Daniel C. Kiper
Institute of Neuroinformatics, University and Swiss, Federal Institute of Technology of Zurich

https://doi.org/10.36866/pn.52.19

When a human observer looks at an object, perception of the object is elicited by the activity of numerous neuronal populations. Most visual neuroscientists today would agree on the neuronal events occurring in the first stages of the visual system. The light reflected from the object’s surface reaches the observer’s eyes, where it is transformed into electrical signals by the photoreceptors of the retina. In daylight, three classes of cone photoreceptors respond differentially to the wavelength composition of the incoming light, thereby providing the physiological basis for our trichromatic color vision system. After processing by the retinal network, the signals representing the object are sent through the optic nerves towards the lateral geniculate nucleus (LGN) of the thalamus. The retinal ganglion cells and LGN neurons represent the object’s physical properties in roughly the same way: a population of cells sensitive to local luminance variations signals the object’s contours, while the object’s colour is represented in two distinct channels: one sensitive to red-green modulations, the other to blue-yellow (Derrington er al. 1984). Thus, in the retina and LGN, signals coding for the object’s physical attributes are treated by separate, distinct neuronal populations. Upon reaching the primary visual cortex (V1, striate cortex), the fate of the signals is more mysterious. Neuroscientists disagree on the nature of the object’s representation in the cortex. Since the pioneering work of Nobel laureates D. Hubel and T. Wiesel (Hubel & Wiesel, 1968), it is well established that cortical neurons can perform operations that earlier level neurons cannot. In particular, most individual cortical neurons are sensitive to the orientation of the object’s contours: a cell excited by a vertical contour will remain silent when seeing a horizontal contour, and vice-versa. The orientation selectivity of cortical neurons is thought to be crucial for the perception of form, a very important cue for object identification. In the cortex, the object’s colour is not represented solely by the two distinct channels (red-green and blue-yellow) described above, but by a population of cells with a homogeneous distribution of preferred colors. An obvious question is to determine whether the same cortical neurons signal simultaneously the orientation of the contours and the object’s colour, or whether these tasks are carried out by different populations of cortical neurons. Surprisingly, this simple question has been the matter of intense debate in the last decades.

On the one hand, a number of researchers reported that colour selective cells in primate V1 are not orientation selective, and vice-versa (Livingstone & Hubel, 1988; Shipp & Zeki, 2002). Furthermore, these reports proposed that colour selective, unoriented cells are found in clusters, corresponding to patches of tissue that stain for the mitochondrial enzyme cytochrome oxidase (CO), the so-called CO­blobs. According to that scheme, information about colour and orientation remains segregated within V1, and through most of extrastriate cortex as well. This scheme implies the existence of a later processing stage in which information about an object’s visual attributes converge to yield a coherent, unified percept. The ‘segregated’ scheme has received indirect support from the discovery of functional streams in extrastriate cortical visual areas, with one stream responsible for object localization (the ‘where’ stream), the other concerned with object identification (the ‘what’ stream) (Ungerleider & Mishkin, 1982). The functional distinction between cells within and without the V1 CO-blobs is thought to be maintained in the extrastriate functional streams. Some results are at odd with this proposal.

Quantitative studies of V1 receptive field properties failed to find a segregation of function between cells within and without the CO-blobs. They found many V1 cells selective for both colour and orientation, and little correspondence between their distribution within V1 and the location of the CO-blobs (Leventhal et al. 1995. Similarly, in V2, the original proposal that color and form are treated by segregated neuronal populations, also corresponding to distinct CO-compartments, has also been challenged (Gegenfurtner et al. 1996). Many reasons have been invoked to account for these discrepancies. They imply the following requirements for any study that aims at resolving this issue.

Clearly, any conclusion drawn from data obtained on small samples of neurons will not be conclusive. A large sample of V1 and V2 cells is necessary to resolve the issue. Furthermore, functional properties of cells are best assessed in an awake animal, because anesthesia can alter the response properties of neurons, particularly in extrastriate cortex. Finally, quantitative methods must be used to measure the cells’ properties and classify them into functional classes. Qualitative classification of cells is difficult even for experienced researchers, and has proven quite unreliable. These conditions were all met by a recent study by Friedman et al. (2003). They recorded the activity of cells in awake, behaving monkeys, and used quantitative methods to characterize each cell. They collected data from a large number of V1 (425) and V2 (417) neurons. They computed indices to quantify each cell’s selectivity for colour and orientation and studied the correlation between these indices. If colour and orientation were treated by independent, distinct populations of neurons, one should observe a negative correlation between selectivity for color and orientation. Their results clearly show that no such correlation exists for any of the indices they used. Their results demonstrate that many colour selective cells are orientation selective, and that non-oriented cells can also be unselective for colour.

These results demonstrate that the physical attributes of an object are not always treated by distinct neuronal populations. Instead, visual neurons are concerned by the features most useful to detect and identify objects in a natural environment. Object borders are important such features and they are, in most normal conditions, defined simultaneously by both a luminance and a color gradient. It is therefore not surprising that the primate visual system evolved neurons specialized for the detection of borders simultaneously defined by these two attributes. These results add to the growing literature showing that the properties of natural images determine the receptive field properties of visual neurons to a large extent. In other words, the lesson for neuroscientists is that to understand the brain, one should focus on the tasks that the brain has to solve, and not solely on the physical attributes of the stimuli used in the laboratory.

References

Derrington AM, Krauskopf J, Lennie P. (1984). Chromatic mechanisms in the lateral geniculate nucleus of macaque. J. Physiol. 357, 241-265

Friedman, S.H. Zhou, H, von der Heydt R. (2003). The coding of uniform colour figures in monkey visual cortex. Journal of Physiology. 548.2, 593-613,

Gegenfurtner KR , Kiper DC, Fenstemaker, SB (1996). Processing of color, form and motion in macaque area V2. Visual Neuroscience 13, 161-172.

Hubel DH, Wiesel TN (1968). Receptive fields and functional architecture of monkey striate cortex. J. Physiol. 195:215-43.

Leventhal AG, Thompson KG, Liu, D, Zhou Y, Ault SJ (1995). Concomitant sensitivity to orientation, direction, and color of cells in layers 2,3 and 4 of monkey striate cortex. Journal of Neuroscience 15, 1808-1818.

Livingstone MS, Hubel DH (1988). Segregation of form, color, movement, and depth: anatomy, physiology, and perception. Science 240, 740-749.

Shipp S, Zeki S. (2002). The functional organization of area V2: Specialization across stripes and layers. Visual Neuroscience 19, 187-210.

Ungerleider LG, Mishkin M. (1982). Two cortical visual systems. In Analysis of Visual Behavior, ed. DJ Ingle, MA Goodale, RJW Mansfeld, Cambridge, MA: MIT Press