Videos

Stefan Treue

Stefan Treue

Professor, Director of the German Primate Center - Leibniz Institute for Primate Research

Who does not fear attention disorders? How many times did we overlook what was right before our eyes? Many are the paranoias of a society drowning in a deluge of information. While we know that attention disorders, like the attention-deficit hyperactivity disorder (ADHD), affects more than 4% of the population and are connected to many neuropsychiatric disorders, the neural circuits and computations underlying attention are still poorly understood. However, it is Stefan Treue, Professor for Cognitive Neuroscience and Biological Psychology at the German Primate Center and Goettingen University, recently honoured with the Leibniz Prize for his experimental study of the primate visual system, particularly that of the macaque monkey, who is providing a more rigorous description of the correlates and signatures of attention in neural activity, and starting to identify the sources of attentional influences on neural activity and perception. In Berlin, Treue explained how he is successfully exploring the central influence of attention on our perception and contributing to an overturning of old ideas about information processing in our nervous system.

Breaking the Wall of Sensory Overload. How Primate Neuroscience Reveals the Mechanisms of Our Perception

Transcription

As we all know, life is a challenge. From the perspective of evolution, it is all about survival and maybe reproduction. Luckily, evolution has endowed us with, maybe not with a defective diamond, but the most complex organ that we know as of today, the brain, to help us in this task of facing this challenge. So, what do we have to do? We have to pick up information from our environment to decide how to act on it. This is not all that different from an artist, who takes a picture and renders the environment on a canvas. More technically speaking, our nervous system is at the interface between the environment that provides the surroundings and provides the information, and choosing amongst many possible actions the one that is most appropriate for survival or maybe reproduction in a given situation. Now how do we achieve this? What do we have to do? What we need are sensors, powerful sensors that pick up the information from the environment and convert that information into what we call a representation, an internal picture of our environment. Based on that internal representation, we can decide how to act. To really be good at this, we need powerful sensors. You can imagine; you can have a huge system with all the powers that you can think of, but without appropriate information about the environment it is of no use for us in terms of survival. Lets look at an example of a particularly powerful sensory system in action. I brought you a little video clip here from a real life situation of survival. If you look at this red fox here in an American National Park, it has to produce and reproduce, and it has to pick up information. It doesn’t seem to be all that much. But survival is at stake: he is very hungry; where is food? Watch his ears carefully. Something is going on. (Laughter) If you look very carefully at his snout, you see that lunch was captured, and he survived another day—thanks to enormously powerful ears that were able to hear the little rodent under many layers of snow. Now we don’t usually run around like this looking for mice, but we have very powerful sensory systems ourselves. Let me switch to the visual system from this example from audition and lets talk about us.

So what is behind our ability to see very well is, of course, not only very sophisticated sensors, but an equally sophisticated nervous system that can process the information that comes in from the sensors, because vision doesn’t end with the eyes. It is the information transfer into the brain. So, here, you see a side view of a human cortex. These arrows sort of indicate how information flows through our brains, because the visual information enters in the back of the head, and is then distributed across a number of areas that make up our visual cortex. If you think that vision is just one of the many tasks we do, it is just one of the senses, and there are many other things that we can do, you might be really surprised to realise that one third of our cortex is devoted to vision—one third of our processing capacity for a single sense. It tells you how important seeing is for us. Now the brain is much more structured than you can see here. If we look at the visual system with a little more detail, and I don’t want you to memorise this picture, but I want you to get this story that information that comes here in the back of the head is then passed through a series of areas to generate something that we call “perception”. The one wall that is breaking down over the last decade or so that I want to tell you about, is the interpretation that vision is mostly about a stream of information from an incoming area through a series of steps, as a one-directional kind of information processing. Let’s keep that thought in mind. These areas are very powerful. How do they achieve their ability to support vision? If we want to know what is happening on a cellular level, not just on the scale of the whole brain, we need to measure the activity of individual nerve cells in a living brain that is interacting—that is picking up visual information. So, we could do this in humans, but it is rare and is only under special circumstances. Instead, we are looking at the Rhesus Macaque monkey: a very similar system to ours, except he has actually devoted half of his cortex division, but that is just the better for us. So, what we want to do is record electrical activity from individual cells in the neo- cortex. Just to give you an idea of the scale here, this is a human hair, and this is a microelectrode that needs to be brought close to an individual neuron in a way that doesn’t hurt the animal; it doesn’t inflict pain on the animal, which works, because our brain doesn’t actually have any pain receptors. That is why it could also be done under special circumstances in humans. Now we can go in there and train an animal to do a visual task, to sit in front of a computer screen and essentially do a video game, tell us by pressing a lever, and that lever is connected to a computer, if the animal sees something to react to things on the screen.


