 Have you ever lost your keys? Have you noticed this doesn't happen if you systematically place them in the same location? Do you know why? How does our brain optimize memory function and what can we learn from it to treat memory loss, to improve education, or even to build artificial brains? Let me start by introducing my friend Mark. Mark is a CEO of a high tech company. He's married to a beautiful career woman and he has two small children. Mark is responsible for driving the keys to school every day and he's usually on a hurry. But on that particular morning, everyone overslept. Mark was so anxious to get everything done on time and on top of it all, he couldn't find his car keys. He usually placed them in the living room table so that he didn't have to search for them in the morning. This coincident presentation of the two objects resulted in a strong association between keys and that particular location. Can you think why does the brain form such associations? We live in an era of information overload. We are continuously bombarded with information about our health, our work, our environment, yet our neural resources do not allow us to store everything. I propose that forming associations is an optimal strategy that maximizes both memory capacity and the use of neural resources. Let's take a look at Mark's brain. Deep inside lies this primitive structure that looks like a seahorse called the hippocampus. The hippocampus is critical for learning and memory formation. This is where memory associations are believed to reside. If you take a closer look, you'll find thousands of neurons communicate through electrical signals. Tiny, mushroom-like structures called synapses serve as communication portals. When activated, they lead to the generation of electrical signals that form the essence of everything our brain does. Now deep inside this complex universe, associative memories are stored in specific cell populations. In fact, we now have the technology to manipulate such memories. If you delete these neurons, the memory will disappear. If you activate them again, the memory will come back. However, to do finer manipulations, one needs to know the molecular mechanism behind such memories. And this is where computational models can help. Our simplified model neurons learn to associate two objects by modifying their synaptic connections, much like real brains do. If you present one object, the other object will be recalled, thus creating a virtual memory. In fact, our models predict that associative memories are formed by strengthening their connections in nearby locations within the same cells. This is energetically efficient because the same memories share molecular resources to be established. So how is this beneficial at the organism level? Let's go back to Mark and his efforts to retrieve his keys. What do you think he did? Well, he tried to identify the last moment that he has seen his keys and the events that followed, hoping that this would give him the answer. This mental exercise back in time started from the moment that Mark drove his car into his garage. He remember locking the door and climbing up the stairs to his house. Upon entering, he heard the phone ringing. He rushed into the study room to pick it up. And to do so, he placed his car keys next to the phone device. Now, this mental exercise recalled a memory of him driving into his garage. If our models are right, this memory also activated a set of neurons that contain the second memory, the phone ringing memory. In turn, the phone ringing memory reactivated neurons that contain the memory of him placing the keys next to the phone device. And that's how he managed to find his keys and drive the keys to school on time. Now, the goal of this presentation was to stimulate your curiosity about why and how our brain forms memory associations. If this is an optimal strategy that evolved in order to maximize performance while operating under limited resources, perhaps there's something to gain by understanding it. And this is where models can help. They serve as a bridge between the real brain and the technological innovations that may emerge by simulating such neuronal functions. Our models can be very detailed, starting from the molecular level of single neurons to the micro circuit and the entire brain regions. What can they be used for? They can predict new treatments for memory loss by pointing to key molecules. They can inspire new educational protocols that capitalize on our innate ability to bind information. Or even inspire new technological applications that lead to smart artificial brains. If the brain is doing it, it might be the best way of doing it. So here is my question for you. What's in it for you? What does the social and technological world have to gain by understanding the brain's memory strategies and adopting our tools to achieve technological advances? Thank you.