The Warp and Weft in the Monkey Inferior Temporal Cortex

     The cerebral neocortex is not a randomly interconnected set of neurons, but a highly structured fabric woven from the gwarph (vertical or radial connections) and the gwefth (horizontal connections).  Although the radial arrangement of neurons and connections can be recognized throughout the entire neocortex, evidence for functionally defined columns (i.e., clusters of neurons with similar characteristics or functions across all or most cortical layers) is available only for a few sensory cortical areas.  The most well-documented area is the primary visual cortex, where functionally defined columns have been known since early 60fs.  This is also the area in which the horizontal axon system has been first described and attracted the attention of many researchers.  In higher association cortices, however, evidence for functional columns is scarce, and the presence of horizontal axon system has just begun to be demonstrated in the past few years.


The Warp and Weft in the Monkey Inferior Temporal Cortex

JRDC Forum for Multi-Disciplinary Researches "Probing into the Brain" (Miyajima, Japan)
“‘“cˆκ˜Y@uƒTƒ‹‰Ί‘€“ͺ—t”ηŽΏ‚̏cŽ…‚Ζ‰‘Ž…vV‹ZpŽ–‹Ζ’cŽεΓˆΩ•ͺ–μƒtƒH[ƒ‰ƒ€i‹{“‡j1994.2.14

Ichiro Fujita
PRESTO, Research Development Corporation of Japan; and Laboratory for Neural Information Processing, Frontier Research Program, RIKEN, Japan.

The cerebral neocortex is not a randomly interconnected set of neurons, but a highly structured "fabric" woven from the warp (columnar connections) and the weft (horizontal connections). Although the columnar arrangement of neurons and connections can be recognized throughout the entire neocortex, evidence for functionally defined columns (i.e., clusters of neurons with similar characteristics or functions across all or most cortical layers) is available only for a few early sensory areas.

The best documented area is the primary visual cortex, where functionally defined columns have been known since early 60's. This is also the system in which horizontal axon system have been first described and attracted many researchers. For higher association cortices, however, evidence for functional columns is scarce, and horizontal axon system has just begun to be demonstrated in the past few years.

My colleagues and I have recently found that area TE of the inferior temporal cortex of the macaque monkey, the final station of the visual pathway responsible for object recognition, consists of columns, each containing cells selective for similar features of objects (1). In the course of searching for anatomical correlate of this columnar organization, we have analyzed the horizontal axon system in this cortex (2). In this presentation, I shall describe the nature of the columnar and horizontal axon system in area TE.

TE cells respond selectively to particular visual features of objects such as shape, color, texture or a combination of these. Stimulus features critical for activating individual TE cells are more complex than those for neurons at earlier stages of the visual system, although they are still simpler than the complexity of a real-world object. Activity of individual TE cells does not signal abstract representation of object. They respond to partial features of objects.

Neurons responsive to similar, but slightly different, object features cluster vertically across the cortical layers, forming columnar organizations. The width of each column is 0.4-0.5 mm. Multiple separate columns respond to similar stimulus features, whereas intervening columns prefer quite different features. The border between two adjacent columns is rather discrete: when a recording electrode is moved across the border by 50 ƒΚm a drastic change will occur in effective stimulus features. Repeated recordings from the same site over a period of 30 days enable us to find neurons with similar selectivities, suggesting that the stimulus selectivity of columns is maintained stable at least over this period. This does not exclude the possibility that TE cells are plastic in adults: they do change their stimulus preferences after a learning of new stimuli as Kobatake et al. (1993) have shown (3). The discreteness and temporal stability of columns provide us with rationale to estimate the number of columns in area TE. One TE is estimated to contain 1300-2000 columns.

Horizontal axons in area TE arborize in a patchy manner as in the striate and prestriate cortices. Size and spacing of the axon patches are larger than those in earlier cortical stages, and closely match those of columns with similar stimulus preferences measured in physiological experiments. These patches takes a cylindrical form spanning layer 1 to 3 or even to 5. This and other anatomical features, such as vertical connections across layers and columnar arborization of afferent fibers from a preceding cortex, may contribute to the shared stimulus preferences among neurons within a column. It is an open question what the functional relation is between columns linked by horizontal axons.

When a monkey sees an object, a particular subset of neurons will be activated by the image. Different component features of the object activate different columns. Composition of the entire image may be represented by the combination of active columns. If techniques such as optical imaging of neural activity can be successfully applied to area TE, this prediction will be directly tested, and the question of the relationship of columns connected by horizontal axons can also be answered. Refinement of such technologies is certainly a key to advance our understanding of the functioning of columns in area TE or of the cerebral neocortex in general.

References

(1) Fujita, I., Tanaka, K., Ito, M., Cheng, K. (1992) Nature, 360:343-346.
(2) Fujita, I., Fujita, T. (1993) Soc. Neurosci. Abstr., 19:971.
(3) Kobatake, E., Tanaka, K., Wang, G., Tamori, Y. (1993) Soc. Neurosci. Abstr., 19:975.


Back