Task-specific, dimension-based attentional shaping of motion processing in area MT

Abstract

Non-spatial selective attention is based on the notion that specific features or objects in the visual environment are effectively prioritized in cortical visual processing. Feature-based attention (FBA) in particular, is a well-studied process that dynamically and selectively enhances neurons preferentially processing the attended feature attribute (e.g. leftward motion). In everyday life, however, behavior may require high sensitivity for an entire feature dimension (e.g. motion). Yet, evidence for feature dimension-specific attentional modulation on a cellular level is lacking. We here investigate neuronal activity in macaque motion-selective medio-temporal area (MT) in an experimental setting requiring the monkeys to detect either a motion change or a color change. We hypothesized that neural activity in MT is enhanced if the task requires perceptual . CC-BY-NC-ND 4.0 International license not peer-reviewed) is the author/funder. It is made available under a The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/058693 doi: bioRxiv preprint first posted online Jun. 13, 2016; -2-sensitivity to motion. Despite identical visual stimulation, we found that mean firing rates were higher in the motion task, and response variability and latency were lower as compared to the color task. This task-specific response modulation in the processing of visual motion was independent from the relation between attended and stimulating motion direction. It emerged already in the absence of visual input, and consisted of a spatially global and tuning-independent shift of the MT baseline activity. The results provide single cell support for the hypothesis of a feature dimension-specific top-down signal emphasizing the processing of an entire feature class.

Assessing the robustness of feature-based selection in visual working memory.

“…A prominent property of feature-based attention is that it can be deployed broadly over space (Carrasco, 2011; Gledhill et al, 2015; Maunsell & Treue, 2006; McAdams & Maunsell, 2000; Saenz et al, 2002; Schledde et al, 2016; Treue & Trujillo, 1999). Feature retro-cues were previously shown to afford selecting more than one item in VWM simultaneously in a change detection task (Heuer & Schubö, 2016).…”

“…On the neurophysiological level, spatial attention has been shown to operate via a neural gain mechanism—that is, by amplifying the neural signal coding information perceived at the attended location—whereas feature-based attention operates by neural tuning, which involves amplification of the relevant signal coupled with suppression of other signals in the visual field (Baldassi & Verghese, 2005; Ling et al, 2009). In crucial opposition to spatial attention, in which the benefit arises when the selection focus is narrowed, attention to specific features operates in spatially extended manner, that is, it coselects goal-relevant features across the whole visual field (Carrasco, 2011; Gledhill et al, 2015; Maunsell & Treue, 2006; McAdams & Maunsell, 2000; Saenz et al, 2002; Schledde et al, 2016; Treue & Trujillo, 1999).…”