Now, my colleagues and I were quite interested to understand how do these two properties of orientation preference and direction preference develop conjointly in the very same animals. And we thought that the way to begin to study this question was, was simply to follow the development of a cohort of animals that were raised under our normal laboratory conditions. And when we did that, what we discovered is displayed here in this slide. what I'm showing you are representations of orientation selectivity and direction selectivity acquired from the very same animal within this cohort. And what we did was we sampled these animals at different stages of development following the time of natural eye opening, which is right around the end of the first month of postnatal life. What we found is that when we looked at animals right around the time of natural eye opening, orientation preference was present, as indicated by this series of orientation columns interdigitating with one other. But there is virtually no evidence of a differential response to opposite directions of motion that began to be visible after a few days of visual experience following eye opening. But it really wasn't until about seven to ten days later that we began to see this differentiation of our orientation columns into domains that prefer opposite directions of motion. So the upshot here when we looked at the developmental curves is that while orientation preference is present, at or just before eye opening and continues to develop throughout the first few weeks of life. There seems to be a real lag in the development of direction preference. There seems to be about a week or so of visual experience that's necessary before there is a steep inflection in the development curve. And then, after about two to three weeks of postnatal life, at least in this animal model mature levels of orientation and direction selection activity are established. So the fact that there seems to be this lag in the development of direction preference suggests that, perhaps, vision in this early period of postnatal life is actually doing more for the development of direction preference than it is for orientation preference. Well, in order to test this idea we did similar kinds of experiments, raising animals under normal conditions, and comparing them to animals that were raised without vision. And here are some data that represent the cohort of animals that were raised with normal visual experience and these animal develop robust cortical maps for orientation, selectivity and direction selectivity. And these symbols here represent the measurements that we're able to obtain from these animals reflecting the degree of development of selective responses to visual stimuli that varied in orientation and direction of motion. So this is a picture of what normal development looks like. When we raised animals in darkness, as I've just told you, these animals do, in fact, self-organize a map of orientation preference. But the degree of selectivity is less than what we would see with the full benefit of visual experience. Well, it was really surprising to us, and actually, quite unexpected was that there was virtually no sign of the emergence of direction selectivity in these animals. In fact, quantitatively speaking, these symbols are clustering around zero, which is equivalent to the absence of a visual stimulus. And this was quite a remarkable dissociation of the development of the circuits that compute directing selectivity from those that are responsible for, at least the essential computations that compute orientation selectivity. Well, we wanted to know at what time in development was this network especially sensitive to the onset of visual experience. And what we found is that if we deprived animals of vision for the first couple of weeks after eye opening and then gave them what we might call late vision. We found that their orientation preference maps returned to their full developmental strength. And you can see that by eye with the robust dark and light distribution of orientation domains in the maps from these animals. However, the data that represented the development of direction selectivity really remained no different than those that were completely deprived throughout the developmental epoch. In fact these data sets are no different from one another, and statistically speaking, not different from the comparison of maps taken from animals with no visual stimulus. So, what we found is that there remained severe impairment in the circuits that would compute direction of motion in the absence of early vision. In fact, it's really only those animals that receive the benefit of early vision that then go on to develop direction domains indicative of maturation of circuits that are computing direction selectivity. This cohort of animals were deprived of vision up until the time of natural eye opening, and thereafter, given a couple of weeks of visual experience. And these animals are requiring strength or robustness of visual computation equivalent to animals that were simply raised under normal laboratory conditions right from birth. So these data suggest quite powerfully I think, that the circuits that compute direction selectivity are indeed more sensitive to early vision than are those circuits that self-organize in order to compute orientation selectivity. And it suggests that there is a very early critical period in visual cortical development when normal visual experience is essential in order to develop this circuitry that allows us to compute movement to one direction or another. You can imagine how important that is for survival in a world that is full of objects that are moving and indeed, as we move through the world we generate our own patterns of visual motion and visual cortex. Because there is this very early critical period then for the development of these circuits, we wondered if it might be possible to actually train these circuity with structured vision. And so, this allows us to get at the same problem, really from a complementary direction. Rather than depriving vision, we want it to know whether we could provide some structured vision and impact the developmental course of the circuits that are necessary to compute this property. And indeed, that turned out to be the case. So what we did is we prepared some animals for our experiments. And while they were anesthetizes and prepared under stable physiological conditions, we were able to expose them to a, a regimen of visual training. And