The article reviews the current understanding of how the brain solves the problem of object recognition, focusing on the inferior temporal cortex (IT) as a key region. The authors propose that core object recognition, the ability to rapidly identify objects despite substantial appearance variation, is achieved through a cascade of reflexive, feedforward computations. They argue that this process involves the IT cortex, which produces a powerful neuronal representation that can support robust, real-time object categorization and identification. The authors highlight the importance of invariance, the ability to recognize objects across different viewing conditions, as the computational crux of object recognition. They describe how the ventral visual stream, a set of cortical areas, houses critical circuits for this task. The IT cortex, in particular, is crucial for object recognition, with evidence suggesting that it can support robust performance even in the face of changes in object position, scale, and context. The article also discusses the nature of IT neuronal responses, noting that IT neurons are not narrowly tuned "object detectors" but rather have broad selectivity and can maintain their object preferences over small changes in object appearance. This diversity in IT neurons contributes to a distributed representation of objects, which is essential for object recognition. Finally, the authors explore the algorithms that might underlie the IT population representation, suggesting that these algorithms likely involve a combination of feedforward and feedback processes. They propose that the IT cortex uses a series of successive re-representations to untangle object identity manifolds, allowing for easy separation of object identities through simple weighted summation codes. Overall, the article provides a comprehensive overview of the current understanding of object recognition in the brain, emphasizing the role of the IT cortex and the computational principles that may underlie this process.The article reviews the current understanding of how the brain solves the problem of object recognition, focusing on the inferior temporal cortex (IT) as a key region. The authors propose that core object recognition, the ability to rapidly identify objects despite substantial appearance variation, is achieved through a cascade of reflexive, feedforward computations. They argue that this process involves the IT cortex, which produces a powerful neuronal representation that can support robust, real-time object categorization and identification. The authors highlight the importance of invariance, the ability to recognize objects across different viewing conditions, as the computational crux of object recognition. They describe how the ventral visual stream, a set of cortical areas, houses critical circuits for this task. The IT cortex, in particular, is crucial for object recognition, with evidence suggesting that it can support robust performance even in the face of changes in object position, scale, and context. The article also discusses the nature of IT neuronal responses, noting that IT neurons are not narrowly tuned "object detectors" but rather have broad selectivity and can maintain their object preferences over small changes in object appearance. This diversity in IT neurons contributes to a distributed representation of objects, which is essential for object recognition. Finally, the authors explore the algorithms that might underlie the IT population representation, suggesting that these algorithms likely involve a combination of feedforward and feedback processes. They propose that the IT cortex uses a series of successive re-representations to untangle object identity manifolds, allowing for easy separation of object identities through simple weighted summation codes. Overall, the article provides a comprehensive overview of the current understanding of object recognition in the brain, emphasizing the role of the IT cortex and the computational principles that may underlie this process.