'Neil Reizinger & Maurer 🌟
Reposted by Melissa Franch, PhD
The Grand Plan for brain & mind research is in terrific hands with the researchers at Baylor College of Medicine!
From an excellent panel discussion across silos to many insightful conversations with PIs and trainees, I'm a bit blown away by what they are up to here. Impactful & exciting.
From an excellent panel discussion across silos to many insightful conversations with PIs and trainees, I'm a bit blown away by what they are up to here. Impactful & exciting.
September 11, 2026 at 1:19 PM
The Grand Plan for brain & mind research is in terrific hands with the researchers at Baylor College of Medicine!
From an excellent panel discussion across silos to many insightful conversations with PIs and trainees, I'm a bit blown away by what they are up to here. Impactful & exciting.
From an excellent panel discussion across silos to many insightful conversations with PIs and trainees, I'm a bit blown away by what they are up to here. Impactful & exciting.
Reposted by Melissa Franch, PhD
And of course, a great source on why cell types like "place cell" aren't a natural kind:
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
The inevitability and superfluousness of cell types in spatial cognition
Discoveries of functional cell types, exemplified by the cataloging of spatial cells in the hippocampal formation, are heralded as scientific breakthroughs. We question whether the identification of c...
www.biorxiv.org
September 9, 2026 at 3:34 PM
And of course, a great source on why cell types like "place cell" aren't a natural kind:
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
Reposted by Melissa Franch, PhD
And in some patients we measured gaze as well. We find that gaze has its own maps, and those are also semi-orthogonalized. So in debates about whether primate hippocampus does gaze or place, the answer is “both, and their codes are rotated”.
September 9, 2026 at 3:34 PM
And in some patients we measured gaze as well. We find that gaze has its own maps, and those are also semi-orthogonalized. So in debates about whether primate hippocampus does gaze or place, the answer is “both, and their codes are rotated”.
Reposted by Melissa Franch, PhD
Autism has long been linked to difficulties with Theory of Mind.
But what exactly is driving these differences? And what happens in the brain?
In our new preprint, we show that adaptive mentalization might be key.
A 🧵 #CogSci #CogNeuro
osf.io/preprints/ps...
But what exactly is driving these differences? And what happens in the brain?
In our new preprint, we show that adaptive mentalization might be key.
A 🧵 #CogSci #CogNeuro
osf.io/preprints/ps...
September 9, 2026 at 2:20 PM
Autism has long been linked to difficulties with Theory of Mind.
But what exactly is driving these differences? And what happens in the brain?
In our new preprint, we show that adaptive mentalization might be key.
A 🧵 #CogSci #CogNeuro
osf.io/preprints/ps...
But what exactly is driving these differences? And what happens in the brain?
In our new preprint, we show that adaptive mentalization might be key.
A 🧵 #CogSci #CogNeuro
osf.io/preprints/ps...
Reposted by Melissa Franch, PhD
New preprint! Posterior cingulate is often linked to memory in humans but decisions in monkeys yet there's little consensus about its functional neuroanatomy. We recorded LFPs+single units in humans during matched value and memory based decisions, finding convergence in dorsal, but not ventral, PCC.
www.biorxiv.org
August 19, 2026 at 4:27 PM
New preprint! Posterior cingulate is often linked to memory in humans but decisions in monkeys yet there's little consensus about its functional neuroanatomy. We recorded LFPs+single units in humans during matched value and memory based decisions, finding convergence in dorsal, but not ventral, PCC.
Reposted by Melissa Franch, PhD
The superposition of these two studies — both incredible — reflect a snapshot of "one big thing" systems neuroscience has yet to resolve.
In mice:
1. Task-related signals are everywhere.
2. Absent large swaths of "everywhere," suprisingly sophisticated learning & behavior remains.
In mice:
1. Task-related signals are everywhere.
2. Absent large swaths of "everywhere," suprisingly sophisticated learning & behavior remains.
September 4, 2026 at 9:36 PM
The superposition of these two studies — both incredible — reflect a snapshot of "one big thing" systems neuroscience has yet to resolve.
In mice:
1. Task-related signals are everywhere.
2. Absent large swaths of "everywhere," suprisingly sophisticated learning & behavior remains.
In mice:
1. Task-related signals are everywhere.
2. Absent large swaths of "everywhere," suprisingly sophisticated learning & behavior remains.
Reposted by Melissa Franch, PhD
Textbook: tracking position during navigation is primarily the job of the hippocampus. Recent findings suggest a more complicated story.
Recording >20k neurons brainwide, we indeed find spatial representations in every region!
With @kenneth-harris.bsky.social and @carandinilab.net
Recording >20k neurons brainwide, we indeed find spatial representations in every region!
With @kenneth-harris.bsky.social and @carandinilab.net
September 4, 2026 at 6:22 AM
Textbook: tracking position during navigation is primarily the job of the hippocampus. Recent findings suggest a more complicated story.
