Semir Zeki's experimental studies of vision in primate monkeys established that different aspects of vision, especially color, motion, and form (shape/orientation) are not processed in the primary visual cortex (now known as V1), but in separate areas in the brain, now called V2, V3, V4, V5/MT, etc.

Making single-cell recording in these areas, Zeki showed that their processes were all happening in parallel, and even at different times. Zeki located color processing in V4, where it occurs 80ms before motion and precedes form by 40ms. This temporal asynchrony led Zeki to be a strong proponent of parallel processing in the brain, though he was very skeptical that the brain is a computer, neither serial or parallel processing.

Zeki made his argument for parallelism in the brain in his 1993 book A Vision of the Brain.

Parallelism, computational neurobiology and neural networks

'Computational neurobiology' has been all the rage for some years and many consider that this approach is fundamental in any effort to unravel the intricacies of the brain. The general argument in favour of a strong computational approach in neurobiological studies is this: The brain is much too complicated an organ to be studied without some guiding theory. Because computers also undertake complex tasks, including that of 'seeing' (robotics), they are the natural source for such guiding theories, which can then be tested by direct experimentation. The argument is not without considerable merits. Theory has always been of importance in the collection of facts, including even the mere anatomical facts. After all, collection of the voluminous facts on the structure of the nervous system by Ramon y Cajal was not done in the void, but with the specific aim of proving the neuron doctrine, the one which supposed that the nervous system was not made of a continuous sheet but of discontinuous elements, the neurons, which connected with each other. It is likely that theory will be much more important in the future, particularly in studies of the higher functions of the brain, which includes the function of seeing. But does the experimental neurobiologist have to rely on the computational neurohiologist for his theory and, if so, to what extent?

In trying to formulate an answer, it is as well to record the striking fact that the principle of parallelism in the cerebral cortex at large, and in the visual cortex in particular, not only antedates the current conversion of computational neurobiologists to the idea, but was derived from what many would regard as pedestrian anatomical studies, not from computational theories or from neural networks, even though the latter might have been the natural source for such an idea. One looks in vain through the pre-1975 computational literature for a clear and explicit statement of the principle of parallelism in the visual cortex. Even after the anatomical demonstration of parallel connections in the visual cortex, and the demonstration of multiple, specialized visual areas, computational neurobiologists spoke in very vague terms about these concepts, if they spoke about them at all.

Marr made a very insightful comment about Horace Barlow's first dogma.

I do agree with one of the thoughts behind this dogma, namely, that there is nothing else looking at what the cells are doing — they are the ultimate correlates of perception. [Editor's italics]

Vision, p.336

David Marr's book, Vision, the Bible of visual computational neurobiology and one which everyone admires yet few understand, makes no mention of the subject of parallelism or of the separate visual areas or of the specializations in the visual cortex, all demonstrated years before he published his book. It is a curious fact that computational neurobiologists should have been so late in seeing the significance of parallelism in vision, even though they now try to make out that they had been there all along. This may be due in part to the prevalence of serial, rather than parallel, computers until relatively recently,- it may be due to a relative ignorance among computational scientists about the precise characteristics and capabilities of parallel computers.8 Whatever the real reasons, it should make us all a little careful in accepting cortical theories derived from those whose tools are mathematics and computers rather than brains. The proper way of understanding the brain is to study the brain.

This is not to say that neural networks and computational neurobiology will not have an important role in cortical studies of the future. There is no doubt that they will and the relationship between brain studies, computational neurobiology and neural networks is therefore worth investigating briefly. One problem with computational neurobiologists is that they are far ahead of their time and yet far behind the times. They are far ahead in believing that neurobiology will have to depend upon theories generated by studying how the nervous system might undertake a task, which implies defining the task precisely, something which neurobiologists haven't been particularly good at. But they are far behind the times in believing that computational neurobiology can come up with theories that are of direct relevance to the neurobiologist, ones which may be put to the experimental test, by ignoring the facts of the nervous system. Marr's book illustrates this perfectly. It is rightly prized for introducing a new way of looking at what the nervous system does, of defining its tasks. But at the same time, by ignoring the facts of the visual cortex, it renders the book of very limited value to anyone who might want to understand how the cerebral visual cortex might be undertaking the tasks thus defined, how it may be functioning. This ignorance derives in part from the profound contempt with which computational neurobiologists regard the experimental neurobiologist — well illustrated in Marr's ignorance of what has come to be one of the most powerful doctrines dictating work on the visual cortex — and in part from the fact that not nearly enough is known about the brain. To be of use to the neurobiologist, a computational approach has to rely on the facts of the nervous system.

A Vision of the Brain, pp.118-119