guest ::
| Home > Publications database > A neurocomputational approach to feature- and space-based attention systems in the human brain |
| Home > Publications database > A neurocomputational approach to feature- and space-based attention systems in the human brain |
| Abstract | FZJ-2016-04801 |
; ; ; ;
2016
Abstract: Introduction:Prior information about the location or features of an upcoming stimulus facilitates its detection and speeds up response times (RTs). Conversely, violations of expectancies induce RT costs. These effects can be investigated with cueing paradigms in which a cue predicts the location or a feature of an imminent target with above chance probability [1]. Even when this probability is unknown or changes over time, subjects infer the probability that the cue will be valid in a given trial on the basis of recent observations according to Bayesian principles [2]. At the neural level, orienting attention after location or feature cues engages a common dorsal frontoparietal network [3]. Reorienting of attention to invalidly cued targets activates a ventral frontoparietal system in the right hemisphere in the spatial domain, and activity in some of these regions is modulated by inferred trial-wise probabilistic beliefs about cue validity [4]. These effects have not been investigated for feature-based attention.Methods:The present study combined computational modeling of RTs with fMRI to compare the neural mechanisms of attentional orienting, reorienting, and probabilistic inference in spatial and feature-based attention systems of the human brain. In two runs, 24 volunteers performed two versions of a probabilistic cueing task with spatial or feature (color) cues in a 3T scanner. The percentage of cue validity (%CV) changed unpredictably between 50, 70, and 90% in each version. Parameters of a hierarchical Bayesian model [5] were estimated on the basis of individual RTs to derive trial-wise estimates of probability-dependent attention, which then entered the analysis of the fMRI data as parametric regressors. Whole-brain analyses were used to identify orienting and reorienting networks in both attention systems. Modulations by probability-dependent attention were investigated by extracting beta estimates for the parametric regressor in these regions and comparing them between the different conditions.Results:Higher levels of experimentally manipulated %CV significantly increased RT costs after invalid cueing. The subject-specific parameters of the Bayesian model - quantifying trial-wise updating of the probability that the cue will be valid - were significantly correlated between the spatial and the feature-based task. In both task versions orienting of attention activated a similar bilateral network comprising the superior parietal lobe, frontal eye fields, and putamen. There were no significant modulations of activity in these areas by probability-dependent attention. Reorienting of attention engaged bilateral superior and inferior frontal gyri (SFG/IFG), the precuneus, left supramarginal gyrus and left inferior parietal lobe (IPL) in the spatial attention task, while feature-based reorienting activated the left IPL. Parametric modulations of reorienting-related activity by trial-wise probability-dependent attention were observed in the left IPL for both attention systems, while modulations in the SFG and precuneus were only significant for spatial attention.Conclusions:These findings provide novel insights into the generality and specificity of the neural and computational mechanisms underlying attentional modulation in different subsystems. At the behavioral level, the updating of the trial-wise probability of valid cueing was correlated between the spatial and the feature-based task, suggesting a link between the mechanisms for probabilistic inference in both systems. In line with this notion, activity in the left IPL was modulated by probability-dependent spatial and feature-based attention. However, in other brain areas reorienting-related activity and probability-dependent effects were selective for the spatial attentional system. These results may help to explain findings from patients with right-hemispheric brain lesions with selective deficits in learning spatial statistical regularities [6].Imaging Methods:BOLD fMRI 2Modeling and Analysis Methods:Bayesian ModelingPerception and Attention:Attention: Visual 1Perception and Attention Other
|
The record appears in these collections: |