The internal representations of early deep artificial neural networks (ANNs) were found to be remarkably similar to the internal neural representations measured experimentally in the primate brain. Here we ask, as deep ANNs have continued to evolve, are they becoming more or less brain-like? ANNs that are most functionally similar to the brain will contain mechanisms that are most like those used by the brain. We therefore developed Brain-Score -a composite of multiple neural and behavioral benchmarks that score any ANN on how similar it is to the brain's mechanisms for core object recognition -and we deployed it to evaluate a wide range of state-of-the-art deep ANNs. Using this scoring system, we here report that: (1) DenseNet-169, CORnet-S and ResNet-101 are the most brain-like ANNs.(2) There remains considerable variability in neural and behavioral responses that is not predicted by any ANN, suggesting that no ANN model has yet captured all the relevant mechanisms.(3) Extending prior work, we found that gains in ANN ImageNet performance led to gains on Brain-Score. However, correlation weakened at ≥ 70% top-1 ImageNet performance, suggesting that additional guidance from neuroscience is needed to make further advances in capturing brain mechanisms. (4) We uncovered smaller (i.e. less complex) ANNs that are more brain-like than many of the best-performing ImageNet models, which suggests the opportunity to simplify ANNs to better understand the ventral stream. The scoring system used here is far from complete. However, we propose that evaluating and tracking model-benchmark correspondences through a Brain-Score that is regularly updated with new brain data is an exciting opportunity: experimental benchmarks can be used to guide machine network evolution, and machine networks are mechanistic hypotheses of the brain's network and thus drive next experiments. To facilitate both of these, we release Brain-Score.org: a platform that hosts the neural and behavioral benchmarks, where ANNs for visual processing can be submitted to receive a Brain-Score and their rank relative to other models, and where new experimental data can be naturally incorporated. computational neuroscience | object recognition | deep neural networks

Non-recurrent deep convolutional neural networks (DCNNs) are currently the best models of core object recognition; a behavior supported by the densely recurrent primate ventral stream, culminating in the inferior temporal (IT) cortex. Are these recurrent circuits critical to ventral stream's execution of this behavior? We reasoned that, if recurrence is critical, then primates should outperform feedforward-only DCNNs for some images, and that these images should require additional processing time beyond the feedforward IT response. Here we first used behavioral methods to discover hundreds of these "challenge" images. Second, using large-scale IT electrophysiology in animals performing core recognition tasks, we observed that behaviorally-sufficient, linearly-decodable object identity solutions emerged ∼30ms (on average) later in IT for challenge images compared to DCNN and primate performance-matched "control" images. We observed these same late solutions even during passive viewing. Third, consistent with a failure of feedforward computations, the behaviorally-critical late-phase IT population response patterns evoked by the challenge images were poorly predicted by DCNN activations. Interestingly, deeper CNNs better predicted these late IT responses, suggesting a functional equivalence between recurrence and additional nonlinear transformations. Our results argue that automatically-evoked recurrent circuits are critical even for rapid object identification. By precisely comparing current DCNNs, primate behavior and IT population dynamics, we provide guidance for future recurrent model development.

242 words 6Significance Statement 97 words 2 7