Tracing Neurons or Mimicking Neurons: A Comparison of Two Techniques
N.R. Buijs · Utrecht University Repository (Utrecht University) · 2020
There is a discrepancy between the field of neuroscience and of computer science. Artificial intelligence is a broad, interdisciplinary field, which involves both of these areas of research. Convolutional neural networks (CNNs), a type of artificial intelligence, have been inspired by biological neurons. In neuroscience, population receptive field mapping (pRF mapping) is used to estimate the receptive fields of voxels (voluminous pixels, a unit to index small groups of neurons close together), to produce visual/sensory/auditory field maps of the brain. In this thesis there is a main focus on the visual pathway, to be able to discuss the processing of visual input for both techniques: pRF mapping and CNNs. The question answered in this thesis is “How can the two techniques pRF mapping and CNNs be advanced by incorporating best practices or concepts from the other technique?” The concept of a receptive field is shared in both areas of research and new insights on them - from either side - could advance both disciplines.