Recording >20k neurons brainwide, we indeed find spatial representations in every region!
With @kenneth-harris.bsky.social and @carandinilab.net
Recording >20k neurons brainwide, we indeed find spatial representations in every region!
With @kenneth-harris.bsky.social and @carandinilab.net
Reposted by Melissa Franch, PhD
🤖🧠NEW PAPER🧠🤖
(The result of an 8-year project!)
LLMs seem very different from symbolic systems. Yet LLMs excel in symbolic domains (e.g., language/code/math). How do they do it?
Our finding: LLM representations have implicit symbolic structure!
Link in thread ⬇️
1/n
(The result of an 8-year project!)
LLMs seem very different from symbolic systems. Yet LLMs excel in symbolic domains (e.g., language/code/math). How do they do it?
Our finding: LLM representations have implicit symbolic structure!
Link in thread ⬇️
1/n
![Overview of the paper.
Title: The Emergent Symbolic Structure of Artificial Neural Networks
Authors: Tom McCoy, Paul Soulos, Tal Linzen, Paul Smolensky
Left: Neural networks encode information in vectors (there is then an image of a vector), yet they excel at tasks long thought to require symbolic structure (there is then an image of a symbolic representation, specifically a syntax tree). How do LLMs do it?
Right: We find that LLM representations can be closely approximated with symbolic structures. This approximation lets us edit the structure of an LLM’s output by editing the structure of its internal representations, as shown. There is then an image of two edits to LLMs. In the first one, the original input is 3 + 6 * 8, with an answer of 51. But if we swap the positions of the 3 and the 6, the output becomes 30. In the second one, the original input is a Python command repeating the list [Z, U] three times, producing [Z, U, Z, U, Z, U]. But if we edit the input in a way that adds a Q at the end of the input, the output becomes [Z, U, Q, Z, U, Q, Z, U, Q].](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:bcsvlv7ubeagj5virdvatadd/bafkreic3zm2rpc4vtqrz2t3y4vcl3t2owp2sxay42vuh5zm5zkg4jhr64e)
September 1, 2026 at 7:21 PM
🤖🧠NEW PAPER🧠🤖
(The result of an 8-year project!)
LLMs seem very different from symbolic systems. Yet LLMs excel in symbolic domains (e.g., language/code/math). How do they do it?
Our finding: LLM representations have implicit symbolic structure!
Link in thread ⬇️
1/n
(The result of an 8-year project!)
LLMs seem very different from symbolic systems. Yet LLMs excel in symbolic domains (e.g., language/code/math). How do they do it?
Our finding: LLM representations have implicit symbolic structure!
Link in thread ⬇️
1/n
Reposted by Melissa Franch, PhD
My dept/area is hiring! We're open to animal species/model and research area. UCD is an amazing place to do nonhuman animal work - major 🐒 center, a band new rodent vivarium in the Psych building, and capacity to house all sorts of other animals!
recruit.ucdavis.edu/JPF07812
recruit.ucdavis.edu/JPF07812
Assistant Professor of Psychology - Biological Bases of Behavior
University of California, Davis is hiring. Apply now!
recruit.ucdavis.edu
August 31, 2026 at 1:08 AM
My dept/area is hiring! We're open to animal species/model and research area. UCD is an amazing place to do nonhuman animal work - major 🐒 center, a band new rodent vivarium in the Psych building, and capacity to house all sorts of other animals!
recruit.ucdavis.edu/JPF07812
recruit.ucdavis.edu/JPF07812
Reposted by Melissa Franch, PhD
Check out our new TICS article! We present an alternative take on the widely observed alignment between various AI models and human brain representations: the Umwelt Representation Hypothesis.
Are brains and artificial neural networks converging onto universal representations?
There is a seductive idea making the rounds in NeuroAI / machine learning: train systems well enough, and they all converge on the same representation of reality (i.e. a unique world model).
We have thoughts™
1/n
There is a seductive idea making the rounds in NeuroAI / machine learning: train systems well enough, and they all converge on the same representation of reality (i.e. a unique world model).
We have thoughts™
1/n
The Umwelt Representation Hypothesis: rethinking Universality
Recent studies reveal striking representational alignment between artificial neural
networks (ANNs) and biological brains, leading to proposals that all sufficiently
capable systems converge on univer...
www.cell.com
August 31, 2026 at 4:48 PM
Check out our new TICS article! We present an alternative take on the widely observed alignment between various AI models and human brain representations: the Umwelt Representation Hypothesis.
Reposted by Melissa Franch, PhD
First Noel Lab Pre-print!
If you are into "peri-personal space" and feel like we need more single cell recordings, this one is for you!
www.biorxiv.org/content/10.6...
