Professor Dr Bradley C. Love | A common mechanism for spatial and concept learning

How do we learn to categorise novel items and what is the brain basis of these acts? For example, after a child is told an animal is a dog, how does that experience shape how she classifies future items? In this talk, I will discuss work using model-based fMRI analyses to understand how people learn categories from examples. Results indicate that the medial temporal lobe (MTL), including the hippocampus, plays an important role in both learning and recognition. Succesful cognitive models, which explain both behavioural and fMRI data, learn to selectively weight (i.e., attend) to stimulus aspects that are task relevant. This form of weighting, or top-down attention, can be viewed as a compression process. I will discuss how the medial prefrontal cortex (mPFC) and the hippocampus coordinate to build low-dimensional representations of learned concepts, as well as how the dimensionality of visual representations along the ventral stream is altered by the learning task. Finally, this general learning mechanism offers a straightforward account of spatial learning, including place and grid cell activity.

Poster