People perceive and experience the same emotion across a wide variety of situations. For example, being chased by an attacker, giving a public speech, and peering over the edge of a cliff all may evoke fear, but the behavioral responses may look quite different. We refer to the ability to identify features of a common emotion across variable contexts (Barrett, 2006; Satpute & Lindquist, 2019) as Òemotion abstractionÓ. This study will create a novel task explicitly designed to measure emotion abstraction across different clinical populations with trait alexithymia – the inability to recognize and describe emotions (Bagby, Parker, & Taylor, 1994). Alexithymia occurs in the majority of individuals on the autism spectrum and about half of individuals with depression (Pogurusse et. al. 2018; Kim et. al, 2008). There is currently no task measuring emotion abstraction while using systematically varying situations. Our task, which is adapted from the Wisconsin Card Sorting Test (Grant & Berg,1948), fills this gap. Participants are instructed to match images that convey a common emotion content (e.g. fear) across diverse perceptual and semantic content. We predict that participants with high trait alexithymia will show worse performance (slower reaction time, atypical matching performance) on the task. Validating this task is the first step toward using it to study variation in emotion abstraction in clinical populations, and also examining its neural underpinnings. Both first and second authors contributed equally.