Conference on Cognitive Computational Neuroscience

Face pareidolia refers to seeing an illusory face in inanimate objects. Emerging neuroimaging evidence suggests that face-selective regions in the primate ventral visual stream respond more to pareidolia compared to control objects (Taubert et al, 2022; Wardle et al, 2020). Here, we characterized face-cell responses to pareidolia images and investigate whether the face-like configuration of pareidolia images is necessary to drive the neuronal responses. For each pareidolia image, we shuffled the four image quadrants, such that the individual features were preserved, but no longer in the original face-like configuration. Overall, we find that face-selective neurons show a stronger response for intact pareidolia compared to match control objects and that face selectivity correlates positively with pareidolia selectivity. Strikingly, quadrant scrambling of the images did not abolish this relationship, suggesting that face-cell responses to pareidolia images is not driven by their perceived “faceness”. Furthermore, an encoding model trained only on non-face images could capture the neural preference for pareidolia and its relation to face selectivity, suggesting that the face and pareidolia selectivity could be explained by common features also present in non-face images.