The reason why GPT-4 fine-tuning is needed for this research is that GPT-4, compared to GPT-3.5, possesses stronger language comprehension and generation capabilities, enabling it to better handle complex scientific data and interdisciplinary knowledge. Research on cognitive science-driven attention mechanism design involves a large amount of specialized terminology and cross-disciplinary content, and fine-tuning GPT-4 ensures that the model generates reports, analyzes data, and provides recommendations with greater precision and professionalism. Additionally, GPT-4 fine-tuning can help optimize research designs and offer more efficient solutions. Given the limitations of GPT-3.5 in handling complex tasks, this research must rely on GPT-4's fine-tuning capabilities to ensure the reliability and innovation of the research outcomes.