Abstract
Recent work (Richler et al., 2019) revealed a reliable domain-general object recognition ability, o, independent from general intelligence. o was originally measured only with novel objects to avoid the problem of variability in experience, but here we extend this work by measuring o for both novel and familiar objects. In addition, we measure performance for ensemble mean judgments, to characterize the extent to which ensemble processing relates across categories and is related to o. Prior work found strong correlations between ensemble coding of face identity and face expression (Haberman Brady & Alvarez, 2015) and between performance on ensemble judgments tasks with different non-face categories (Chang & Gauthier, VSS 2020), but the relation to o has not been investigated. Here, we collected data from 285 subjects using images of three categories of novel objects (Ziggerins, Sheinbugs and Greebles) and three categories of familiar objects (birds, planes, Transformers). For each category, subjects performed three tasks: a learning task (similar to the Novel Object Memory Test and the Vanderbilt Expertise Test), a matching task (same/different) and an ensemble mean judgment task for arrays of four items. We used Confirmatory Factor Analysis to estimate on for novel objects and of for familiar objects using indicators from the learning and matching tasks, as well as an ensemble perception factor. Results revealed that the model fit well (RMSEA= .038), with an almost perfect correlation between on and of (.985) and a strong correlation between the ensemble perception factor and on and of (.63 and .69, respectively). The results support a common domain-general ability across familiar and novel objects, despite the variability in experience that subjects have with familiar categories. They also support recent findings of a domain-general ability for ensemble processing, which shows a robust relation with o.