If you are into "peri-personal space" and feel like we need more single cell recordings, this one is for you!
www.biorxiv.org/content/10.6...
An egocentric map prioritizing peri-personal space in the mouse rostro-lateral visual area
All physical interactions between an organism and its environment occur within the space immediately adjacent to and surrounding its body, its peripersonal space (PPS). This space has been extensively...
www.biorxiv.org
August 29, 2026 at 3:16 PM
First Noel Lab Pre-print!
If you are into "peri-personal space" and feel like we need more single cell recordings, this one is for you!
www.biorxiv.org/content/10.6...
If you are into "peri-personal space" and feel like we need more single cell recordings, this one is for you!
www.biorxiv.org/content/10.6...
Reposted by Melissa Franch, PhD
With agentic AI’s rapid infiltration into science, researchers need to develop AI use policies and practices, writes @nicoschuck.bsky.social. #neuroskyence
www.thetransmitter.org/artificial-i...
www.thetransmitter.org/artificial-i...
How, when and why to use agentic AI in our neuroscience labs
With agentic AI’s rapid infiltration into science, researchers need to develop AI use policies and practices.
www.thetransmitter.org
August 24, 2026 at 4:46 PM
With agentic AI’s rapid infiltration into science, researchers need to develop AI use policies and practices, writes @nicoschuck.bsky.social. #neuroskyence
www.thetransmitter.org/artificial-i...
www.thetransmitter.org/artificial-i...
Reposted by Melissa Franch, PhD
First study out! Using a new cooperative paradigm and multi-agent inverse RL, we show that mice spontaneously adopt leader/follower roles, and the prefrontal cortex encodes these role dynamics and an egocentric social value map of the partner’s position. (1/3)
www.nature.com/articles/s41...
www.nature.com/articles/s41...
Asymmetric prefrontal representations for leader–follower dynamics - Nature
Mice spontaneously form leader and follower roles during cooperation, and the medial prefrontal cortex encodes these role dynamics and an egocentric social value map of the partner’s position.
www.nature.com
August 20, 2026 at 1:56 PM
First study out! Using a new cooperative paradigm and multi-agent inverse RL, we show that mice spontaneously adopt leader/follower roles, and the prefrontal cortex encodes these role dynamics and an egocentric social value map of the partner’s position. (1/3)
www.nature.com/articles/s41...
www.nature.com/articles/s41...
Reposted by Melissa Franch, PhD
Excited to share our new paper: "Probing the content of semantic representations in body-selective regions"
We propose a framework based on object co-occurrence to interpret semantic representations of natural scenes predicted by LLM embeddings.
doi.org/10.1162/IMAG...
We propose a framework based on object co-occurrence to interpret semantic representations of natural scenes predicted by LLM embeddings.
doi.org/10.1162/IMAG...
Probing the content of semantic representations in body-selective regions
Abstract. Recent advances in neural networks trained on natural language have revealed that category-selective regions encode complex semantics and contextual information of natural scenes in addition...
doi.org
August 11, 2026 at 3:02 PM
Excited to share our new paper: "Probing the content of semantic representations in body-selective regions"
We propose a framework based on object co-occurrence to interpret semantic representations of natural scenes predicted by LLM embeddings.
doi.org/10.1162/IMAG...
We propose a framework based on object co-occurrence to interpret semantic representations of natural scenes predicted by LLM embeddings.
doi.org/10.1162/IMAG...
Reposted by Melissa Franch, PhD
New perspective piece out in @cp-neuron.bsky.social with Zaid Zada, @adelegoldberg.bsky.social, and Uri Hasson! We try to articulate some of our excitement about LLMs and discuss what kinds of insights they might provide into the neural computations supporting natural language in the human brain.
August 18, 2026 at 1:04 AM
New perspective piece out in @cp-neuron.bsky.social with Zaid Zada, @adelegoldberg.bsky.social, and Uri Hasson! We try to articulate some of our excitement about LLMs and discuss what kinds of insights they might provide into the neural computations supporting natural language in the human brain.
Reposted by Melissa Franch, PhD
Our latest profile is here! Dr. Lea Duncker (@leaduncker.bsky.social) uses statistical and theoretical approaches to study the neural basis of flexible behavior. Follow the link below to learn more!
#StoriesOfWiN #WomenInNeuroscience
www.storiesofwin.org/profiles/202...
#StoriesOfWiN #WomenInNeuroscience
www.storiesofwin.org/profiles/202...
Dr. Lea Duncker — Stories of WiN
uses statistical and theoretical approaches to study the neural basis of flexible behavior
www.storiesofwin.org
August 5, 2026 at 8:24 PM
Our latest profile is here! Dr. Lea Duncker (@leaduncker.bsky.social) uses statistical and theoretical approaches to study the neural basis of flexible behavior. Follow the link below to learn more!
#StoriesOfWiN #WomenInNeuroscience
www.storiesofwin.org/profiles/202...
