Jones Seminar: Scaling Laws vs Neural Laws—Toward more natural artificial vision

The remarkable progress of modern computer vision has been propelled by the relentless logic of scaling laws: bigger models, more data, more compute, predictably better performance. On benchmarks like ImageNet, deep networks now match or even surpass human accuracy. Yet beneath these headline results, the alignment with human vision is fragile: On deceptively simple probes from the cognitive sciences, even the largest models drop to near-chance, and on ImageNet itself, models that reach human accuracy do so by strikingly different visual strategies—a divergence that, troublingly, widens with scale.

In this talk, I will argue that the path to more natural artificial vision lies not in pushing scaling laws further, but in a deeper engagement with the neural laws of biological vision: developmental principles that shape how brains learn to see, and architectural constraints that impose strong inductive biases on cortical processing. I will share recent work from my lab on two such laws. On the learning side, I will present evidence that pairing the right predictive objectives with naturalistic video—sequences of object transformations like those the developing brain encounters—pulls deep networks toward markedly more human-like visual strategies, and I will point to active, embodied vision as the next frontier for capabilities such as spatial perspective-taking that passive video alone cannot deliver. On the architectural side, I will show how recent advances in state space models let us scale cortical recurrent feedback into a brain-inspired, linear-time alternative to transformer self-attention—one that closes the gap on reasoning benchmarks where transformers fail to train, and that on real-world vision traces more favorable compute- and parameter-efficiency curves than transformers.

Together, these results point toward a future in which scaling laws and neural laws are in agreement rather than in tension, and in which computer vision, in dialogue with the brain sciences, helps build AI systems that are not only more capable but more aligned with the kind of intelligence we ultimately seek to understand and emulate.

About the Speaker(s)

Thomas Serre
Professor of Cognitive and Psychological Sciences and Computer Science, Brown U

Thomas Serre is the Thomas J. Watson Sr. Professor of Science at Brown University and professor of cognitive and psychological sciences and computer science. He is faculty director of the Center for Computation & Visualization and associate director of the Center for Computational Brain Science, and is affiliated with the Carney Institute for Brain Science, the Center for Theoretical Physics and Innovation, and the Data Science Institute. He is a Fellow of the European Laboratory for Learning and Intelligent Systems and holds an international chair in artificial intelligence at the Artificial and Natural Intelligence Toulouse Institute in France. He received his PhD in neuroscience from MIT and his MSc in electrical engineering and computer science from Télécom Bretagne. His research focuses on understanding the computational principles of biological vision and developing brain-inspired models of visual intelligence.

Contact

For more information, contact Joyce Xiao at joyce.xiao@dartmouth.edu .