Neural Mechanisms of Presaccadic Attention and Foveal Prediction

• Main Author: Coop, Shanna Hong
• Format: Dissertation
• Language: English
• Published: ProQuest Dissertations & Theses 01.01.2022
• Subjects: Cognitive psychology; Neurosciences
• ISBN: 9798790654510

Human vision relies on rapid eye movements (saccades) to bring peripheral visual targets to the fovea for high resolution inspection. An open question is how peripheral and foveal processing are integrated through a saccade to maintain the continuity of attentional selection for the target of the saccade as it jumps from the periphery to the fovea. It is known that attention leads eye movements producing perceptual enhancements at the saccade target immediately before saccades, called presaccadic attention. Tracking stimuli across saccades is also thought to involve a predictive visual remapping that anticipates the sensory effect of saccades, called visual remapping, and for saccade targets brought to the fovea, foveal remapping. Recent research has shown that processing in peripheral and foveal vision is not independent, but rather optimally integrated to create a more stable percept of objects (Wolf & Schütz, 2015; Ganmor et al., 2015). We hypothesize that presaccadic attention selects target features before the saccade in order to prime post-saccadic foveal processing to select the same features and thus preserve continuity of attention for the same target across the saccade as it is remapped to the fovea. Under this hypothesis, feature attention should be an automatic and obligatory component of saccade planning and it should influence post-saccadic foveal processing. The series of studies in this dissertation investigate the underlying mechanisms for presaccadic attention and foveal remapping. We explore each of these processes in the marmoset monkey using a combination of controlled saccadic behaviors in foraging tasks, eye-tracking, and neurophysiology.In Chapter 2, we first examined whether marmosets exhibit presaccadic attention in the middle temporal area (MT) similar to other primates using a simple saccade foraging task. We establish that presaccadic attention in the marmoset alters neural activity in the same manner as in macaques with increases in firing rates and increases in stimulus sensitivity. Recent human psychophysical studies further suggest that presaccadic attention may automatically engage feature selection for the target (Li et al., 2016; Ohl et al., 2017). At the single unit level, feature gain for the target would predict a narrowing in feature tuning. Thus we also examined how neural tuning is modulated in presaccadic attention to test for such narrowing. We find that changes in tuning across the pooled population of neurons show additive and gain increases in firing rate, consistent with early studies of covert attention, but subsets of neurons do exhibit significant narrowing in tuning and could potentially support the feature enhancements seen at the psychophysical level. This opens a question as to how we might determine which neurons in the population influence the perceptual read-out, either by their position in the cortical circuit or by their correlation to behavior.In Chapter 3, we took advantage of the marmoset’s smooth cortical surface to measure neural effects of presaccadic attention as a function of laminar position and type of spike waveform in MT. We first identified the location of input layer with ±100-micron accuracy using a current source density (CSD) method (Mitzdorf, 1985). We then classified isolated single units into narrow and broad spiking categories based on their waveform duration similar to previous studies in macaque V4 (Mitchell et., 2007; Nandy et al., 2017). It is thought that narrow spiking neurons represent local interneurons in cortex, while broad spiking neurons are a diverse set predominantly containing larger pyramidal neurons that project to other areas, and thus, would be more likely to carry signals related to perception. Further, it is known that neurons in superficial layers project information forward in cortex, and thus would also be more likely to convey signals necessary for perception or decision making. The results from this work indicate that the superficial layers of MT show a significant gain increase and sensitivity compared to input and deep layers, while there is a non-selective additive increase across all layers. We find the increase in sensitivity in superficial layers is specific to the broad-spiking neurons and not the narrow-spiking. These results suggest that presaccadic attention uses the MT laminar circuit in distinctive ways and that subsets of neurons exhibiting enhancements in tuning may be preferentially positioned to influence perception, while other cell classes play more regulatory roles in the circuit.Last in Chapter 4, we investigated whether the foveal representation in area MT reflects a prediction of the presaccadic target feature that could facilitate attention continuity across the saccade. Recent studies in both fMRI and EEG suggest that early visual cortex in humans is involved in foveal remapping of the saccade target such that predictive feature information is present in foveal cortical representations immediately before the saccade (Knapen et al 2016; Edwards et al., 2018). Our results support these findings at the single unit level by providing evidence that foveal MT neurons receive feature predictions about a saccade target before it enters their receptive fields. Future studies should examine circuit level implementation of feature prediction in the fovea and its role in active vision.