Xinchi Yu^1,2 (), Ellen Lau^1,2; ¹Program in Neuroscience and Cognitive Science, University of Maryland, ²Department of Linguistics, University of Maryland

Behavioral and neural evidence suggests that human visual working memory (VWM) is implemented by a limited set (~ 3) of pointers, binding features for each object together respectively (Thyer et al., 2022). What has been more controversial is whether different visuo-cognitive domains – for example, systems engaging with biological entities vs. inanimate objects – share the same limited set of pointers, or maintain their own distinct sets. In the current study, we tested whether biological motions (BM) and complex shapes share the same set of pointers, by asking participants to memorize 3 complex shapes, along with 1 or 3 biological motions. We find that, holding more biological motions in the 3- vs. 1-BM condition does not affect Cowan’s K for complex shapes when the probe is a new object (object-change trials) across two experiments (Experiments 1 and 2, N=24 each), but does reduce Cowan’s K for complex shapes when the probe is an old object at a different location (location-change trials; Experiment 2, N=24). The interference effect in the location-change trials, which require binding object features to locations, supports the hypothesis that biological and non-biological objects share the same set of VWM pointers. The contrasting non-interference effect in the object-change trials suggests that similar previous results can be accounted for by the use of a simple feature familiarity strategy that doesn’t require pointers for binding.

Acknowledgements: This work is supported by NSF #1749407 (to E.L.).