Unlocking Cognitive Capabilities and Analyzing the Perception-Logic Trade-off
Abstract
Recent advancements in Multimodal Large Language Models (MLLMs) pursue omni-perception capabilities, yet integrating robust sensory grounding with complex reasoning remains a challenge, particularly for underrepresented regions. In this report, we introduce the research preview of MERaLiON2-Omni (Alpha), a 10B-parameter multilingual omni-perception tailored for Southeast Asia (SEA). We present a progressive training pipeline that explicitly decouples and then integrates "System 1" (Perception) and "System 2" (Reasoning) capabilities. First, we establish a robust Perception Backbone by aligning region-specific audio-visual cues (e.g., Singlish code-switching, local cultural landmarks) with a multilingual LLM through orthogonal modality adaptation. Second, to inject cognitive capabilities without large-scale supervision, we propose a cost-effective Generate-Judge-Refine pipeline. By utilizing a Super-LLM to filter hallucinations and resolve conflicts via a consensus mechanism, we synthesize high-quality silver data that transfers textual Chain-of-Thought reasoning to multimodal scenarios. Comprehensive evaluation on our newly introduced SEA-Omni Benchmark Suite reveals an Efficiency-Stability Paradox: while reasoning acts as a non-linear amplifier for abstract tasks (boosting mathematical and instruction-following performance significantly), it introduces instability in low-level sensory processing. Specifically, we identify Temporal Drift in long-context audio, where extended reasoning desynchronizes the model from acoustic timestamps, and Visual Over-interpretation, where logic overrides pixel-level reality. This report details the architecture, the data-efficient training recipe, and a diagnostic analysis of the trade-offs between robust perception and structured reasoning.
Cite
@article{arxiv.2602.23730,
title = {Unlocking Cognitive Capabilities and Analyzing the Perception-Logic Trade-off},
author = {Longyin Zhang and Shuo Sun and Yingxu He and Won Cheng Yi Lewis and Muhammad Huzaifah Bin Md Shahrin and Hardik Bhupendra Sailor and Heng Meng Jeremy Wong and Tarun Kumar Vangani and Yi Ma and Qiongqiong Wang and Minh Duc Pham and Ridong Jiang and Jingtao Li and Jingyi Liao and Zhuohan Liu and Yanfeng Lu and Manas Gupta and Ai Ti Aw},
journal= {arXiv preprint arXiv:2602.23730},
year = {2026}
}