we asked whether it might be possible to actually observe the development of these circuits in the brain. And indeed, as we looked over an experiment that lasted about a day, what we saw is that after about eight to ten hours of visual training, we began to see the emergence of structure in our selectivity maps for direction. Indeed, we begin to see the emergence of these light and dark domains that are beginning to take shape in a fairly robust way after about 12 hours of visual training. So, what was especially remarkable to us is that, at the end of the experiment, when we tested these animals for their responses to all possible directions of motion, what we found is that the differentiation of direction domains was really only present along the axis of motion that was used for training purposes. So this provides perhaps some of the best evidence that we know of in the field of neuroscience. That visual experience can actually instruct the development of circuits in the, in the cerebral cortex. Now we've gone on to study this phenomenon at the cellular level using this method I mentioned a few moments ago, to photon imaging of calcium signals. And this has allowed us to watch cell by cell the emergence of direction selectivity and direction preference in the visual cortex. And what we're able to do is to follow the same cells over time here we're looking down in a very small region of the visual cortex and after about three hours of visual training, we can observe the very same cells in the living animal. here, we can identify many of the same cells that are giving rise to a fluorescent signal after we've provided a die that is taken up by these cells, and can report neural signals based on calcium fluxes that are present within these cells. And in the initial characterization, what we find are fields of cells that have some degree of direction preference. So these are three different animals here and what we're looking is their responses and preferences before training. And then, after about three to six hours of training, though, what I hope you can appreciate is that, in each of these cases, there is an increase in the numbers of neurons that prefer the direction of visual training. In this top case, training was done with motion to the left. And what we see is an increase in the numbers of neurons that prefer that leftward motion. In the middle case the training stimulus was moving down to the right. And, we see a modest increase in the numbers of cells in the strength of selectivity indicated by the size of the arrows. In this field of neurons following this bit of training with a bias towards movement down and to the right. And the same is true in the third case, which already had a fairly significant bias of cells that preferred rightward motion and then after three to six hours of visual training. those cells became even stronger in their preference and a few more emerged. So when we looked at our entire data, data set before and after we saw indeed a statistically significant increase or a rightward shift in this cumulative pot of neurons that acquired the direction that was presented by the training stimulus. So what has this taught us about the role of visual experience and the development of direction preference. Well, what we've learned is that the neural circuits that underlie direction columns don't seem to have the capacity to self-organize in the way that those circuits that were responsible for orientation columns self organized. Rather, the differentiation of those circuits to compute direction selectivity must be instructed by the presence of a visual stimulus. We've also learned that the window of opportunity for the benefit of this motion training is very brief. so this means that early experience is absolutely critical to avoid what might be a long-lasting impairment in the ability to differentiate directions of visual motion. I would suggest that in order to understand the construction of neural circuits in the brain, we need to examine multiple circuits that exist at different stages of cortical processing. We shouldn't just rely on one gold standard. despite how valuable that standard might be for a certain line of investigation. We should always be cautious about generalizing the principles that we've learned from just one model system to another. We should be somewhat leery of this two-phase view of development, that is, this idea of a pre-critical period followed by a critical period. I think that for many circuits in the brain, I would suggest that this two-phase view will not be adequate to account for the influence of experience on brain development. Thus, sensorimotor experience can influence the construction of neural circuits. As soon as the brain becomes responsive to environmental signals. So to put this all together, I would suggest that genetic specification, self-organization and sensory motor experience interact concurrently, not just sequentially. But as soon as all of these influences could possibly interact with one another, they are competent to do so. And this interaction shapes the ongoing development and refinement of neural circuits in an early phase of life and potentially across the entire life span. Well, what are some implications of these findings? I would suggest that one implication that should not be lost on us is that normal sensorimotor experience has a profound effect on the formation and the maturation of neural circuits in the cerebral cortex. Especially those circuits that reflect the computations performed by networks of neurons that sit just beyond the early input layers of the cortex. They're very likely many properties of cortical networks that depend upon sequence or timing based signals like direction selectivity. That may not develop at all, without the benefit of normal experience in an early critical period. And this is perhaps, the strongest implication of the studies that I've showed you today. I think what we've discovered is that abnormal sensorimotor experience In these early critical periods, can lead to lasting functional impairment. And that abnormal experience might be the absence of experience, or it more likely is going to be experience that is filtered through some kind of peripheral abnormality in the formation of the sensory organ or in a musculoskeletal unit that will impact the way that organ is operated in early life. So to sum this all up, I would say that normal sensorimotor experience in early life is truly critical for lifelong health and wellness.
