Computational Memory Lab

A large amount is known about brain activity phenomena that are associated with memory tasks. More recently, the field of cognitive neuroscience of memory has been evolving from a descriptive approach (e.g., "neural correlates of...") to research that builds directly upon the rich knowledge we have accumulated over decades of behavioural memory research.

We specifically focus on brain-activity research that is intimately connected to behavioural experimental research and computational or mathematical cognitive models, with the goal of increasing what we can learn from memory-related brain-activity measures. At the same time, brain activity can provide complementary lines of evidence that we can use to inform and constrain our understanding of the cognitive processes that support memory, and even constrain our formal models. Ultimately, we want to be able to construct models of how brains remember that are true to both behaviour and neurophysiology

By asking new kinds of questions about the brain-behaviour relationship, and by developing novel ways to analyse our data, we are already beginning to learn new things about memory behaviour, memory-related brain activity and even the relationship between the two.

Highlights : Computational Modelling, Behavioural Experiments

We're trying to account for a diverse range of memory behaviour using simple, parsimonious models. Running behavioural experiments provides important clues as to what may differentiate and unite different memory paradigms, as well as providing valuable data for model testing/fitting. This research also sets the stage for brain-activity studies.

Highlights : EEG and fMRI grounded in behaviour and connected to models

The brain-activity experiments we do are grounded in behavioural experiments and mathematical models of behaviour. We employ two basc methods of recording brain activity: Electroencephalography (EEG) and functional magnetic resonance imaging (fMRI). Much of what we study starts with the Subsequent Memory Effect (Sanquist et al. 1980), in which we try to figure out what aspects of brain activity while people study materials is associated with later memory for those materials.

Highlights: Development/adaptation of novel data analysis methodology. Asking new kinds of questions

.

To make the connection between brain activity and memory behaviour (i.e., how does the brain produce behaviour?) more direct, we need to figure out new ways of asking this kind of question, which includes grounding our brain-activity experiments in behavioural research, and we also need to develop novel analysis methodology techniques to be able to ask these new kinds of questions better. We have been using the multivariate method, partial least-squares (PLS) thus far. We are also investigating machine learning methods such as support vector machines and relevance vector machines, which are very popular right now, but it is not yet clear how to apply them to cognitive brain data.