People can recognize objects and concepts at different semantic levels, such as knowing that a pet is an animal, a dog, a German shepherd, and “Fidoâ€. These levels can be remembered or forgotten differently. When people first learn a concept, the brain encodes the relevant information at each of these levels into memory. Each time a memory is retrieved, its underlying neural patterns are reactivated, including its multiple semantic levels. Yet, the neural representation of these levels of memories and reactivations are poorly understood. This project will examine the human brain mechanisms that support memory for concepts, and the ways in which learning and retrieval affect their neural representations. Functional magnetic resonance imaging (fMRI) experiments and advanced analytical methods will investigate the brain activity of healthy adults as they encode and retrieve new concepts. The research will examine how processing different semantic levels during encoding and retrieval influences a concept's neural representations and a person’s memory. The findings may apply to learning, including for formal education and for behavioral interventions. The project will include workshops for college freshmen on cognitive neuroscience of learning and retrieval applied to study skills.
The project has two objectives: 1) to test how the level of semantic granularity, processed during encoding and retrieval, impacts neural representation and memory; 2) to test how training protocols strengthen different levels of neural representations for concepts. The first set of experiments will examine how the degree of semantic granularity, invoked during learning and retrieval, determines the specificity of neural reactivation and final memory. They will also test whether consistency in the levels of semantic granularity in encoding and retrieval influences neural patterns and memory. The second set of experiments will examine how testing, re-exposure, and sleep play a role in this influence. Throughout, studies will employ fMRI to record patterns of brain activity as participants encode new pairings between words and images. The encoding and retrieval procedures will be manipulated to emphasize the item, category, or theme of presented concepts. Data analysis methods will include representational similarity analysis and machine learning. These will test alternative neural models for how semantic granularity can be represented in neural activity patterns within, and across, brain regions. These experiments will increase understanding of how neural representations of new concepts are molded by the manner in which they are encoded and reactivated.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
