POSTER DETAILS RELATED POSTERS

POSTER DETAILS

Spatial and feature-selective attention interact multiplicatively in multiple-demand network

Nadene Dermody, Romy Lorenz, Alexandra Woolgar

Date / Location: Monday, 11 July 2022 / S04-075
Add this poster event to your 🗓
Aim: Distinct neural effects are associated with attention deployed to a location (spatial attention) or to a particular object feature (feature attention), but it is less clear how these types of attention interact to affect information processing. To examine this, we compared the effects of spatial and feature attention on stimulus representations within a frontoparietal “multiple-demand” (MD) network posited to play a critical role in attentional control. Method: Participants (N=30, 16 female, 14 male, mean age=27.2) underwent fMRI while they covertly attended to one of two objects, presented left and right of a fixation cross (spatial attention manipulation), and reported the attended object's colour ("red" or "green") or shape ("X-shaped" or "flat") (feature attention manipulation) via button press (Fig1, Panel A). We used multivariate pattern analysis to measure coding of attended and unattended stimulus information. Results: We found a significant multiplicative interaction between spatial and feature attention in MD (F(1,30)=69.924, p<.001) and visual cortices (F(1,30)=17.304, p<.001). Moreover, stimulus decoding was only above chance for the attended feature of the attended object (BF10>100)(Fig1, Panel B). We additionally found these task-relevant representations were organised along two major dimensions, reflecting physical stimulus properties and task difficulty. fig1.png Conclusion: Our results suggest spatial and feature attention interact multiplicatively, selectively enhancing coding of the attended feature of the attended object. Rather than boosting processing of whole objects or relevant features across space, selective attention in the MD system – at least in this difficult perceptual task – appears to reflect all-or-nothing tuning to behaviourally relevant information.