#StoriesOfWiN #WomenInNeuroscience
www.storiesofwin.org/profiles/202...
Reposted by Melissa Franch, PhD
(1/8) Our approach for untangling brain-wide interactions is out today in @cp-neuron.bsky.social!
Current-Based Decomposition (CURBD) uses RNNs constrained by real neural data to reveal the input driving neurons, uncovering how brain regions talk to each other.
doi.org/10.1016/j.ne...
🧵👇
Current-Based Decomposition (CURBD) uses RNNs constrained by real neural data to reveal the input driving neurons, uncovering how brain regions talk to each other.
doi.org/10.1016/j.ne...
🧵👇
August 7, 2026 at 3:38 PM
(1/8) Our approach for untangling brain-wide interactions is out today in @cp-neuron.bsky.social!
Current-Based Decomposition (CURBD) uses RNNs constrained by real neural data to reveal the input driving neurons, uncovering how brain regions talk to each other.
doi.org/10.1016/j.ne...
🧵👇
Current-Based Decomposition (CURBD) uses RNNs constrained by real neural data to reveal the input driving neurons, uncovering how brain regions talk to each other.
doi.org/10.1016/j.ne...
🧵👇
Reposted by Melissa Franch, PhD
US scientists: Going into fall 26, how are you thinking about things?
Following Fall 24 events and a tumultuous spring 25, I remember a convo with a colleague in which we agreed we just needed to wait things out until fall 25 and we'd have some clarity about the landscape (eg for funding). /1
Following Fall 24 events and a tumultuous spring 25, I remember a convo with a colleague in which we agreed we just needed to wait things out until fall 25 and we'd have some clarity about the landscape (eg for funding). /1
August 7, 2026 at 12:15 PM
US scientists: Going into fall 26, how are you thinking about things?
Following Fall 24 events and a tumultuous spring 25, I remember a convo with a colleague in which we agreed we just needed to wait things out until fall 25 and we'd have some clarity about the landscape (eg for funding). /1
Following Fall 24 events and a tumultuous spring 25, I remember a convo with a colleague in which we agreed we just needed to wait things out until fall 25 and we'd have some clarity about the landscape (eg for funding). /1
Reposted by Melissa Franch, PhD
For a neural circuit to be "specialized", does it need specialized neurons? In our new work - now out in @NatureNeuro - we address this question in the BLA, using representational geometry to propose a new population-centric way of defining circuit specialization. 🧵👇 1/
August 6, 2026 at 9:00 AM
For a neural circuit to be "specialized", does it need specialized neurons? In our new work - now out in @NatureNeuro - we address this question in the BLA, using representational geometry to propose a new population-centric way of defining circuit specialization. 🧵👇 1/
Reposted by Melissa Franch, PhD
CCN time--come say hi!
@cogcompneuro.bsky.social
@cogcompneuro.bsky.social
August 3, 2026 at 6:07 PM
CCN time--come say hi!
@cogcompneuro.bsky.social
@cogcompneuro.bsky.social
Reposted by Melissa Franch, PhD
And on Wednesday afternoon it's time for our Community Event that focuses on what neural encoding models are useful for, where I'll be showcasing several cool projects on EEG encoding during scene perception led by @niklasmuller.bsky.social! 2026.ccneuro.org/community-ev...
Community Event – So You Have a Highly Predictive Model, Now What? Translating Neural Encoding Models into Scientific and Engineering Impact
2026.ccneuro.org
July 31, 2026 at 8:05 PM
And on Wednesday afternoon it's time for our Community Event that focuses on what neural encoding models are useful for, where I'll be showcasing several cool projects on EEG encoding during scene perception led by @niklasmuller.bsky.social! 2026.ccneuro.org/community-ev...
Reposted by Melissa Franch, PhD
Then on Wednesday, double whammy: in the morning I'll present a poster on our hot-of-the-press EEG decoding study on neural representational dynamics between images vs. video, led by @annewzonneveld.bsky.social (Poster C85) openreview.net/forum?id=njG...
Poster Presentation
2026.ccneuro.org
July 31, 2026 at 8:05 PM
Then on Wednesday, double whammy: in the morning I'll present a poster on our hot-of-the-press EEG decoding study on neural representational dynamics between images vs. video, led by @annewzonneveld.bsky.social (Poster C85) openreview.net/forum?id=njG...
Reposted by Melissa Franch, PhD
This project is one of our greatest undertakings as a lab, with Calli at the helm (in terms of data collection, transcription of speech features during naturalistic dialogue, and computational modeling), and we're excited to get everyone's feedback!
August 3, 2026 at 3:32 AM
This project is one of our greatest undertakings as a lab, with Calli at the helm (in terms of data collection, transcription of speech features during naturalistic dialogue, and computational modeling), and we're excited to get everyone's feedback!