Table 2.

Comparative overview of the type of signals used for the decoding of spatial attention in the studies discussed in Table 1.

	Study	Signal	# Recorded signals	Coverage	Classification	Available trials
Non-invasive	van Gerven and Jensen (2009)	MEG/Alpha power (8–14 Hz) occipito-parietal channels	275 DC Squid axial gradiometers	Full head	Support Vector Machine w. linear kernel	128 trials per condition
	Treder et al. (2011)	EEG/Alpha power	64 channels	Full head	Logistic regression with L2 regularizer	40 trials per position
	Morioka et al. (2014)	EEG/Alpha and Beta Power guided by NIRS	EEG: 64 channels. NIRS: 49 channels.	Full head EEG. Parietal and occipital NIRS	Sparse logistic regression	88 trials per position
	Andersson et al. (2011)	fMRI/7T	Two ROIs: right vs. left and left vs. right	Full head		8 trials (13s) per condition
	Andersson et al. (2012)	fMRI		Full head		9 trials (13s) per position
Invasive in humans	Gunduz et al. (2012)	ECoG/Power from all frequency bands	86, 78, 64, 103, and 88 electrodes	Left hemisphere, 1 subject right hemisphere. No occipital lobe	Stepwise regression and Bayesian classification	40 trials per position (except for one subject: 80)
	Andersson et al. (2011)	ECoG/High gamma power	64 electrodes	Left parietal-occipital cortex		20 left/20 right/39 center
Invasive in non-human primates	Rotermund et al. (2013)	ECoG/Power, phase coherence and difference	36 and 37 electrodes	Extrastriate area V4 and portions of V1/V2 along lunate sulcus	Support Vector Machine w. Gaussian kernel	~several hundred trials
	Astrand et al. (2014a)	SUA/average spiking rate	131 neurons	Single hemisphere, Right or Left Frontal Eye Fields	Linear Regression with regularization	60 trials per condition
	Astrand et al. (2014b)	MUA/average spiking rate	48 channels	Both hemispheres, Frontal Eye Fields	Linear Regression with regularization	60 trials per position
	Armstrong et al. (2009)	SUA/average spiking rate	40 neurons	Right Frontal Eye Field	Support Vector Machine w. linear kernel	$>$10 trials per condition (leave-one out procedure)
	Zhang et al. (2011)	SUA/average spiking rate	187 neurons	Anterior Inferior Temporal cortex, single hemisphere	Correlation Coefficient Classifier	12 trials per stimulus (leave-one-out procedure)

This includes the type of recording method, the number of recorded signals, brain coverage, the type of classifier used for the decoding, the available number of trials per condition.