Now, in the original experiments done by Hubel and Wiesel and then many experiments done by many hundreds of workers since then. we know that if such a micro electrode is inserted into the cortex vertically, that is, in the radial axis of the cortical microcircuit. What we find is a series of neurons that we might encounter at different depths through the cerebral cortex that pretty much all report the same aspects of the visual stimulus. That is if we plotted their spacial receptive fields we all we find that all of these spatial receptive fields more or less overlap, across the depth of this recording. And when we look and test for the orientation selectivity of these cells, we find that they are all pretty much sharing the same orientation preference. In this example, again, all preferring angles very much near 90 degrees. Now a different story would be observed if the angle of the electrode penetration is made much steeper. And that's what's shown in this experiment. So now the micro electrode is coming in the cortex at a very sharp angle, such that we are passing across a set of cortical micro columns progressing from one to the next. And what we find when we move across the cortex in this way, is that receptive field position slowly begins to drift from one location through adjacent locations until finally we reach a separation of a full receptive field after we advance a micro electrode about a millimeter or so in the cortical tissue. If we consider what happens with the advance of that micro electrode with respect to the orientation selectivity of the neurons, we find a similar shift. In one position, we may find neurons that prefer horizontal. And as we progress across the cortex, moving, let's say, in hundred micron increments, what we might see is a gradual progression of orientation preference, as we move from one sight to the next. And finally, we may discover that after movement of about half a millimeter or so, that we might have a tuning function that is entirely non-overlapping with the first neuron recorded, in this series across a penetration. So this suggests that there is a columnar organization of circuitry where all the neurons in the column tend to respond to the same stimulus, and that this columnar circuitry is organized in some systematic fashion, such that neighboring locations across the cortical surface are representing nearby aspects of the visual stimulus. And there is a smooth progression of change that occurs. Now, since the time that David Hubel and Torsten Wiesel did their experiments that earned them the Nobel Prize additional methods have been developed. Some of which allow neuroscientists to look at large scale patterns of activity. Not at the single cell level, but at the level of large populations of neurons over several millimeters of visual cortex. One such method Is known as optical imaging of intrinsic signals and it's based upon the fact as brain regions become active, they draw oxygen off of the hemoglobin that's being supplied to local regions. That creates a bit of oxygen debt, and as a consequence arterials dilate and blood flow increases to those regions. And there is exquisite coupling between brain activity and blood supply that makes these signals a robust indicator of neural activity. Now the method in question is called optical imagining, because we don't need radioactivity as we would with PET imaging. We don't need a magnet as we would with MRI imaging. rather, using experimental animals and in patients undergoing neurosurgical exposure of the brain, it's possible to directly illuminate the brain surface with light and record the subtle changes in the absorbance of light that happens as this oxygen and blood flow dynamic changes in brain tissue. So what you're looking at here is the results of an experiment that I actually performed. And in this study, we looked at the responses of the visual cortex in an experimental animal where the cortex is exposed to light and visual stimuli are presented on a screen in front of the anesthetized animal. So this panel that we have here in the upper right hand side of the image is a view of a the living brain and all the little lines that you see, these are all blood vessels that are supplying the brain with the blood that it needs to be active. Now if we take an image of the brain with no stimulus on the screen and compare that to an image taken while a stimulus is presented, we can subtract those images and look for the very small change in the absorbance of light that is a reflection of brain activity. When we do, we have the image that is shown here on the left. So we're looking at roughly about 8 millimeters by 6 millimeters to the visual cortex. And what we see are a series of dark spots and small bands. And these dark spots and bands are cortical columns that were activated by the stimulus. Now it's interesting to look at the literature from the time of Hubel and Wiesel's original writings back in the 60s and 70s and see how even though they were probing the brain with micro-electrodes, they still had a sense of the organization that this method makes now quite clear. And that is that there are columns of neurons that are responding to the same visual stimulus that are distributed across the visual cortex. And columns shift their preferences gradually from 1 location to the next. Such that, if we compared the responses, let's say, to a horizontal stimulus, with those to, let's say, a vertical stimulus, what we would find is that the columns that are present in these 2 images precisely interdigitate. So the image we see here in the upper right is the result of the subtraction of the images acquired while the animal viewed horizontal stimuli and vertical stimuli. And it's a way of looking at how selective