Semantic knowledge

We live in a world absolutely jam packed with stuff and a million things to know about. We learn all about it, and are capable of quite impressive acts of inference and generalization. It is important to characterize our semantic learning system in mechanistic terms, and to relate these mechanisms to the brain.

One line of inquiry we are pursuing is whether separate learning mechanisms are necessary for learning about things and learning about events. In the jargon of the field, the distinction is between "taxonomic" (think: context-free definitions) and "thematic" (think: context-bound associations) knowledge. Through careful behavioral experimentation and computational modeling, we are evaluating where there is a difference in how people utilize knowledge of different kinds.

Perceived similarity

Everyone has different experiences, which leads to learning different concepts, which then manifest as differences in perception and comprehension of the world. In short, although we often talk about words and objects having "a meaning", there are potentially important individual differences in concept representation that have very real psychological consequences.

In one line of inquiry we are exploring the relationship between individual differences in the representation of emotion concepts and self-report symptoms of depression. Emotion concepts are hard to define, but by having participants make similarity judgments, we can infer the meaning of each concept with respect to its dissimilarity from other concepts.

Neurocognitive representation

Semantic knowledge, like most cognitive and memory abilities, is supported on distributed networks of neural activity. This is challenging to study, because neuroimaging produces complex data which provide measurements at thousands of points in the brain simultaneously. A cognitive neuroscientist needs not only to look for which points correspond to a cognitive function, but for combinations of points. The number of possible combinations is practically infinite.

We are experimenting with exciting new analysis techniques that are capable of discovering sparse, distributed patterns of information in complex (but structured!) datasets like that produced by fMRI, EEG, and other neuroimaging techniques.

Emotion and Cognition

What is an emotion? Are you happy? Sad? Somber? Gleeful? Nostalgic? How we label a feeling, and the associations that label has, has consequences for our thoughts and behavior.

We are conducting experiments to understand how the semantic relationships among emotion words relate to personal experience in typical and atypical samples with respect to mental heath.

Learning and "Representational Development"

Every new experience is perceived and comprehended in light of our prior experiences. Life is all about progression. If you want to progress as quickly as possible, however, you need to array the right experiences in the right order.

Using computational models and lots of simulation work, we seek to gain insight into how representations develop over a sequence of experiences so that we are ultimately able to optimize those sequences. The end goal is translational: improving learning outcomes but optimizing the selection and sequencing of experiences.