Abstract
The common misperception of faces in inanimate objects, known as face pareidolia, can be considered an error of face detection. Examination of these errors has the potential to reveal new insight into the functional organization of category-selective cortex in primates. Previously we showed that rhesus monkeys (Macaca mulatta) perceive illusory faces in the same images (e.g. eggplant, bell pepper, coffee) that humans do (Taubert et al., 2017, Current Biology; Taubert et al., 2018, PNAS), demonstrating that this experience is not unique to humans. Here we examined the macaque brain’s response to face pareidolia by directly adapting a fMRI paradigm previously used in humans (Wardle et al., 2017, bioRxiv). The brain’s response to face pareidolia is particularly interesting in terms of understanding category-selective areas in visual cortex, as these stimuli are simultaneously perceived as both objects and faces. We used contrast agent enhanced functional imaging in awake macaques (N=4) in a 4.7T Bruker vertical MRI scanner. The on-off block design used the same paired stimuli as the human experiment: examples of face pareidolia and matched objects belonging to the same category. We categorized regions-of-interest in the superior temporal sulcus as place-selective, object-selective or face-selective, based on independent data from functional localizer runs. Interestingly, the data revealed several differences in the response to these stimuli compared to the human brain. For example, the only face-selective area that reliably responded more to examples of face pariedolia than to matched objects in the macaque brain was located in the prefrontal cortex. This region has been implicated in behavior towards socially relevant information. Overall, our results indicate that objects that belong to more than one semantic category may be processed differently by the macaque and human brain; these differences inform our understanding of high-level perception in both species.
Acknowledgement: The NIMH intramural Research Program