Abstract
Vision is an important function that guides behaviors. However, our understanding of the visual system in the brain remains incomplete. In this dissertation, I combine behavioral, electrophysiological, and computational approaches in two model organisms — tree shrews and mice — to study the vision from two perspectives: visually guided behavior and its implicated neural mechanism, and motion processing on single neuron level and its implicated computational principle. For the former perspective, I studied visual decision-making in tree shrews; for the latter perspective, I studied motion representation in the superior colliculus (SC) of both tree shrews and mice.
In Chapter 1, I review the historical view of vision and visually guided behavior, summarize key anatomical and physiological properties of the SC as a subcortical visual structure, and highlight accumulating findings that link SC activities to cognitive processes. I conclude by proposing the tree shrew as an intermediate model for comparative vision research, lying between rodent and primate visual systems.
Chapter 2 presents our work on establishing and characterizing visual decision-making behaviors in freely moving tree shrews. Using a two alternative forced choice contrast-discrimination task with manipulations of trial-delay schemes, I demonstrate that tree shrews rely on a stimulus-independent process to guide their choice behaviors. The comparison between two forms of racing diffusion models fit to the choice and response time data further shows a potential mechanism for these task-dependent non-sensory decision signals can arise from a time accumulation process.
In Chapter 3, I shift focus to the mouse SC and ask how individual neurons represent "plaids", a type of complex motion patterns. Using asymmetric plaid stimuli, I show that mouse SC neurons do not implement the classic intersection-of-constraints (IOC) rule, which is followed by primate perception and cortex neurons to integrate the motion, but instead compute a probabilistically constrained vector sum (VS) of component directions. By examining optokinetic reflex (OKR) behaviors to plaids, I further demonstrate that this probabilistically constrained VS computation directly drives reflexive eye movements. The findings raise intriguing questions about subcortical visual motion processing and its functional significance.
Chapter 4 extends the motion‐processing investigation to the tree shrew SC. Through in-vivo electrophysiological recordings and OKR behavioral measurements, I establish that tree shrew SC neurons are tuned to spatial frequency, temporal frequency, orientation, speed, and motion direction in a manner comparable to mouse SC, although the exact tuning ranges exhibit a priority on processing rapid motion stimuli. In response to symmetric and asymmetric plaids, tree shrew SC neurons also implement a VS‐like representation of pattern direction, mirroring the mouse results. Conversely, tree shrew SC demonstrates a lower level of pattern selectivity compared to mice, and the VS computation is not limited by the probabilistic constraints observed in mice. The findings show both the conserved functionality and the specialized adaptation of the tree shrew SC.
Finally, Chapter 5 discusses these cross-species findings and explores their broader implications in light of the literature on other species. I first discuss the tree shrew’s exceptionally fast behavioral responses and the implication of the underlying neural mechanisms and drive force. Next, I call attention to the importance of subcortical visual functions, with an emphasis on the SC. Comparative perspective highlights how subcortical processing supports rapid behaviors, while cortical circuits can augment precision under the requirement of behavioral context and species needs. Finally, I briefly discuss the SC function in a broader range of behaviors including abstract cognitive processes and urge for a integrative view on the SC.
In summary, this dissertation demonstrates a collection of my investigations of vision with a comparative perspective. These cross‐species insights illuminate how evolution potentially shapes the trade-offs in visual processing, and they underscore the SC’s pivotal role in bridging visual inputs with behavior.
