Data Science and Neuroeconomics
Over the past 2 decades, the field of neuroeconomics has shown to be a better predictor of outcomes than epidemiological surveys. Much of this stems from neuroscience’s ability to understand the unconscious processing behind behaviors. I have studied a number of unconscious systems that relate to behaviors, such as statistical learning (how we automatically pick up on patterns in the environment) to familiarity.
To understand learning mechanisms of the brain, I use a number of Psychological, Biological, and Imaging Techniques. During my dissertation I explored how learning a working-memory task changes the prefrontal cortex and discovered that memory in this area is also related to implicit (unconscious) buffering of information (Click for more information: 1, 2, 3, 4)
To explore these implicit learning mechanisms, I explored brain areas that project to the prefrontal cortex, and designed Psychological experiments to explore how the brain automatically picks up on environmental patterns (Statistical Learning and Familiarity). I discovered that brains track the probability past events to make predictions of the future. Interestingly, instead of the brain increasing in activity, it actually reduces it’s activity - as though the predictions are leading to a ‘Zen’ state. In other words, if the brain knows what’s going to happen, then it can reduce metabolic resources. (Click for more information: 1, 2, 3)
To probe this learning circuitry, I decided to explore the neural limits of the most basic statistical learning I could think of - familiarity. I discovered that this reduction in activity (Prediction Suppression) primed the brain to then respond more robustly to following images. Interestingly, different species have different processing frequencies: human brains could process familiar images at 180 ms (5.6 Hz), while monkeys were faster at 120 ms (8.3Hz) (Click for more information: 1)
To explore these neural mechanism further, I joined the Visual Memory Laboratory at the University of Pennsylvania. We developed a technique that allowed us to measure familiarity even after SINGLE exposures using Neural Population Decoders. We’ve been able to track the memorability of images, categorical boundaries, and even to map forgetting at the neural level. (Click for more information: 1, 2)