Untangling the Animacy Organization of Occipitotemporal Cortex

Abstract

Some of the most impressive functional specialization in the human brain is found in occipitotemporal cortex (OTC), where several areas exhibit selectivity for a small number of visual categories, such as faces and bodies, and spatially cluster based on stimulus animacy.Previous studies suggest this animacy organization reflects the representation of an intuitive taxonomic hierarchy, distinct from the presence of face-and body-selective areas in OTC.Using human fMRI, we investigated the independent contribution of these two factors -the face-body division and taxonomic hierarchy -in accounting for the animacy organization of OTC, and whether they might also be reflected in the architecture of several deep neural networks. We found that graded selectivity based on animal resemblance to human faces and bodies masquerades as an apparent animacy continuum, which suggests that taxonomy is not a separate factor underlying the organization of the ventral visual pathway.

The representational hierarchy in human and artificial visual systems in the presence of object-scene regularities

“…As an example, view-invariant features represented in face- [48,49] and hand-selective regions [50,51] reflect domain-specific computations: the former to support identity recognition [52], the latter to support action understanding [50]. In agreement, our results show that in addition to a large division between animal and scene representations, within each domain, representational content reflects the type of computations these networks support: animacy features in animal-selective areas [4,5,53] and layout navigational properties in scene-selective areas [7,54]. We can show this representational diversity because in our study we included separate behavioral-relevant dimensions for objects (i.e., animacy continuum) and background scenes (i.e., navigational properties), while this was typically not done in previous studies.…”

“…Further, human neuroimaging has shown that rich domain-specific object spaces characterise the different processing channels. As an example, behaviorally relevant object dimensions are encoded in object and scene areas: animate features such as the eyes/mouth useful for identifying living entities in the former (6)(7)(8), and spatial layout informative for scene navigation in the latter (9,10,89). At the same time, statistical regularities of the world influence perception (11,12,13,16), and interaction between object and scene processing has been shown at the neural level (13-15, 33, 34, 87).…”

“…While visual information is fully intertwined at the retina level, soon through the human hierarchical visual pathway, information pertaining to the different object domains is processed in largely separated brain networks [e.g., 2, 3] as confirmed by human neuroimaging revealing rich domain-specific object spaces characterizing the different processing channels. As an example, behaviourally relevant image dimensions are encoded in object and scene areas: animate features such as the eyes/mouth, useful for identifying living entities in the former [4][5][6], and spatial layout, informative for scene navigation in the latter [7][8][9]. At the same time, statistical regularities of the world influence perception [10][11][12][13] as shown by faciliatory effects of context observed during object recognition [14] and evidence for interaction between object and scene processing at the neural level [15][16][17][18].…”

A Highly Selective Response to Food in Human Visual Cortex Revealed by Hypothesis-Free Voxel Decomposition

“…We also did not see evidence for previously-proposed selectivities for small inanimate objects 45, 47 or tools 48, 49 , although these selectivities may be located more on the lateral than ventral surface of the brain, outside the search window used here. Of course, there are many reasons why selectivities that exist in the brain might not be detected using fMRI, but the failure of previous findings from fMRI to emerge from the current analysis raises questions about whether those selectivities might already be better accounted for by the components found here 50 .…”

Disentangling five dimensions of animacy in human brain and behaviour

“…Apart from the abovementioned dimensions of animacy, several other human-centred interpretations of animacy have been proposed. Recently reported animacy-related concepts that explain variance in the ventral visual stream fMRI measurements are human-likeness (Rosenthal-von der Pütten et al, 2019), humanness (Contini et al, 2020), resemblance to human faces and bodies (Ritchie et al, 2021), and capacity for self-movement and thought rather than face presence (Proklova & Goodale, 2022). Another similar concept is the animacy continuum, where objects are perceived as more animate when they are more similar to humans (e.g., images of monkeys would be perceived as more animate than insects, even though both species belong to the animal category; Connolly et al, 2012).…”