Lecture 1.3: James DiCarlo - Neural Mechanisms of Recognition Part 1

MIT RES.9-003 Brains, Minds and Machines Summer Course, Summer 2015
View the complete course: ocw.mit.edu/RES-9-003SU15
Instructor: James DiCarlo
Neural circuits underlying object recognition. Feedforward processing in the ventral visual stream from the retina to inferior temporal cortex. Models to decode IT signals to infer object identity and predict human recognition behavior in cluttered scenes.
License: Creative Commons BY-NC-SA
More information at ocw.mit.edu/terms
More courses at ocw.mit.edu

[@beppenonantola216]

This is an awesome talk! Really enjoyed it

[@CharlesVanNoland]

I'm 100% confident that the object recognition capabilities of brains are not a purely visual task that works like how modern ANN AI techniques approach it. It's an experience-understanding task. The brain as a whole has a lifetime of experience with objects that can have different percepts from different orientations and lighting situations while still being an object. I believe neural networks can get really close but there will always be larger edge cases with them than brains have because treating it like a purely visual task is simply not as robust and resilient as treating it like an experience-understanding task.

[@jonathangrant7788]

Actually this is why CNN + Transformers work so well at object recognition

[@zphuo]

Does the different guys see same object trigger the same position in nervous system??

[@BronxGrrlX]

I believe different agents (human, animal, etc) have similar neural maps for the same image, but not identical.