Pattern classification of neural signals
A number of projects in the lab involve the examination of the patterns of neural activity elicited while people search through their memories for recently experienced events. We are examining human neural activity using both functional magnetic resonance imaging (fMRI) and scalp electroencephalography (scalp EEG). Using fMRI, we have found that the topographic pattern of blood flow across many brain sites can be used to determine the category of an item while it is being viewed, as well as just before the person recalls the item during a memory test. Using scalp EEG, we have found that the topographic pattern of oscillatory neural activity can be used to decode this same information, but on a much quicker time-scale.
Computational modeling of the human memory system
The Context Maintenance and Retrieval model (CMR) is a computational model of the human memory system. It is one of a class of retrieved context models that explains why, when people search through their memories, the memories come to them in a particular order. The basic dynamics of the CMR model come from from the Temporal Context Model of human memory (TCM), which, among other things, explains why people tend to remember things that happened nearby in time one after another. CMR builds upon this idea and explains more generally, why all kinds of things that are similar to one another tend to be remembered one after another.
Behavioral investigations of memory dynamics
We ask people to study a series of items (e.g., a list of words, or a series of pictures) and then after they've had a chance to study all of the items, we ask them to remember those items, aloud, in any order. By looking carefully at the likelihood that people will remember particular items, and the order in which people remember the items, we can learn a lot about how the memory system works.