I can tell you about these tasks within minutes. We have the problem that most of our monkeys don’t speak English or German, so we need to train them. The training takes months sometimes. During the training the monkey learns to solve a particular task with the visual information that is presented on a computer screen. Once the animal is trained, we can record the activity of single neurons in the brain and see how the activity of the single neuron represents the information that is presented in front of the animal on the screen. So, we can sort of watch the brain encoding the information. The monkey is rewarded whenever he is doing the task right with something to drink. You hear the clicking of the valve. Here is a view over his shoulder onto the computer screen. With time, you can build up quite complicated tasks where they have to respond to events on the computer screen. As I said, when the training is over, we can then start recording the electrical activity. You have to know, you hear it in the back there: this crackling sound are individual pulses what are called action potentials of individual nerve cells. They fire these action potentials to communicate. With a microelectrode we can pick them up in response to a particular visual stimulus. There is a lot of amplification and computer power necessary to do this, but then you can see, or you can hear, the brain in action on the level of an individual neuron. Let me show this little example of a neuron that we have been recording from. Before I start it, I need to explain to you what is happening. You are looking at a copy of the computer screen in front of the monkey. This sort of circle is the window that every neuron has into the world. Every neuron just sees a little chunk of the reality around it and picks up information based on the stimulus that is presented within that receptive field. So, if I start moving this now, you hear the pulses. Every one of them is one action potential from a neuron, and you can plot the activity as a function of the direction inside the receptive field. What you will see is that this particular neuron cares about motion to the left. It is a filter for the direction of motion. It encodes moving stimuli and signals to us, or the brain in general, what direction of motion is present in the environment.


Now you have to imagine that there are millions of these neurons all tuned to different directions, and as a concert across all these neurons they encode the visual information. This is the flow from the eye up into higher centres of the brain that I talked to you about at the beginning. It is a very powerful machinery and gives us high resolution in terms of our perceptual ability. Now, imagining that you have these neurons not only for direction of motion but for colour, for the orientation of edges, for the depth in a visual scene; you can imagine that the visual system breaks down an image into its local properties—each of them handled by specialised nerve cells. It is extremely well understood how that works, but there is a problem. As powerful as this machinery is, it delivers too much information. It is more than we can handle in a given situation. Something is lost, because our processing capacity is not sufficient to cover everything that our sensors pick up. That is a prediction I am making to you here; it is a claim, and it is claim that we are usually not aware of. If you look at me, you don’t have the impression that I am incomplete because some parts of me you can’t process. You don’t have the impression that there are gaps in your visual perception. So, that has to be studied a little more carefully than how we do this in this sort of introspection here. Let me show you a little movie from an American campus taken by colleagues, who were investigating the visual ability of this person in the middle. Here is the experimenter. They are both interacting. His very powerful visual system picks up a lot of information about the scene. Now the experimenter creates a short diversion and changes the visual scene massively, trying to see if the person on the right notices anything. Many of the subjects do not notice anything. Some of them get really concerned and run away. If you ask them what happened, they say, “Something strange happened, but I don’t know what it was.” So, as powerful as our visual systems are, we choose to analyse only a small portion of the incoming information. This might be surprising to you, and you might not believe this, but the organisers of the meeting asked me to try this with you as the subjects. Even though you might not be white haired and a lot younger than the guy here, you might not have noticed some of the things that have changed in the room while I was giving my presentation. Who noticed anything? I see very few hands. I turn to my colleagues here, who all volunteered for this experiment, and maybe you hold up what you changed. So, all of them turned around their badges into these bright blue badges just about a minute ago, while your attention was directed elsewhere, but your eyes were clearly able to see this.


This might be a little small for the people in the back, but for the people in the back, did you notice that we now have blue lights on the walls there? Maybe we can switch back to the correct illumination. We just switched the colour of the wall for you, and I hope, and I think it is right, most of you didn’t notice that. So, that shows you that we are very selective about our perception. In fact, the small amount of the information that we can process, we have to be very careful about, because if it is irrelevant information, then you are wasting your processing powers on something that you don’t really need for survival, and at the same time you might be missing something critical. This sophisticated cognitive process that underlies this is what we call “attention”. Now, how does this attentional system interact with the neurons I told you about, the neurons that pick up sort of the physical properties from the environment and simply encode the physical properties—nothing about attention or meaning or importance in that processing. So, lets go back to this little experiment. This is what you have seen so far. You have seen the data from this neuron, or this receptive field, and how this neuron is tuned and responds to a certain direction of motion. If I let the experiment go on a little bit and now instruct the animal, different from before, to instruct the animal to direct its attention on this stimulus, and I will code this in red when this happens. So, I now I am changing the instructions; the animal is now attending to the stimulus, and as you can see in the red plot here, the neuron becomes much more responsive. It doesn’t lose its selectivity; it still encodes the direction of motion. But the same physical stimulus has now become more powerful in terms of its signal that it passes on into the brain. That is presumably what happens if we use the attention to select information from our environment. It becomes much more prominent. So, attention creates a representation of our environment that is not a one-to-one copy, but rather emphasises the things that we care about and de-emphasises others. The basis for this are these neurons that I just introduced you to. So, to visualise this effect, if we think of this as the physically correct representation of a given scene where people are crossing these crosswalks, if you now imagine that the neurons prefer motion, it will de-emphasise non-moving parts. That is sort of a sensory representation. Now you are focusing your attention maybe on these two people here in the middle. What that will do to your internal representation, it will enhance them and push down everybody. These people are moving to the left; so you are not only attending to the location but also their direction, which will mean that the poor woman over here, who is at the wrong place and moving in the wrong direction, essentially becomes invisible to you, because your internal representation is missing that crucial piece of information from the environment.


So, to summarise, we have a highly sophisticated system for picking up information—very powerful. But to handle this amount of data, we need an equally sophisticated selection system, the attentional system. What we then have is a problem if there are deficits in the system. Many of you have heard of attentional deficit hyperactivity disorder, a problem with processing ability on an everyday scale. So, perception is an active process. It is not a passive copy of our environment. Therefore, if we go back to the picture at the beginning, while it is a very nice artistic rendition, it might not actually represent what we do. I much more like this picture, drawn by the same artist three years later, which I think captures the generative aspect of vision—seeing what will be there in the future. Thank you for your attention.

back