is the response. That is how well can we differentiate the responses to one visual angle horizontal, from another vertical and we can repeat this method over and over again with different kinds of stimuli different pairs of orthogonal stimuli and generate a family of images that reflect the interdigitation of columns that are representing orthogonal orientations. Now there's a different way to represent this data that I'd like to show you here by looking now at a dynamic representation of a set of these images where we will just loop around and see how the illusion of motion in these images gives us some insight into the organization of the functional map for orientation preference. What I'd like to show you are some data from my lab and the lab of my colleagues. Where we've looked at different experimental animals and have measured the distribution of orientation columns in their brains. And we've created an animation that runs through all the images that were required as the orientation of the visual stimulus was changed. So if I start this animation, and it will just run over and over. What you can see, I think, is this beautiful dynamic array of functional representation in the brain. Of course, it looks like something's moving in the brain, nothing's moving in the brain. What is changing is the distribution of functional activity over time as the visual stimulus is rotated. And one point I would like for you to appreciate is really how similar these patterns look in three very different species of animals. the Treeshrew is a small animal that makes its home in Southeast Asia. The Galago is a Prosimian primate from Central and South America. And the third of course is a domesticated carnivore. So we're looking at the visual cortical patterns from animals that diverged about 65 million years ago. And yet there are very similar patterns and seemingly a highly conserved cortical network that's giving rise to these patterns. And if we look in even more detail I find it interesting that we can localize certain locations in these images where it looks like these dark domains of activity are swirling around small points, or singularities. And we call these points pinwheel centers. And it appears as though the organization of this representation of orientation preference is one of a large collection of pinwheel structures that are arrayed across the visual cortex. Well one way that we can represent this kind of a dynamic set of experimental data in a static fashion is to perform a little bit of image math and then apply a color code which allows us to show what we call the map of orientation preference. So we're looking at the very same region of the visual cortex in an experimental animal. And now we're using color to represent the preferred orientation at each location in the visual cortex. So for example, where we see the color red is representation of horizontal visual stimuli. So these are all cortical columns or bands of cortex that prefer horizontal among all possible orientations. And where we see blue regions in the brain, or rather brain regions that have been colorized blue, these are cortical columns that are representing near vertical. And between red and blue are all possible orientations that are represented in the brain. Now, the dark bands that we see, these are structures that we can't resolve the preference. But with newer techniques that allow us to see the activities of individual neurons in the living brain, we know that these zones are really not dark at all. Rather, they have cells that represent orientation just as robustly with as sharp tuning functions as any other place in the brain. But rather these cells with different preferences come together in very tight spaces. And with this method, we can't resolve their differences but with a cellularly based method of resolution called two photon imaging of calcium signals, it's possible to detect the preferences of neighboring neurons. Well when we do so, what we discover is that there are locations such as right here, right here, and there, for example, where there are point singularities, around which a set of cortical columns are arrayed in a radially symmetrical fashion. And at that point singularity, nearby neurons have very different orientation preferences. Such that we're looking at the center of one of these pinwheel formations. Well, I can go on and on about this as my colleagues and I have in our research. I won't belabor this point any further. But rather, I wanted to give you a bit of an inside look at the circuitry and the physiology of this wonderful part of the mammalian brain, the primary visual cortex. And all of this really has just been to illustrate one computational property, and that is the property of orientation preference. Now we and others have done experiments where we've plotted out the activity maps, not just of, let's say a horizontal bar moving along a particular axis of motion. But let's say, a horizontal bar that's either drifting in just one direction or drifting in the opposite direction. And what we discover is that we may find a domain in the brain that represents a horizontal stimulus but that domain may be subdivided. And across half of its region, there may be cells that prefer upward motion of that horizontal bar. And in the other half of that domain, we may have a clustering of neurons that prefer downward motion of that horizontal bar. This is what we call direction selectivity. And as Hubel and Wiesel discovered, there are numerous neurons throughout the visual pathways that respond best to motion presented in just one direction with much weaker responses, if at all, in the opposite direction. So, neurons in the visual cortex can be characterized by their orientation and directional selectivity.