A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
The neocortex is organized in columns perpendicular to the surface, where neurons sharing stimulus preferences group together (e.g., orientation tuning in visual cortex). Columns tile the cortex systematically; adjacent columns process adjacent sensory space. Within columns connectivity is dense; between columns, it's sparser and longer-range.
Use multi-electrode arrays to map columns. Record systematically and map receptive fields.
Columns are anatomically isolated—they communicate extensively. All columns are identical—they vary across cortical areas.
From your knowledge of the primary motor cortex and neuronal cell types, you understand that the cerebral cortex contains diverse neurons organized to process and generate signals. Cortical organization describes the architectural principles that allow the neocortex — just 2-4 mm thick — to perform the vast range of computations underlying perception, movement, and thought. The two fundamental organizational axes are layers (horizontal, parallel to the surface) and columns (vertical, perpendicular to the surface).
The neocortex has six layers, numbered I (outermost) to VI (deepest), each with a characteristic mix of cell types and connection patterns. Layer IV is the primary recipient of sensory input from the thalamus — it is thick in sensory cortices and thin in motor cortex, which receives less direct thalamic input. Layers II and III contain pyramidal neurons that project to other cortical areas, forming the cortico-cortical connections that link distant brain regions. Layers V and VI contain large pyramidal cells that project downward — layer V to subcortical targets like the spinal cord and brainstem, layer VI back to the thalamus. This laminar organization means that information flows through a cortical area in a stereotyped sequence: input arrives in layer IV, is processed and elaborated in layers II/III, and output exits from layers V/VI. Think of each cortical area as a circuit board with a consistent wiring diagram, even though the specific computations vary by region.
Perpendicular to the layers, neurons are organized into columns — vertical groups of cells spanning all six layers that share similar response properties. The concept was first demonstrated by Vernon Mountcastle in somatosensory cortex, where he found that all neurons encountered in a vertical electrode penetration responded to the same type of skin stimulus (e.g., light touch versus deep pressure) at the same body location. Hubel and Wiesel later showed that neurons in a single column of primary visual cortex all prefer edges at the same orientation. Adjacent columns prefer slightly different orientations, and a full rotation of preferred orientations (0° through 180°) is covered in a systematic progression across about 1 mm of cortex — a structure called a hypercolumn. This columnar tiling means that each small patch of cortex contains a complete set of feature detectors for a small region of sensory space.
The columnar principle has important functional consequences. Within a column, neurons are densely interconnected — they share information vertically across layers, allowing each layer's specialized connectivity to operate on the same input features. Between columns, connections are sparser and often travel longer distances through horizontal fibers in layers II/III. These lateral connections preferentially link columns with similar response properties (e.g., columns preferring the same orientation), creating networks that can coordinate responses across larger spatial scales. This architecture explains phenomena like contour integration in vision, where the brain links aligned edge segments into continuous contours. However, it is important to recognize that cortical columns are not rigid, anatomically walled-off compartments — they are statistical tendencies in the organization of connectivity and response properties, and their prominence varies across cortical areas and species.