Cognitive tasks are used to probe neuronal activity during functional magnetic resonance imaging (fMRI) to detect signs of aberrant cognitive functioning in patients diagnosed with schizophrenia (SZ). However, nonlinear (inverted-U-shaped) associations between neuronal activity and task difficulty can lead to misinterpretation of group differences between patients and healthy comparison subjects (HCs). In this paper, we evaluated a novel method for correcting these misinterpretations based on conditional performance analysis.

Participants included 25 HCs and 27 SZs who performed a working memory (WM) task (N-back) with 5 load conditions while undergoing fMRI. Neuronal activity was regressed onto: 1) task load (i.e., parametric task levels), 2) marginal task performance (i.e., performance averaged over all load conditions), or 3) conditional task performance (i.e., performance within each load condition).

In most regions of interest, conditional performance analysis uniquely revealed inverted-U-shaped neuronal activity in both SZs and HCs. After accounting for conditional performance differences between groups, we observed few difference in both the pattern and level of neuronal activity between SZs and HCs within regions that are classically associated with WM functioning (e.g., posterior dorsolateral prefrontal and parietal association cortices). However, SZs did show aberrant activity within the anterior dorsolateral prefrontal cortex.

Interpretations of differences in neuronal activity between groups, and of associations between neuronal activity and performance, should be considered within the context of task performance. Whether conditional performance-based differences reflect compensation, dedifferentiation, or other processes is not a question that is easily resolved by examining activation and performance data alone.

认知任务用于探测功能磁共振成像 (fMRI) 期间的神经元活动，以检测诊断为精神分裂症 (SZ) 的患者认知功能异常的迹象。然而，神经元活动和任务难度之间的非线性（倒 U 形）关联可能导致对患者和健康比较受试者 (HC) 之间的组差异的误解。在本文中，我们评估了一种基于条件性能分析来纠正这些误解的新方法。

参与者包括 25 名 HC 和 27 名 SZ，他们在接受 fMRI 时执行了具有 5 个负载条件的工作记忆 (WM) 任务 (N-back)。神经元活动回归到：1）任务负载（即参数任务水平），2）边际任务性能（即所有负载条件下的平均性能），或3）条件任务性能（即每个负载条件下的性能）。

在大多数感兴趣的区域，条件性能分析独特地揭示了 SZ 和 HC 中的倒 U 形神经元活动。在考虑了组间条件性能差异后，我们观察到 SZ 和 HC 在经典与 WM 功能相关的区域（例如，后背外侧前额叶和顶叶联合皮质）内的神经元活动模式和水平几乎没有差异。然而，SZ 确实在前背外侧前额叶皮层内表现出异常活动。

应在任务绩效的背景下考虑对组间神经元活动差异以及神经元活动与绩效之间关联的解释。基于条件的绩效差异是否反映了补偿、去分化或其他过程，这不是一个仅通过检查激活和绩效数据就能轻松解决的问题。