Convolutional Neural Networks (CNNs) surpass human-level performance on visual object recognition, yet their behavior differs from humans in important ways. One prominent example is that CNNs trained on ImageNet exhibit a texture bias, while humans show a strong shape bias. Although CNN shape bias can be increased through e.g., data augmentation or additional training, the source of this discrepancy remains unclear. Developmental research suggests that one factor driving human shape bias is that during early childhood, toddlers tend to fill their field-of-view with close-up objects. We operationalize this close-up as a zoom-in on objects during CNN training, which increases shape bias without additional training or data augmentation. Systematic manipulation of background-object ratios during training reveals a strong inverse correlation with shape bias. Notably, zooming in on objects, thereby more closely emulating child vision, aligns classification accuracy and shape bias between humans and CNNs. Finally, we achieve a near human-like shape bias when using a developmentally-inspired background-object ratio for training and shape bias assessment. These findings demonstrate that a simple adjustment to image datasets — zooming in on objects — can produce human-like shape bias. This suggests that human learning strategies offer a promising avenue for developing human-aligned, efficient, and robust vision CNNs.

The human brain efficiently represents visual inputs through specialized neural populations that selectively respond to specific categories. Advancements in generative modeling have enabled data-driven discovery of neural selectivity using brain-optimized image synthesis. However, current methods independently generate one sample at a time, without enforcing structural constraints on the generations; thus, these individual images have no explicit point of comparison, making it hard to discern which image features drive neural response selectivity. To address this issue, we introduce Brain Activation Control Through Image Variation (BrainACTIV), a method for manipulating a reference image to enhance or decrease activity in a target cortical region using pretrained diffusion models. Starting from a reference image allows for fine-grained and reliable offline identification of optimal visuo-semantic properties, as well as producing controlled stimuli for novel neuroimaging studies. We show that our manipulations effectively modulate predicted fMRI responses and agree with hypothesized preferred categories in established regions of interest, while remaining structurally close to the reference image. Moreover, we demonstrate how our method accentuates differences between brain regions that are selective to the same category, and how it could be used to explore neural representation of brain regions with unknown selectivities. Hence, BrainACTIV holds the potential to formulate robust hypotheses about brain representation and to facilitate the production of naturalistic stimuli for neuroscientific experiments.

What can we learn from comparing video models to human brains, arguably the most efficient and effective video processing systems in existence? Our work takes a step towards answering this question by performing the first large-scale benchmarking of deep video models on representational alignment to the human brain, using publicly available models and a recently released video brain imaging (fMRI) dataset. We disentangle four factors of variation in the models (temporal modeling, classification task, architecture, and training dataset) that affect alignment to the brain, which we measure by conducting Representational Similarity Analysis across multiple brain regions and model layers. We show that temporal modeling is key for alignment to brain regions involved in early visual processing, while a relevant classification task is key for alignment to higher-level regions. Moreover, we identify clear differences between the brain scoring patterns across layers of CNNs and Transformers, and reveal how training dataset biases transfer to alignment with functionally selective brain areas. Additionally, we uncover a negative correlation of computational complexity to brain alignment. Measuring a total of 99 neural networks and 10 human brains watching videos, we aim to forge a path that widens our understanding of temporal and semantic video representations in brains and machines, ideally leading towards more efficient video models and more mechanistic explanations of processing in the human brain.

Neural responses in visual cortex adapt to prolonged and repeated stimuli. While adaptation occurs across the visual cortex, it is unclear how adaptation patterns and computational mechanisms differ across the visual hierarchy. Here we characterize two signatures of short-term neural adaptation in time-varying intracranial electroencephalography (iEEG) data collected while participants viewed naturalistic image categories varying in duration and repetition interval. Ventral- and lateral-occipitotemporal cortex exhibit slower and prolonged adaptation to single stimuli and slower recovery from adaptation to repeated stimuli compared to V1-V3. For category-selective electrodes, recovery from adaptation is slower for preferred than non-preferred stimuli. To model neural adaptation we augment our delayed divisive normalization (DN) model by scaling the input strength as a function of stimulus category, enabling the model to accurately predict neural responses across multiple image categories. The model fits suggest that differences in adaptation patterns arise from slower normalization dynamics in higher visual areas interacting with differences in input strength resulting from category selectivity. Our results reveal systematic differences in temporal adaptation of neural population responses between lower and higher visual brain areas and show that a single computational model of history-dependent normalization dynamics, fit with area-specific parameters, accounts for these differences.

