 You may not be aware of this, but our visual world lies somewhere between a visual illusion and reality. We actually extract nuggets of information from our environment and fill in the rest of the environment by using predictions and expectations of what we expect to see. Unconsciously, we try to predict what we will next encounter in the environment in order to efficiently process incoming information. But sometimes predictions can go wrong. For example, one time I was approaching a business with pink and purple lights. I expected this business to be a bar or a dance club. To my surprise, it happened to be a T-Mobile store. But most of the time, we are successful in our predictions, and therefore, our visual experience comes from not just a veritable analysis of light falling on our eyes, but rather an intermingle of our lifetime of experience and our perceptual input that gives rise to our visual experience. So what does it mean to see? What does it mean to understand our visual world? It means going well beyond just looking at a scene. For example, because this is a beach scene, we have certain expectations and stereotypical associations of objects that we expect to see and certain actions and behaviors. Therefore, to see is not trivial. Computers are still bad at this problem. Although there has been amazing progress over the last few years, computers would still have a hard time telling you why a 40-foot rubber duck in a river is unusual. I want to understand how we are so good at this problem. And by we, I mean our brains, of course, who do all the hard work for us. And the place to look for answers is a network of brain regions that process the visual environment. These areas process stimuli such as scenes like a beach, more than other stimuli such as faces and nonsense images. And moreover, these areas don't process scenes all equally. The differences in brain response between scene stimuli can reveal meaningful clues about how the brain represents information about a scene. And it can help reveal the code the brain uses in order to understand our visual world. For example, a key question is, how does the brain break apart a scene? What are those visual nuggets that help us predict what else is going on in the scene? Just like a sentence can be broken down into words, how can a scene be broken down into meaningful units? If we could answer this question, we might have a better idea of how the brain represents scenes. But unfortunately, our brains don't tell us what they're doing. And we have terrible self-reflection about how to explain our brain processing. So therefore, we can't just sit in the chair and decide what are the most important aspects of a scene. How does the brain understand what are the critical parts of a scene? By having a lifetime of experience analyzing scenes. As I like to say, the brain is the biggest big data analyzer out there. So how can we replicate this analysis power? This is where computer vision and machine learning tools come in. They can analyze big data very well and can approximate the type of processing that the brain may be doing in understanding scenes. By funneling huge amounts of data, computers can define important aspects of scenes and therefore reveal the building blocks the brain may be extracting from the environment in order to understand the full visual experience. Therefore, we asked whether computers could help us understand brain function better than we could understand brain function with our own self-awareness. So we compared computers analysis to scenes, to brains analysis to scenes, to our own self-reflective analysis of scenes. And it turned out that computers did really well in explaining our brains analysis of scenes, whereas self-reflection did not. Finding the similarity between brains and computers is a huge step forward in understanding how the brain represents scenes. Because not only does it add computers analyze scenes in a similar way, but it can tell us how it does it. It can tell us the building blocks of this scene that the brain may be using to extract from the environment to give us our full visual experience. And the more we can reveal about the information that we process to give our full visual experience, the more we can look at how the human experience can shape our perceptual world. This, in turn, can guide engineering of computer systems to help us become better at integrating technology in our everyday life. Therefore, there is a bidirectional, intimate relationship between computational technology, cognitive neuroscience, and everyday life. The more that we can help one another, the more that we can use technology efficiently and seamlessly in our everyday life. So the question that I would like to leave you with is, what will you do with computers that can see and understand? Thank you.