%0 Journal Article
%T Evidence that recurrent circuits are critical to the ventral stream¡¯s execution of core object recognition behavior
%J -
%D 2019
%R https://doi.org/10.1038/s41593-019-0392-5
%X Non-recurrent deep convolutional neural networks (CNNs) are currently the best at modeling core object recognition, a behavior that is supported by the densely recurrent primate ventral stream, culminating in the inferior temporal (IT) cortex. If recurrence is critical to this behavior, then primates should outperform feedforward-only deep CNNs for images that require additional recurrent processing beyond the feedforward IT response. Here we first used behavioral methods to discover hundreds of these ¡®challenge¡¯ images. Second, using large-scale electrophysiology, we observed that behaviorally sufficient object identity solutions emerged ~30£¿ms later in the IT cortex for challenge images compared with primate performance-matched ¡®control¡¯ images. Third, these behaviorally critical late-phase IT response patterns were poorly predicted by feedforward deep CNN activations. Notably, very-deep CNNs and shallower recurrent CNNs better predicted these late IT responses, suggesting that there is a functional equivalence between additional nonlinear transformations and recurrence. Beyond arguing that recurrent circuits are critical for rapid object identification, our results provide strong constraints for future recurrent model development
%U https://www.nature.com/articles/s41593-019-0392-5