The computational and neural basis of visual metacognition
Description
Metacognition is the ability to judge the accuracy of our own decisions. Metacognitive ability is known to be imperfect but the nature of this imperfection is still not understood. I will present a new model of metacognitive imperfection that assumes two separate noise sources: sensory noise that affects both the perceptual decision and the confidence ratings, and metacognitive noise that only affects the confidence ratings. The model makes the counterintuitive prediction that higher sensory noise should lead to better metacognitive efficiency. I will present a series of studies that confirm this prediction and shed light on the nature of this metacognitive noise. Finally, I will link different components of our confidence model to the function of different areas of the prefrontal cortex. Together, these results build the foundation for a mechanistic understanding of visual metacognition.
Speaker Bio
Dobromir Rahnev is Assistant Professor of Psychology at the Georgia Institute of Technology. He earned a B.A. in Psychology from Harvard University and a Ph.D. in Psychology from Columbia University, after which he completed a postdoctoral training in cognitive neuroscience at UC Berkeley. Rahnev’s research focuses on perceptual decision making and confidence generation with an emphasis on creating computational models that explain human behavior and linking these models to the neural representations in human brains.