Search for: All records

Humans are able to recognize objects based on both local texture cues and the configuration of object parts, yet contemporary vision models primarily harvest local texture cues, yielding brittle, non-compositional features. Work on shape-vs- texture bias has pitted shape and texture representations in opposition, measuring shape relative to texture, ignoring the possibility that models (and humans) can simultaneously rely on both types of cues, and obscuring the absolute quality of both types of representation. We therefore recast shape evaluation as a matter of absolute configural competence, operationalized by the Configural Shape Score (CSS), which (i) measures the ability to recognize both images in Object-Anagram pairs that preserve local texture while permuting global part arrangement to depict different object categories. Across 86 convolutional, transformer, and hybrid models, CSS (ii) uncovers a broad spectrum of configural sensitivity with fully self- supervised and language-aligned transformers – exemplified by DINOv2, SigLIP2 and EVA-CLIP – occupying the top end of the CSS spectrum. Mechanistic probes reveal that (iii) high-CSS networks depend on long-range interactions: radius- controlled attention masks abolish performance showing a distinctive U-shaped integration profile, and representational-similarity analyses expose a mid-depth transition from local to global coding. A BagNet control, whose receptive fields straddle patch seams, remains at chance (iv), ruling out any “border-hacking” strategies. Finally, (v) we show that configural shape score also predicts other shape- dependent evals (e.g.,foreground bias, spectral and noise robustness). Overall, we propose that the path toward truly robust, generalizable, and human-like vision systems may not lie in forcing an artificial choice between shape and texture, but rather in architectural and learning frameworks that seamlessly integrate both local-texture and global configural shape 

Self-supervised Vision Transformers (ViTs) like DINOv2 show strong holistic shape processing capabilities, a feature linked to computations in their intermediate layers. However, the specific mechanism by which these layers transform local patch information into a global, configural percept remains a black box. To dis- sect this process, we conduct fine-grained mechanistic analyses by disentangling patch representations into their constituent content and positional information. We find that high-performing models demonstrate a distinct multi-stage processing signature: they first preserve the spatial localization of image content through many layers while concurrently refining their positional representations. Compu- tationally, we show that this is supported by a systematic "local-global handoff," where attention heads gradually shift to aggregating information using long-range interactions. In contrast, models with poor configural ability lose content-specific spatial information early and lack this critical positional refinement stage. This positional refinement is further stabilized by register tokens, which mitigate a common artifact in ViTs; repurpose low-information patch tokens into high-norm ’outliers’ to store global information, causing them to lose their local positional grounding. By isolating these high-norm activations in register tokens, the model better preserves the visual grounding of each patch, which we show also leads to a direct improvement in holistic processing. Overall, our findings suggest that holis- tic vision in ViTs arises not just from long-range attention, but from a structured pipeline that carefully manages the interpl 

Abstract Among the animals on this planet, dogs are uniquely adapted for life with humans, a status that exposes them to risks of human-mediated traumatic experiences. At the same time, some lineages of dogs have undergone artificial selection for behavioral phenotypes that might increase risk or resilience to stress exposure, providing an opportunity to examine interactions between heritable and acquired traits. In a large-scale study (N = 4,497), English-speaking dog guardians reported on their dogs’ life histories, current living environments, and provided observer ratings of dog behavior using the Canine Behavior Assessment and Research Questionnaire (C-BARQ). Our analysis revealed that adverse experiences in the first six months of life, such as abuse and relinquishment, were significantly associated with increased aggression and fearfulness in adulthood, even when accounting for factors such as acquisition source, sex, and neuter status. Additionally, effects of adversity on fearful and aggressive behavior systematically varied at the breed level, suggesting heritable factors for risk and resilience for developing particular phenotypes. Our findings establish that breed ancestry and individual experience interact to show fear and aggressive behavior in pet dogs, confirming that socioemotional behavior is shaped by gene-environment interactions. 

Abstract In this paper, we show how relaxation techniques can be used to establish the existence of an optimal contract in the presence of information asymmetry. The method we illustrate was initially motivated by the problem of designing optimal brokerage fees, but it does apply to other optimal contract problems in which (i) the agent controls linearly the drift of a diffusion process, (ii) the direct dependence of the principal’s and the agent’s objectives on the strategy of the agent is of a special form, and (iii) the space of admissible contracts is compact. This method is then applied to establish the existence of an optimal brokerage fee in a market model with a private trading signal observed by the broker’s client, but not by the broker. 

Abstract Intracellular electrophysiology is essential in neuroscience, cardiology, and pharmacology for studying cells’ electrical properties. Traditional methods like patch-clamp are precise but low-throughput and invasive. Nanoelectrode Arrays (NEAs) offer a promising alternative by enabling simultaneous intracellular and extracellular action potential (iAP and eAP) recordings with high throughput. However, accessing intracellular potentials with NEAs remains challenging. This study presents an AI-supported technique that leverages thousands of synchronous eAP and iAP pairs from stem-cell-derived cardiomyocytes on NEAs. Our analysis revealed strong correlations between specific eAP and iAP features, such as amplitude and spiking velocity, indicating that extracellular signals could be reliable indicators of intracellular activity. We developed a physics-informed deep learning model to reconstruct iAP waveforms from extracellular recordings recorded from NEAs and Microelectrode arrays (MEAs), demonstrating its potential for non-invasive, long-term, high-throughput drug cardiotoxicity assessments. This AI-based model paves the way for future electrophysiology research across various cell types and drug interactions. 

Deep neural network models provide a powerful experimental platform for exploring core mechanisms underlying human visual perception, such as perceptual grouping and contour integration—the process of linking local edge elements to arrive at a unified perceptual representation of a complete contour. Here, we demonstrate that feedforward convolutional neural networks (CNNs) fine-tuned on contour detection show this human-like capacity, but without relying on mechanisms proposed in prior work, such as lateral connections, recurrence, or top-down feedback. We identified two key properties needed for ImageNet pre-trained, feed-forward models to yield human-like contour integration: first, progressively increasing receptive field structure served as a critical architectural motif to support this capacity; and second, biased fine-tuning for contour-detection specifically for gradual curves (~20 degrees) resulted in human-like sensitivity to curvature. We further demonstrate that fine-tuning ImageNet pretrained models uncovers other hidden human-like capacities in feed-forward networks, including uncrowding (reduced interference from distractors as the number of distractors increases), which is considered a signature of human perceptual grouping. Thus, taken together these results provide a computational existence proof that purely feedforward hierarchical computations are capable of implementing gestalt “good continuation” and perceptual organization needed for human-like contour-integration and uncrowding. More broadly, these results raise the possibility that in human vision, later stages of processing play a more prominent role in perceptual-organization than implied by theories focused on recurrence and early lateral connections. 

In this paper, we describe efforts of an alliance to increase Pell-grant eligible and first-generation student access to active conference participation by systematically including considerations for student basic needs as well as developing professional science skills and knowledge that aligns with industry and graduate school pathways in computer science. We describe how an alliance creates the structure and flexibility for systematic care for student needs and local innovation to improve educational practice regarding conference participation. We describe our lessons learned for improving access to conferences as well as provide recommendations for increasing student access to professional conference benefits.