Pereira fMRI Dataset

该数据集包含了5位成年受试者的功能性磁共振成像（fMRI）记录，其中4位为英语母语者，1位为双语者。这些记录涉及与认知语言处理相关的神经活动表现。刺激材料包括180个孤立刺激和627个连续刺激，时间分辨率为两秒。数据规模涉及5位受试者，研究任务旨在探究LLM（大型语言模型）表示与大脑认知处理信号的对应关系。

This dataset comprises functional magnetic resonance imaging (fMRI) recordings from 5 adult participants, including 4 native English speakers and 1 bilingual individual. The recordings capture neural activity associated with cognitive language processing. The stimulus materials consist of 180 isolated stimuli and 627 sequential stimuli, with a temporal resolution of 2 seconds. Involving a total of 5 participants, this study aims to explore the correspondence between Large Language Model (LLM) representations and the neural signals underlying brain cognitive processing.

搜集汇总

背景与挑战

背景概述

Pereira fMRI Dataset是一个研究从大脑激活数据解码语言意义的项目，通过语义空间向量表示和高效采样方法，能够解码具体和抽象主题的句子，并分析句子间的语义相似性。