Unravelling the spindle

optic tectum on one side of the brain and found that the axons from the retina that were severed in the surgery sprouted new projections which crossed over to terminate in the optic tectum on the same side of the frog's brain. When he examined the visually guided behaviour of these unfortunate creatures, Ingle found that they showed 'mirror-image' feeding — directing their snapping movements to positions in space that were mirror-symmetrical to the location of the prey objects. They also showed mirror-image predator avoidance. These results provided pretty clear evidence that the optic tectum has a critical role in the visual control of these patterns of behavior in the frog — and also converged nicely with our own findings in rats. But, remarkably, not all the visually guided behaviour in the rewired frogs was mirror-imaged. Ingle found that the same frogs showed quite normal avoidance of obstacles as they hopped from one place to another. As it turns out, the reason they showed normal visual control of obstacle avoidance is quite straightforward; the retinal projections to the pre-tectum, a structure in the thalamus just in front of the optic tectum, were still intact and had not been redirected to the opposite side of the brain. (Not surprisingly, lesion studies in frogs and rodents have since shown that this structure has a critical role in the visual control of obstacle avoidance.) Thus, Ingle argued, there are at least two independent visual pathways in the frog: a tectal pathway, which mediates visually elicited prey-catching and predator avoidance, and a pre-tectal pathway which mediates visually guided locomotion around barriers. Ingle's experiments, like Schneider's before him, suggested that there is modularity in the organization of the visual pathways of the vertebrate brain. But the modularity that Ingle was talking about is very different from the 'what' versus 'where' story. In short, Ingle had demonstrated the existence of different 'visuomotor' modules, not simply visual modules. In fact, it turns out that there are five or more visuomotor modules in the amphibian brain, each with its own set of retinal inputs and each controlling different arrays of motor outputs. There is no single 'multi-purpose' representation of space residing somewhere in the animal's brain; instead, visual input about spatial location is transformed in different ways for different purposes. Ingle's paper was a watershed in my own thinking about the organization of the vertebrate visual system. It …