Research

The human visual system employs extensive long-range feedback circuitry, where feedforward and feedback connections iteratively refine interpretations through reentrant loops (Di Lollo, 2012). Inspired by this neuroanatomy, a recent computational model incorporated long-range modulatory feedback into a convolutional neural network (Konkle & Alvarez, 2023). While this prior work focused on injecting an external goal signal to leverage feedback for category-based attention, here we investigated its default operation: how learned feedback intrinsically reshapes representational geometry without top-down goals. Analyzing activations from this model across two passes—feedforward versus modulated—on ImageNet data, we examined local (within-category) and global (between-category) structure. Our results demonstrate that feedback significantly compacts category clusters: exemplars move closer to prototypes, and the local structure improves as more near neighbors fall within the same category. Notably, this occurs while largely preserving global structure, as between-category distances remain relatively stable. An exploratory analysis linking local and global changes suggested a positive relationship between local compaction and prototype shifts. These findings reveal an emergent “prototype effect” where fixed long-range feedback automatically refines local representations, potentially enhancing categorical processing efficiency without disrupting overall representational organization. This suggests intrinsic feedback dynamics might contribute fundamentally to perceptual organization.

Advancements in deep neural networks have led to significant progress in computer vision and natural language processing. These networks, trained on real-world stimuli, develop high-level feature representations of stimuli. It is hypothesized that these representations, stemming from different inputs, should converge into similar conceptual systems, as they reflect various perspectives of the same underlying reality. This paper examines the degree to which different conceptual systems can be aligned in an unsupervised manner, using feature-based representations from deep neural networks. Our investigation centers on the alignment between the image and word representations produced by diverse neural networks, emphasizing those trained via self-supervised learning methods. Subsequently, to probe comparable alignment patterns in human learning, we extend this examination to models trained on developmental headcam data from children. Our findings reveal a more pronounced alignment in models trained through self-supervised learning compared to supervised learning, effectively uncovering higher-level structural connections among categories. However, this alignment was notably absent in models trained with limited developmental headcam data, suggesting more data, more inductive biases, or more supervision are needed to establish alignment from realistic input.

Inspired by the concept of the male gaze (Mulvey, 1975) in literature and media studies, this paper proposes a framework for analyzing gender bias in terms of female objectification—the extent to which a text portrays female individuals as objects of visual pleasure. Our framework measures female objectification along two axes. First, we compute an agency bias score that indicates whether male entities are more likely to appear in the text as grammatical agents than female entities. Next, by analyzing the word embedding space induced by a text (Caliskan et al., 2017), we compute an appearance bias score that indicates whether female entities are more closely associated with appearance-related words than male entities. Applying our framework to 19th and 20th century novels reveals evidence of female objectification in literature: we find that novels written from a male perspective systematically objectify female characters, while novels written from a female perspective do not exhibit statistically significant objectification of any gender.