A sum of destructions

Pablo Picasso once said “To me painting is a sum of destructions. I paint a motif, then I destroy it.” Unknowingly, he had an intuition about visual processing. In fact, our current understanding is that retinas, quite like Picasso, break an image into its parts. The first man to lay the foundation of this idea was Haldan Keffer Hartline, a contemporary of Picasso.

Fig 1. Eye of Limulus Polyphemus, or horseshoe crab.

Fig 1. Eye of Limulus Polyphemus, or horseshoe crab.

In the 30s and 40s, Hartline was inspired by the work of another great neurophysiologist Baron Adrian who was the first to record neural activity from single nerve fibers. Hartline learned this technique from Adrian and used it to investigate how individual retinal cells respond to light stimulation. To do so he chose a model organism that ironically is named after the cyclops in the Odyssey: Limulus Polyphemus. Ironically because this invertebrate arthropod in fact has multiple light sensitive organs, two of which are compound eyes (a common structure in invertebrates). If we take a closer look at a cross-section through one of these eyes we see that these eyes are composed of many parallel light processing units, called ommatidia, each with a tiny lens, light detector cells, and an output neuron with a large axon that Hartline could record from (fig 1).

Hartline would dissect out the retina, place the electrode on the axon, and periodically increased the light intensity, or luminosity as he called it, over the course of a few seconds. The first test he did was a series of stimulations in which the luminosity increased in a stepwise manner (fig 2). He learned that the fold change in light intensity is encoded in the rate of nervous impulses. He also observed that the pattern of response shows a higher rate at the onset of stimulation, followed by a relaxation to a lower steady firing rate indicative of some form of adaptation to the stimulus.

Fig 2. Responses of the eccentric cell of the ommatidium to step-increasing light intensity. Arrows indicate the time of stimulus onset (light-on), the subsequent decrease in firing rate, and stimulus offset (light off).

Fig 2. Responses of the eccentric cell of the ommatidium to step-increasing light intensity. Arrows indicate the time of stimulus onset (light-on), the subsequent decrease in firing rate, and stimulus offset (light off).

Fig 3. Eye of the Rana Catesbiana, or American Bullfrog. Photoreceptors are shown in green, interneurons in red, and ganglion cells in purple.

Fig 3. Eye of the Rana Catesbiana, or American Bullfrog. Photoreceptors are shown in green, interneurons in red, and ganglion cells in purple.

These discoveries motivated Hartline to investigate whether similar principles of information processing could also be found in the vertebrate visual system. To do so he went to the American Bullfrog. The bullfrog, like all other vertebrates and cephalopods, and unlike Limulus, has a simple eye (fig 3). Here light comes in through a pupil and a single large lens to reach the retina. If we look at this tissue in cross-section we can see a beautiful laminar organization. All the way at the back are the photoreceptors. They signal to interneurons (central layer) that then synapse onto the retinal ganglion cells, the output projection neurons of the retina. It is from the axons of these output cells that Hartline recorded light responses.

He first tested these cells for their response to the same flash-of-light stimulus he had used in Limulus. He immediately noticed a striking difference. Some responded to light onset in the same way as the output cells of the ommatidium. Others, however, responded to light offset, and a final group responded to both light onset and offset. These different responses have become famous as ON, OFF and ON-OFF, three classes of retinal ganglion cells that have since been found in many other vertebrate eyes. Hartline realized that there are different types of retinal ganglion cell types that encode features of the visual scene. This notion laid the foundation for the study and discovery of many specialized cell types in the retina.

Fig 4. Responses of different types or retinal ganglion cells in the eye of the bullfrog. The red lines indicate the time of light offset.

Fig 4. Responses of different types or retinal ganglion cells in the eye of the bullfrog. The red lines indicate the time of light offset.

In further experiments, that I won’t show here, he recorded from cells while presenting them with a variety of local and widefield stimuli. He became interested in how cells are affected by their neighbors, and returned to Limulus to study this further. As he had done before, he recorded from the eccentric cell axon while shining light onto its photosensitive cells, and then onto the surrounding ommatidia (fig 4). He observed that stimulation of surrounding neurons reduced firing in the recorded cell, a phenomenon he termed lateral inhibition. This mechanism enhances the visual system’s ability to see edges and create contrast.

Fig 5. Lateral inhibition in the eye of the Limulus. The diagram on the left shows the ommatidium that Hartline was recording from in blue, stimulated by constant light. In green are surrounding ommatidia onto which he shone light during an interval of a few seconds. On the right the response of the recorded ommatidium changes (the cells shows a decrease in firing rate) during light stimulation of the surrounding cells.

Fig 5. Lateral inhibition in the eye of the Limulus. The diagram on the left shows the ommatidium that Hartline was recording from in blue, stimulated by constant light. In green are surrounding ommatidia onto which he shone light during an interval of a few seconds. On the right the response of the recorded ommatidium changes (the cells shows a decrease in firing rate) during light stimulation of the surrounding cells.

Hartline’s work gave us several discoveries that have been greatly important in neuroscience. His observation about intensity being encoded by the rate of nervous impulses has since become one of the founding principles of visual and sensory neuroscience at large. Cell types have been characterized morphologically and molecularly as well as functionally in many neural systems. Inhibition has been shown to play a key role in the ability of retinal circuits to dissect visual scene into features. Research in the decades following Hartline’s work have shown that these principles apply to other sensory systems as well, and even to higher order cortical circuits.