arXiv subjects

Melika Honarmand

Publications and source records attributed to Melika Honarmand.

4 recordsLinked to original sources

arXiv2026

Reward Valuation in Vision Language Models: Causal Mechanisms Underlying Anhedonia

Recent Vision-Language Models capture increasingly complex aspects of human cognition. Here we ask whether this alignment extends to reward valuation, which we assess in a mechanistic framework built on clinical tests that were developed to evaluate anhedonia and motivational deficits in major depressive disorder. In the brain, anhedonia is frequently linked to dysregulation in the Nucleus Accumbens (NAc) and the broader dopaminergic reward system. While neuroimaging has localized these deficits, establishing a causal link between NAc activity and specific behavioral symptoms remains a challenge. We use these ideas from neuroscience to functionally identify reward-anticipatory units in vision language models, and test their causal role via targeted perturbations. Perturbing NAc-selective units induces behavioral effects that mirror human anhedonia: the model shifts toward low-effort, low-reward options in effort-based decision-making tasks. Crucially, our results reflect a specific deficit in reward valuation and anticipation rather than a loss of task capability: the perturbed model maintains baseline performance when reward-based choice is removed. This induced vulnerability further aligns with clinical anhedonia and motivation scales, including DARS and MAP-SR. Taken together, these results reveal reward valuation circuits in AI models that parallel those in humans.

cs.LG↗

arXiv2026

Inducing Dyslexia in Vision Language Models

Dyslexia, a neurodevelopmental disorder characterized by persistent reading difficulties, is often linked to reduced activity of the visual word form area (VWFA) in the ventral occipito-temporal cortex. Traditional approaches to studying dyslexia, such as behavioral and neuroimaging methods, have provided valuable insights but remain limited in their ability to test causal hypotheses about the underlying mechanisms of reading impairments. In this study, we use large-scale vision-language models (VLMs) to simulate dyslexia by functionally identifying and perturbing artificial analogues of word processing. Using stimuli from cognitive neuroscience, we identify visual-word-form-selective units within VLMs and demonstrate that they predict human VWFA neural responses. Ablating model VWF units leads to selective impairments in reading tasks while general visual and language comprehension abilities remain intact. In particular, the resulting model matches dyslexic humans' phonological deficits without a significant change in orthographic processing, and mirrors dyslexic behavior in font sensitivity. Taken together, our modeling results replicate key characteristics of dyslexia and establish a computational framework for investigating brain disorders.

cs.CL↗

arXiv2025

Fast and flexible long-range models for atomistic machine learning

Most atomistic machine learning (ML) models rely on a locality ansatz, and decompose the energy into a sum of short-ranged, atom-centered contributions. This leads to clear limitations when trying to describe problems that are dominated by long-range physical effects - most notably electrostatics. Many approaches have been proposed to overcome these limitations, but efforts to make them efficient and widely available are hampered by the need to incorporate an ad hoc implementation of methods to treat long-range interactions. We develop a framework aiming to bring some of the established algorithms to evaluate non-bonded interactions - including Ewald summation, classical particle-mesh Ewald (PME), and particle-particle/particle-mesh (P3M) Ewald - into atomistic ML. We provide a reference implementation for pyTorch as well as an experimental one for JAX. Beyond Coulomb and more general long-range potentials, we introduce purified descriptors which disregard the immediate neighborhood of each atom, and are more suitable for general long-ranged ML applications. Our implementations are fast, feature-rich, and modular: They provide an accurate evaluation of physical long-range forces that can be used in the construction of (semi)empirical baseline potentials; they exploit the availability of automatic differentiation to seamlessly combine long-range models with conventional, local ML schemes; and they are sufficiently flexible to implement more complex architectures that use physical interactions as building blocks. We benchmark and demonstrate our torch-pme and jax-pme libraries to perform molecular dynamics simulations, to train ML potentials, and to evaluate long-range equivariant descriptors of atomic structures.

physics.chem-ph↗

arXiv2023

An Assessment of PC-mer's Performance in Alignment-Free Phylogenetic Tree Construction

Background: Sequence comparison is essential in bioinformatics, serving various purposes such as taxonomy, functional inference, and drug discovery. The traditional method of aligning sequences for comparison is time-consuming, especially with large datasets. To overcome this, alignment-free methods have emerged as an alternative approach, prioritizing comparison scores over alignment itself. These methods directly compare sequences without the need for alignment. However, accurately representing the relationships between sequences is a significant challenge in the design of these tools. Methods:One of the alignment-free comparison approaches utilizes the frequency of fixed-length substrings, known as K-mers, which serves as the foundation for many sequence comparison methods. However, a challenge arises in these methods when increasing the length of the substring (K), as it leads to an exponential growth in the number of possible states. In this work, we explore the PC-mer method, which utilizes a more limited set of words that experience slower growth 2^k instead of 4^k compared to K. We conducted a comparison of sequences and evaluated how the reduced input vector size influenced the performance of the PC-mer method. Results: For the evaluation, we selected the Clustal Omega method as our reference approach, alongside three alignment-free methods: kmacs, FFP, and alfpy (word count). These methods also leverage the frequency of K-mers. We applied all five methods to 9 datasets for comprehensive analysis. The results were compared using phylogenetic trees and metrics such as Robinson-Foulds and normalized quartet distance (nQD). Conclusion: Our findings indicate that, unlike reducing the input features in other alignment-independent methods, the PC-mer method exhibits competitive performance when compared to the aforementioned methods especially when input sequences are very varied.

q-bio.QM↗