Microsaccades: Our Visual Framerate

in Opinion
February 25th, 2012

Dr. Frank Werblin at UC Berkeley has dedicated nearly his entire academic life to the study of the eye and visual processing. More recently Dr. Werblin has completed his model of the retinal processing system he has deemed “The Retinal Hypercircuit”. The Hypercircuit itself is made up of the five classical retina cell types: Photoreceptor, Horizontal, Bipolar, Amacrine and Retinal Ganglion Cells, but more recently, a collaborative effort has identified over 50 morphologically different cell types. Of this vast array of unique cell types the most variance falls in the morphology of the Amacrine cells, which offer horizontal properties in the Inner Plexiform Layer between the Bipolar and Ganglion Cells. Although the mechanics behind the Hypercirtuit are fascinating, what I find arguably more important is the output of the system, a topic which Werblin has indirectly stumbled upon, but which I believe could potentially lead to an incredibly progressive line of research.

Amacrine Cell Types

Amacrine Cell Types

In his studies, Werblin has stumbled upon a couple novel properties of Ganglion Cell output which may hold more information about visual processing than meets the eye. First, he has discovered that the stream of visual information to the brain is actually not a “stream” at all, rather a set of discrete pieces of information sent at different times and with different inherent signal properties. Another finding was that each of the 12 morphologically distinct Ganglion Cells processes only a certain type of stimuli. Perhaps, as an example, one type of Ganglion Cell processes the signal that “something is looming” while another “something is moving to the right”. Although this organization is indeed the work of the Hypercircuit and is mind-blowingly complex, Werblin has shown that the signal leaving the each Retinal Ganglion Cell encodes a certain stimulus quality, discrete in time and purpose. There is a problem with this theory though: We actually perceive a “stream”!

In a recent e-mail correspondence with Dr. Werblin I asked whether there were certain signals that were processed more quickly than others. If you think about it this must be true. When someone jumps out of a bush at night and scares you, you are not at first aware that the person is wearing a red coat, rather that they are there, human and dangerous! This was indeed the response I received from Dr. Werblin. He said enthusiastically, “YES SOME MORE QUICKLY THAN OTHERS” and further that they are probably organized in an evolutionarily advantageous processing order. But again, there is a problem with this, we perceive stream…And not only that but another potent flaw in this discrete logic: If we were to be a constant delay between visual cues, say, we perceive “thereness” before we perceive “color” from birth to death that delay would be constantly growing. Hypothetically, if there were a 0.001ms difference between when we perceive “shape” and “color”, for every second of life we would be adding 0.001ms to the total gap between our perception of shape and color. Visually, we could represent this as two divergent graphs, each having a slope corresponding to the rate at which we perceive that discrete visual quality. As these graphs grow over time they are divergent which means that if our system worked in this simple discrete manner then by late life we would be perceiving the shape of a stimulus, only to receive the color information a day or so later.

Divergence of Visual Stiumulus Quality Arrival Time

Divergence of Visual Stiumulus Quality Arrival Time

So, in essence, it cannot be this simple, there must be something we are missing. Sadly, no research has ever been dedicated to answering this question so for now we can only guess and I think I have a solution that works.

Remember, the visual information is discrete in time and quality. The cut of visual information from the retina to the brain, or what actually makes the signals discrete, is done by a short eye movements called a Microsaccades which occur thousands of times per second. I believe, and again this is all speculation, that these Microsaccades have developed to sync up all of our visual data from all of the different Ganglion Cells. I think that they function to hold the feed of visual information until all of the visual qualities have been analyzed by their respective locations in the brain and the proper reactions have been then initiated. Once our brain has reacted to one of these discrete packages defined by microsaccadic borders our eyes can let up, the saccade can stop, and our eyes can again be soaked in a new set of electromagnetic data for analysis. If we could further analyze which qualities are encoded and sent by the retina in what order we may be able to extract information about downstream processing or evolutionary importance, but until then this idea may have to lie in the realm of mere hypothesis.

Again, I have to stress that this is speculation based on emergent data not intended for this use, but, as a closing remark I would like to mention one thing: When I posited this idea to Dr. Werblin himself he responded with simply, “MAKES SENSE!!!”. Further, for anyone interested in the more detailed workings of the hypercircuit, on The Werblin Lab’s website, which is linked just below, there is a visual walk-through detailing the entire logic and processes behind and within it.

Werblin Retinal Hypercircuit – Werblin Lab

Tagged , , , , , , , , , , , , , ,

Post Your Comment