These findings demonstrate that with appropriate inductive biases, specifically, dual-pathway architectures for multi-scale motion processing and training objectives focused on dynamic visual tasks, ANNs can develop functionally useful representations of motion-defined forms that exhibit better alignment with the visual regions involved in processing dynamic visual signals.
Abstract
The human visual system integrates both static and dynamic information to support form and shape perception, yet the computational principles underlying the integration of motion for object recognition remain unclear. Artificial neural networks (ANNs) offer a computational framework for developing and testing hypotheses about these principles: if ANNs trained on motion-related tasks develop representations that align with brain activity and support object categorization, this would suggest that the training objectives and architectural constraints of these networks may capture key aspects of motion processing in biological visual systems in general, and motion processing for object recognition, in particular. Here, we investigated this question using “object kinematograms”, stimuli in which object form is conveyed solely through motion cues. We measured neural responses of two higher regions of the lateral and the dorsal visual pathways, respectively, with strong sensitivity to dynamic cues from objects: lateral occipitotemporal cortex (LOTbio), and left supramarginal gyrus (SMGlh), as well as primary visual cortex (V1). We compared brain responses to representations extracted from two neural networks: SlowFast, a dual-pathway architecture trained on action recognition that processes slow- and fast-varying visual information with cross-pathway integration, and DorsalNet, a model of the primate dorsal visual pathway trained on embodied self-motion estimation. Representational similarity analysis revealed distinct representational profiles across brain areas, demonstrating functional specialization in motion-based form processing. LOTbio was best characterized by the slow pathway of the SlowFast model, whereas SMGlh showed strong similarity to both models. Critically, we found that representations aligned with brain activity also better supported behavioral function: the full SlowFast model, incorporating both slow and fast pathways, outperformed other models in few-shot categorization of object kinematograms and showed the highest similarity to human perceptual judgments. These findings demonstrate that with appropriate inductive biases, specifically, dual-pathway architectures for multi-scale motion processing and training objectives focused on dynamic visual tasks, ANNs can develop functionally useful representations of motion-defined forms that exhibit better alignment with the visual regions involved in processing dynamic visual signals.
Evidence suggests that connectivity should be treated as a mechanism of occipital plasticity, and structural findings alone do not yet explain how anatomical connectivity supports functional recruitment, how structural profiles relate to functional connectivity, or how structure-function coupling develops with sensory...
Xiang-zhi Xiao, Yu Zhang, Jia-Cheng Zheng et al.· Frontiers in Human Neuroscie...· 0 citations
Results demonstrate that the joint encoding of color purity and motion distinguishes the functional roles of the four HVAs, whereas tuning preferences segregate them into two groups, and suggest that joint encoding may be related to the primary visual cortex neurons with projections to multiple targets.
Zimin Dai, Jin Chang, Deyong Gong et al.· Neuroscience Bulletin· 0 citations
It is shown that distinct cortical spatial representations similar to those found in POR and RSC can be learned along disparate visual pathways (originating in SC and V1), suggesting that the varying features encoded in different visual pathways contribute to the distinct spatial properties in downstream cortical areas...
Yanbo Lian, Patrick A. LaChance, Samantha Malmberg et al.· Science Advances· 0 citations
Discriminating between two stimuli requires that their neural representations become separable by downstream readouts. This can be achieved geometrically, by disentangling the manifolds that population responses form in neural state space. Such reorganization has been observed in associative and motor areas, but never...
Julien Corbo, L. R. Çağlar, O. Erkat et al.· bioRxiv· 0 citations
It is established that an internal model of the world can serve as a primary driver of activity in sensory cortex, and under illumination, visual stimuli gain- modulated this path-invariant activity rather than replacing it with stimulus-driven responses.
Kyu Hyun Lee, Philip Adenekan, Fan Gao et al.· bioRxiv· 0 citations
This work used a variational autoencoder to generate video clips predicted to correspond to maximal activity in each brain region, and identified a network of 11 cortical regions that support active navigation and are tuned for complex feature combinations that capture dynamic scenes that incorporate both world events...
Tian-Jiao Zhang, Cheol Jun Cho, Jack L. Gallant· bioRxiv· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.