As large language models (LLMs) increasingly operate as autonomous agents in social contexts, evaluating their capacity for prosocial behavior is both theoretically and practically critical. However, existing research has primarily relied on static, economically framed paradigms, lacking models that capture the dynamic evolution of prosociality and its sensitivity to structural inequities. To address these gaps, we introduce ProSim, a simulation framework for modeling the prosocial behavior in LLM agents across diverse social conditions. We conduct three progressive studies to assess prosocial alignment. First, we demonstrate that LLM agents can exhibit human-like prosocial behavior across a broad range of real-world scenarios and adapt to normative policy interventions. Second, we find that agents engage in fairness-based third-party punishment and respond systematically to variations in inequity magnitude and enforcement cost. Third, we show that policy-induced inequities suppress prosocial behavior, propagate norm erosion through social networks. These findings advance prosocial behavior theory by elucidating how institutional dynamics shape the emergence, decay, and diffusion of prosocial norms in agent-driven societies.

Despite their ubiquity, wireless earbuds remain audio-centric due to size and power constraints. We present VueBuds, the first camera-integrated wireless earbuds for egocentric vision, capable of operating within stringent power and form-factor limits. Each VueBud embeds a camera into a Sony WF-1000XM3 to stream visual data over Bluetooth to a host device for on-device vision language model (VLM) processing. We show analytically and empirically that while each camera’s field of view is partially occluded by the face, the combined binocular perspective provides comprehensive forward coverage. By integrating VueBuds with VLMs, we build an end-to-end system for real-time scene understanding, translation, visual reasoning, and text reading; all from low-resolution monochrome cameras drawing under 5mW through on-demand activation. Through online and in-person user studies with 90 participants, we compare VueBuds against smart glasses across 17 visual question-answering tasks, and show that our system achieves response quality on par with Ray-Ban Meta. Our work establishes low-power camera-equipped earbuds as a compelling platform for visual intelligence, bringing rapidly advancing VLM capabilities to one of the most ubiquitous wearable form factors.

Immersive and believable NPC dialogue requires characters that feel intentional. They remember specific information about themselves, follow through on their goals, and stay true to their personalities across long conversations. We introduce Post-Thinking, a technique that maintains a rolling reflection trace across chat turns. After each response, effectively in the dead time between LLM queries, the model generates a trace reflecting its current goals, emotional state, and narrative intentions. This trace is kept in context and conditions the next response, directing the conversation while designed to add zero perceivable latency to the end user. Critically, each trace is generated with the previous n reflection traces still visible in context, allowing the character’s inner state to compound and evolve naturally throughout extended conversations. As a preliminary study, we synthetically annotated conversations from seed datasets and interviewed human experts to assess their quality in preparation for fine-tuning. We expect the reflection generation pass to help actively ground the character as it encourages the model to explicitly surface aspects of the character definition most relevant to the current moment, counteracting the prompt drift that typically degrades consistency over long exchanges.

Motion deblurring aims to recover clear visual information from a blurred image. Existing methods still face challenges in handling blur with a large spatial extent or high severity, despite the practical importance of such blur conditions in real-world applications involving dynamic scenes and fast motions. To tackle this issue, we decompose the blurred image into multiple frequency band images using asymmetric complex steerable pyramid frequency filters. These band images capture the image structures (edges and textures) across multiple frequency ranges and orientations, with amplitude and phase encoding the spatial intensity and local positional information. Moreover, the correlation image, generated by a three-phase correlation image sensor, is introduced to identify the blurred region and compensate for the degradation of the band image. Consequently, we construct a dual-stream complex-valued neural network and devise a dynamic complex convolution to extract representative features of the band image. An efficient integration strategy including feature interaction and fusion is developed to tightly combine the band and correlation images. Experimental results show that our method achieves competitive performance on the GoPro and Adobe240 datasets. The paired t -test further confirms that our method significantly outperforms existing state-of-the-art approaches in challenging scenes with global and extreme blurs.

Deep neural network models generally enhance super-resolution (SR) reconstruction of remote sensing images but may distort feature edge details. Recovering low-resolution (LR) remote sensing images with clear texture and high-fidelity edge details is challenging. Recent approaches improve feature fusion in the frequency domain, yielding promising results. We propose a dual-domain optimization network (DDOM) based on discrete Fourier transform (DFT) and frequency-domain complex-valued neural networks. Unlike end-to-end approaches in the image domain, DDOM incorporates frequency-domain information via DFT transformation operators, preserving high-level semantics (phase) and low-level statistical information (magnitude). Using lightweight complex neural networks and Swin Transformer architecture, the frequency-domain and image-domain subnetworks are designed. The dual-domain data consistency constraints ensure the positivity of model optimization. Extensive experiments show superior performance over existing methods in quantitative and qualitative evaluations. The proposed scheme’s robustness is verified on additional datasets. Code and model configurations are available athttp://github.com/YB-Cheng/DDOM.