Scales of measurement for interrogating the human visual system with fMRI

Event details

Functional magnetic resonance imaging (fMRI) provides a unique non-invasive window into the human visual system, that can now operate across a broad spectrum of spatial scales. At the macro scale, conventional-resolution imaging combined with group-averaging strategies successfully mapped functionally selective modules throughout the visual cortex. While offering a high signal-to-noise ratio (SNR), these coarse-grained measurements primarily inform cognitive architectures—the classic "box-and-arrow" models—by prioritizing global functional specialization over individual micro-architecture.
At the same time other researchers used higher-resolution imaging and mapped responses onto the cortical surface of individuals to reveal retinotopic maps of the visual system. This scale allows precise characterization of visual field topographies on an individual basis, explicitly revealing the biological divisions of the cortical visual system. These divisions can then be probed for their function properties. The approach works particularly well in early visual areas where retinotopic representations are large and as a result easy to identify.
Today, ultra-high-field fMRI achieves resolutions of ~1mm that are capable of resolving mesoscopic structures, including cortical layers and functionally selective columns. Although mesoscopic imaging faces severe trade-offs in SNR and heightened sensitivity to motion and physiological noise, it fundamentally shifts the analytical paradigm to directly interrogate feedforward and feedback canonical micro-circuits and functional organization within visual maps.
This talk will review how trading off SNR for spatial specificity across these three measurement regimes reshapes our understanding of visual processing, tracing the trajectory of fMRI from testing functionally selective modules to probing underlying neurobiology.