How the eye anticipates

Though light takes some time to reach the eye, there is a system by which animals anticipate the movement of a moving object. Now, scientists can explain it

 
Published: Wednesday 30 June 1999

 There is more to vision than< it takes anywhere from one-thirtieth to one-tenth of a second for a visual stimulus falling on the retina of the eye to evoke neural activity in the brain. This time is long enough for a rapidly moving object to cover a fair distance, implying that when we see such an object, it should appear to be some distance behind its actual location.

However, both everyday experience and careful experimental measurements indicate that this is not the case: humans are quite good at spotting where a moving body actually is. This suggests that there must be a mechanism that helps our brain use the information that something is moving at a certain rate to constantly extrapolate, or predict, where the body will be in the future. This means that what we 'see' is really the anticipated position of the object, rather than its actual position at that instant.

An elegant technique that demonstrates this is to let a moving bar of light sweep across a subject's retina. At some instant a second, fixed bar is flashed briefly in coincidence with the moving bar. When asked what they see, the subjects invariably answer that the flash was behind the moving bar. It turns out that this form of visual compensation works within speeds of about 0.3-0.9 millimetre per second (mm/sec) on the retina and breaks down by 4 mm/sec. (To get a feeling for these figures, a speed of 1 mm/sec implies that in bright light, an image can flash across the pupil of the eye in about one second. Failure to compensate would correspond to making a 10 per cent error in estimating the actual position.)

M J Berry and colleagues at Harvard University have recently shown -- rather unexpectedly -- that the process of anticipation can begin right in the retina itself; it does not have to wait until the brain gets to work ( Nature , Vol 398, p334-336).

The experiments were carried out with isolated retinas from an amphibian known as the tiger salamander. Computers were used to generate appropriate visual stimuli, which were presented onto the photoreceptor cells of the retina. Electrical recordings were made from ganglion cells. These are also in the retina, but are situated two cell layers behind the photoreceptors. In the visual system, these are the first cells in the pathway that generate the characteristic spike-shaped electrical 'firing pattern' of nerve cells. Recordings of electrical activity were made by placing the retinas with their ganglion cells in contact with a series of conducting electrodes.

As expected, a moving bar of light caused a moving train of electrical spikes to travel along the ganglion cells. But the profile of the electrical activity showed up something simple but unexpected. The activity was a maximum not at the centre of the bar but rather at its front edge, where the image profile just begins.

The Harvard group also suggests a model for how the retina accomplishes this feat, a model that basically assumes knowledge of the speed of movement. Irrespective of the correctness of the model, here is evidence for an early step in visual perception that makes sure that a frog can direct its tongue to the exact location of a fly.

Subscribe to Weekly Newsletter :

Comments are moderated and will be published only after the site moderator’s approval. Please use a genuine email ID and provide your name. Selected comments may also be used in the ‘Letters’ section of the Down To Earth print edition.