A computational psychiatric approach to understanding category representation in autism spectrum disorder

分类是一项重要的认知技能，它使我们能够构建世界，将以前获得的知识转移到新的情况下，并与环境快速互动。为了确定刺激是否属于给定的类别，可以将该项目与存储的类别样本或类别的抽象平均值（原型）进行比较。重要的是，个体必须在两种策略之间进行灵活的选择，因为最优表征策略依赖于特定的分类问题，自闭症谱系障碍（ASD）个体倾向于关注细节，而对上下文信息的关注较少。尽管在许多分类任务中表现完好，但早期的证据表明，注重细节的风格会导致在抽象原型方面的特定困难，这可能是社交困难的前兆。然而，目前尚不清楚这些困难是否可以通过对基于范例的类别表示的偏见来弥补。计划中的项目将采用一种高度敏感、成熟的方法，通过结合认知建模和多元神经成像分析，区分基于范例和原型的类别表示。这种方法将使我们能够在个人层面上理清分类策略。分层多元回归分析将用于调查正式的ASD诊断和自闭症特征的程度是否对分类风格的选择有独立的贡献，以及这些影响是否对抑郁和焦虑的群体差异有强大的影响。结果将与社会类别学习任务的表现和临床测量（即社会功能，重复行为/限制兴趣，生活质量）有关。自上而下对感觉区域的影响也促进了类别知识的有效获取，从而使用先验的类别知识来指导知觉加工。在患有ASD的个体和自闭症特征较高的神经典型人群中，感觉加工通常较少受到先前（“次先验”理论）经验的调节。间接证据表明，这种现象延伸到类别学习，并可以解释为什么ASD的类别知识习得有时会减慢。因此，计划中的项目将采用一个完善的范例，适合于揭示抽象类别知识自上而下的影响对点模式的感觉类别表征的调节。结果将与社会类别学习任务的次先验测量和临床测量有关。拟议的项目旨在促进对ASD基本认知过程的理解，以及它们与典型ASD症状的关系。潜在的临床意义包括识别自闭症谱系障碍诊断的神经元生物标志物，以及为自闭症谱系障碍患者提供适应（社交技能）训练的建议。

Categorization is a vital cognitive skill that allows us to structure the world, transfer previously acquired knowledge to new situations and interact quickly with the environment. To decide whether a stimulus belongs to a given category, the item can be either compared to stored category exemplars or to the abstract average of the category (the prototype). Importantly, individuals have to choose flexibly between the two strategies, as the optimal representational strategy is dependent on the particular categorization problem, Individuals with autism spectrum disorder (ASD) tend to focus on details and to pay less attention to contextual information. Despite intact performance in many categorization tasks, early evidence suggests that the detail-focused style leads to specific difficulties in the abstraction of prototypes, which could be a precursor to social difficulties. However, it is unclear whether these difficulties are compensated by a bias towards exemplar-based category representation.The planned project will employ a highly sensitive, well-established approach to differentiate exemplar- and prototype-based category representation by the combination of cognitive modeling with multivariate neuroimaging analyses. This approach will enable to disentangle categorization strategies on the individual level. Hierarchical multiple regression analyses will be used to investigate whether formal ASD diagnosis and the degree of autistic traits make independent contributions to the choice of categorization style and whether these effects are robust against group differences in depression and anxiety. Results will be related to performance in a social category learning task and clinical measures (i.e. social functioning, repetitive behavior/restricted interests, quality of life).The efficient acquisition of category knowledge is also facilitated by top-down influences on sensory areas, such that prior category knowledge is used to guide perceptual processing. In individuals with ASD and neurotypical persons high in autistic traits, sensory processing is often less modulated by prior experience in (‘hypo-prior’-theory). Indirect evidence suggests that this phenomenon extends to category learning and could explain why category knowledge acquisition in ASD is sometimes slowed down. The planned project will therefore employ a well-established paradigm suitable to uncover the modulation of sensory category representation of dot-patterns by top-down influences of abstract category knowledge. Results will be related to measures of hypo-priors in a social category learning task and clinical measures.The proposed project seeks to advance the understanding of basic cognitive processes in ASD and how they related to classical ASD symptoms. Potential clinical implications include the identification of neuronal biomarkers for ASD diagnosis and suggestions for adaptations of (social skills) trainings for individuals with ASD.