Abstract
A central goal in neuroscience is to understand how processing within the ventral visual stream enables rapid and robust perception and recognition. Recent neuroscientific discoveries have significantly advanced understanding of the function, structure and computations along the ventral visual stream that serve as the infrastructure supporting this behaviour. In parallel, significant advances in computational models, such as hierarchical deep neural networks (DNNs), have brought machine performance to a level that is commensurate with human performance. Here, we propose a new framework using the ventral face network as a model system to illustrate how increasing the neural accuracy of present DNNs may allow researchers to test the computational benefits of the functional architecture of the human brain. Thus, the review (i) considers specific neural implementational features of the ventral face network, (ii) describes similarities and differences between the functional architecture of the brain and DNNs, and (iii) provides a hypothesis for the computational value of implementational features within the brain that may improve DNN performance. Importantly, this new framework promotes the incorporation of neuroscientific findings into DNNs in order to test the computational benefits of fundamental organizational features of the visual system.
Highlights
A central goal in cognitive and computational neuroscience is to understand how processing within the ventral visual stream enables rapid and robust recognition and classification of the visual input
The ventral visual processing stream emerges in V1—the first cortical visual area that resides in the calcarine sulcus [3]—through a series of occipital visual areas, and ends in high-level visual regions in ventral temporal cortex (VTC), whose activation predicts visual perception and recognition [4,5,6,7,8]
Two important insights have emerged from neuroscience research: (i) the functional organization of the ventral visual stream is structured and (ii) it is reliable across individuals
Summary
A central goal in cognitive and computational neuroscience is to understand how processing within the ventral visual stream enables rapid and robust recognition and classification of the visual input. Significant advances in computational models including hierarchical deep neural networks (DNNs) and technological advances that enable training DNNs using large and labelled image sets [21] have brought machine performance in recognition and classification of visual images to a level that rivals human performance [18,22,23,24] This computational work has led to two important insights: (i) neurally inspired architectures trained with millions of images can produce optimal, human-like performance [22,23] and (ii) DNNs that learn by optimizing a behaviourally relevant cost function—such as categorization—better predict neural responses and representations in the primate and human brain, respectively, compared to other DNNs [18,25,26]. We consider similarities and differences between the functional architecture of the brain and DNNs, as well as provide a hypothesis for the computational value of this feature
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have