Understanding others in everyday situations requires multiple types of information processing (visual, auditory, higher order…) which implicates the use of multiple neural circuits of the human brain. Here, using a multisensory paradigm we investigate one aspect of social understanding less explored in the literature: instead of focusing on the capacity to infer what a specific person is thinking, we explore here how people with high functioning autism (HFA) and matched controls with typical development (TD) infer the "population thinking". For this we created an audio-visual 'social norm inference' task. Participants were required to imagine how most people would judge the appropriateness of vocal utterances in relation to different emotional visual contexts. Behavioral findings demonstrated that HFA individuals show more interindividual variability in these judgments despite equal within-participant reliability relative to TD. This was also the case for judgements of the valence of these vocalizations when presented in isolation. At the neural level, multivoxel pattern analysis of functional magnetic resonance imaging data revealed strikingly similar neural representations between HFA and TD participants at the group level across different hierarchical levels and neural systems. However, analyses at the individual-participant level revealed that the "Temporal Voice Area" (TVA) shows more interindividual variability in the HFA group, both for neural representations and functional connectivity. Thus, this larger neural idiosyncrasy in a high-level auditory area matches with the larger behavioral idiosyncrasy in HFA individuals, when judging auditory valence and its adequacy in different social scenarios. These results suggest that idiosyncrasy in task-relevant sensory areas in HFA participants could underlie their greater difficulties to estimate how others can think.

中文翻译：

---

具有高自闭症功能的成年人在判断情感性发声反应的适当性时，会表现出特质的行为模式，神经表征和“语音区域”的连通性。

在日常情况下理解他人需要多种类型的信息处理（视觉，听觉，高阶…），这意味着要使用人脑的多个神经回路。在这里，我们使用多感官范式来研究文献中较少探讨的社会理解的一个方面：与其着重于推断特定人在想什么的能力，不如说是在研究具有高自闭症（HFA）并与之相匹配的对照的人。典型开发（TD）推断“人口思维”。为此，我们创建了一个视听“社会规范推断”任务。要求参与者想象大多数人将如何判断与不同的情感视觉环境相关的话语发音。行为学发现表明，尽管参与者内部的信度与TD相等，但这些判断中的HFA个人仍表现出更大的个体差异。当单独呈现这些声音时，也需要对这些声音的效价进行判断。在神经水平上，功能磁共振成像数据的多体素模式分析显示，在不同层次级别和神经系统的组水平上，HFA和TD参与者之间的神经表示极为相似。但是，在个人参与者级别的分析显示，“时空语音区”（TVA）在HFA组中显示出更大的个体差异，无论是神经表示还是功能连接。因此，当判断听觉价及其在不同社会情景下的适当性时，在高级听觉区域中较大的神经特质与在HFA个体中的较大行为特质相匹配。这些结果表明，HFA参与者与任务相关的感觉区域的特质可能是他们更大的困难，难以估计其他人的想法。