We Know It When We See It What the Neurobiology of Vision Tells Us About How We Think

Understanding the eye is only the first step to understanding how we see. Vision is as much about perceiving, which means understanding how the brain works. We Know It When We See It is the swansong of neuroscientist and ophthalmologist Richard Masland, who passed away in December 2019. His research career spanned over four decades and he was lauded internationally for his work on retinal neurobiology. In this book, his love for teaching shines through, and it was a pleasure to join him on one last trip through the neurobiology of vision.

Masland follows the process of vision from its start when light enters the eye and hits the retina, to the processing of information in the brain's visual areas. This logical start-to-finish approach is as much a journey through the history of scientific discovery as it is a story that proceeds from well-established fact to informed speculation.

One of the things that I never appreciated about the eye is just how complex the retina is (until you realise that it is effectively a part of the brain that is pushed out into the eye during development). Masland introduces the various types of cells, how they were discovered, and how they function. Three layers of cells – photoreceptor cells, bipolar cells, and ganglion cells – are interlinked at two levels by horizontal cells and amacrine cells. All of these exist in many types, not all of which we understand. Interestingly, no single type numerically dominates. Instead, the retina contains roughly equal proportions of each type. Some are sensitive to motion, others to light-dark contrasts, that is, edges. Some are densely clustered, others spread out more diffusely. The result is that the eye does a surprising amount of pre-processing before passing information to the brain: "the visual image is decomposed by the retina into roughly thirty parallel streams, each reporting on its own specific feature of the visual world" (p. 87).

What happens next is where things get more complicated, but also more exciting. As Masland points out, the vision of vision that he describes here is at the frontier of developing knowledge and deviates from the textbooks. The optic nerve carries information from the retina to two brain centres, the superior colliculus and the lateral geniculate nucleus or LGN. The former is charged with visual orienting: what part of the image presented to the brain is of interest, and what should the eyes focus on next? The latter passes information to the visual cortex at the back of the brain. But the LGN is not just a relay station. A large number of neurons also return information from the visual cortex, forming an "immense feedback circuit" of uncertain function.

The visual cortex in turn is not a single entity, but a patchwork of visual areas distributed throughout the brain. Masland frankly admits that from this point forward we really do not understand things all that well. Everything seems to communicate with everything else. He highlights some of the areas we do understand, and I found those dedicated to the recognition of faces particularly interesting. As with the retina, information here is broken down, with different cell types responding to different elements: some to the pattern of two neighbouring dots (eyes), others to the presence of a nearby vertically oriented line (a nose). "It has been proposed that what these cells are doing is measuring a selection of parameters [...] to decide whether something is a face" (p. 115). The system is not fool-proof, and though Masland does not explicitly mention it, it explains the phenomenon of pareidolia: the seeing of faces in everyday objects.

From here, Masland goes into the fuzzy concepts of perception, object recognition, and memory formation. Vision is as much about seeing as it is about remembering: we learn to recognize objects. This is the result of neurons forming connections, so-called Hebb synapses, after repeated activation together. Someone once memorably phrased this as "neurons that fire together wire together". This wiring together forms nerve nets that feature as much in neuroscience as they do in machine learning. As Masland shows, these two disciplines meet each other in the challenge of understanding vision. The formation of nerve nets based on modifiable synapses is also an incredibly elegant evolutionary solution to one of life's conundrums. How do you develop a perception machine without knowing in advance what you will have to learn to recognize throughout life? Life's answer: develop a general pattern recognition machine. A final topic Masland considers is the transition of perception into thinking and consciousness. This is a grey area, and he gladly admits to entering speculative territory as he attempts to give the reader a rough idea of what we do understand.

This exhilarating dive down the rabbit hole of neurobiology is lightened up by the occasional biographical sketch of key players: Donald Hebb of the above Hebb synapse, student-of-the-retina Brian Boycott who is commemorated by the Boycott Prize, connectome-pioneer Winfried Denk who has developed mind-blowing microscopy techniques, and several others. Masland's forte is to not drown you in technical detail, giving you the principles instead. And by adding the daily grind at the lab bench, he brings you back down from the heady territory of theoretical neurobiology. Appropriately for a book about vision, what helped me grasp the ideas presented here were the numerous, small, inline illustrations, the way a teacher might combine writing and drawing on a blackboard. This feels like a lost art form, but interrupting your text with a small doodle or graphic can be incredibly helpful.

Amidst all this, Masland decidedly underplays the applied side of his work. He was interested in clinical applications and, as per his webpage at the Harvard Medical School Department of Ophthalmology, "a potential therapy for blindness, based partly on his work, is currently under clinical trial". It will fall to other writers to celebrate his achievements in these fields. His obituary in Neuron mentions that, amazingly, he wrote this book from start to finish after having been diagnosed with cancer. While I am saddened he will not enlighten us further, we can all be grateful that he was able to finish this book.