Brain-like representational straightening of natural movies in robust feedforward neural networks
This work addresses the challenge of bridging biological and artificial vision by providing a bio-plausible mechanism for representational straightening, though it is incremental as it builds on prior findings in neural representations and robustness techniques.
The study tackled the problem of achieving representational straightening—a decrease in curvature in visual feature representations for natural movie sequences—in artificial feedforward neural networks, which previously lacked this biological hallmark. The result showed that robustness to input noise, through adversarial training and Random Smoothing, induced straightened feature codes, improved predictions of primate V1 neural data by up to 20% over baselines, and enabled linear interpolation to generate intervening movie frames.
Representational straightening refers to a decrease in curvature of visual feature representations of a sequence of frames taken from natural movies. Prior work established straightening in neural representations of the primate primary visual cortex (V1) and perceptual straightening in human behavior as a hallmark of biological vision in contrast to artificial feedforward neural networks which did not demonstrate this phenomenon as they were not explicitly optimized to produce temporally predictable movie representations. Here, we show robustness to noise in the input image can produce representational straightening in feedforward neural networks. Both adversarial training (AT) and base classifiers for Random Smoothing (RS) induced remarkably straightened feature codes. Demonstrating their utility within the domain of natural movies, these codes could be inverted to generate intervening movie frames by linear interpolation in the feature space even though they were not trained on these trajectories. Demonstrating their biological utility, we found that AT and RS training improved predictions of neural data in primate V1 over baseline models providing a parsimonious, bio-plausible mechanism -- noise in the sensory input stages -- for generating representations in early visual cortex. Finally, we compared the geometric properties of frame representations in these networks to better understand how they produced representations that mimicked the straightening phenomenon from biology. Overall, this work elucidating emergent properties of robust neural networks demonstrates that it is not necessary to utilize predictive objectives or train directly on natural movie statistics to achieve models supporting straightened movie representations similar to human perception that also predict V1 neural responses.