Alpha Power and 1/f Slope Provide Independent Decoding of Visual Spatial Attention
Thomas Donoghue1, Bradley Voytek1; 1University of California, San Diego (UCSD)
The focus of visual spatial attention has previously been demonstrated to be predictable from the topography of posterior alpha oscillations (~8-12 Hz). Here we extend these findings by exploring how individual variation in alpha oscillations, as well as background 1/f activity relate to attentional focus. In the first experiment (n=15), participants completed a peripheral visual detection task with concurrent electroencephalography, in which pre-trial cues indicated which of two possible locations participants should attend (left or right). Examining the time period between cue onset and stimulus presentation, we replicate that the topography of posterior alpha oscillations is predictive of attentional focus, using a support vector machine (SVM) classifier. We demonstrate that predictions can be improved by taking the oscillation-specific power at subject-defined peak frequencies, and that 1/f background activity, as measured by the slope of the power spectral density, is also predictive of attentional focus, with the best decoding using a combination of alpha and slope features. In a second experiment (n=14), a variant of the detection task was used with four peripheral locations that could be cued with varying degrees of spatial specificity. Here we replicate that alpha and slope predict the spatial focus of covert attention, across multiple locations, and find that these physiological parameters additionally predict the spatial extent of attentional allocation. Collectively, these experiments expand upon prior work on how alpha oscillations relate to attention and demonstrate how 1/f slope adds additional information to the decoding of both the locus and extent of covert visual attention.
Topic Area: ATTENTION: Spatial