Selective brain responses to objects arise within a few hundreds of milliseconds of neural processing, suggesting that visual object recognition is mediated by rapid feed-forward activations. Yet disruption of neural responses in early visual cortex beyond feed-forward processing stages affects object recognition performance. Here, we unite these discrepant findings by reporting that object recognition involves enhanced feedback activity (recurrent processing within early visual cortex) when target objects are embedded in natural scenes that are characterized by high complexity. Human participants performed an animal target detection task on natural scenes with low, medium or high complexity as determined by a computational model of low-level contrast statistics. Three converging lines of evidence indicate that feedback was selectively enhanced for high complexity scenes. First, functional magnetic resonance imaging (fMRI) activity in early visual cortex (V1) was enhanced for target objects in scenes with high, but not low or medium complexity. Second, event-related potentials (ERPs) evoked by target objects were selectively enhanced at feedback stages of visual processing (from ~220 ms onwards) for high complexity scenes only. Third, behavioral performance for high complexity scenes deteriorated when participants were pressed for time and thus less able to incorporate the feedback activity. Modeling of the reaction time distributions using drift diffusion revealed that object information accumulated more slowly for high complexity scenes, with evidence accumulation being coupled to trial-to-trial variation in the EEG feedback response. Together, these results suggest that while feed-forward activity may suffice to recognize isolated objects, the brain employs recurrent processing more adaptively in naturalistic settings, using minimal feedback for simple scenes and increasing feedback for complex scenes.

The human visual system is thought to use features of intermediate complexity for scene representation. How the brain computationally represents intermediate features is, however, still unclear. Here we tested and compared two widely used computational models — the biologically plausible HMAX model and Bag of Words (BoW) model from computer vision against human brain activity. These computational models use visual dictionaries, candidate features of intermediate complexity, to represent visual scenes, and the models have been proven effective in automatic object and scene recognition. We analyzed where in the brain and to what extent human fMRI responses to natural scenes can be accounted for by the HMAX and BoW representations. Voxel-wise application of a distance-based variation partitioning method reveals that HMAX explains significant brain activity in early visual regions and also in higher regions such as LO, TO while the BoW primarily explains brain acitvity in the early visual area. Notably, both HMAX and BoW explain the most brain activity in higher areas such as V4 and TO. These results suggest that visual dictionaries might provide a suitable computation for the representation of intermediate features in the brain.

The visual world is complex and continuously changing. Yet, our brain transforms patterns of light falling on our retina into a coherent percept within a few hundred milliseconds. Possibly, low-level neural responses already carry substantial information to facilitate rapid characterization of the visual input. Here, we computationally estimated low-level contrast responses to computer-generated naturalistic images, and tested whether spatial pooling of these responses could predict image similarity at the neural and behavioral level. Using EEG, we show that statistics derived from pooled responses explain a large amount of variance between single-image evoked potentials (ERPs) in individual subjects. Dissimilarity analysis on multi-electrode ERPs demonstrated that large differences between images in pooled response statistics are predictive of more dissimilar patterns of evoked activity, whereas images with little difference in statistics give rise to highly similar evoked activity patterns. In a separate behavioral experiment, images with large differences in statistics were judged as different categories, whereas images with little differences were confused. These findings suggest that statistics derived from low-level contrast responses can be extracted in early visual processing and can be relevant for rapid judgment of visual similarity. We compared our results with two other, well- known contrast statistics: Fourier power spectra and higher-order properties of contrast distributions (skewness and kurtosis). Interestingly, whereas these statistics allow for accurate image categorization, they do not predict ERP response patterns or behavioral categorization confusions. These converging computational, neural and behavioral results suggest that statistics of pooled contrast responses contain information that corresponds with perceived visual similarity in a rapid, low-level categorization task.