Dissociation of attentional state and behavioral outcome using local field potentials.

Behavioral outcome (i.e., whether a target is detected or missed) depends on attentional state and potentially other factors related to decision-making, which could in turn modulate the power and phase of neuronal oscillations. Here we investigated whether attentional state (i.e., whether attention is inside or outside the receptive fields of neurons) and behavioral outcome are distinguishable using the power and phase of local field potential (LFP) recorded from electrode arrays in area V4 of two male rhesus monkeys performing an attentional task under different cuing conditions. Since attention also strongly modulates pairwise measures such as spike count correlation and phase consistency which are typically measured across trials, we developed novel methods to obtain single-trial estimates of these measures. Surprisingly, while attentional location was best discriminated using gamma and high-gamma power, behavioral outcome was best discriminated by alpha power and steady-state visually evoked potential. Power outperformed absolute phase, although single-trial gamma phase consistency provided good attentional discriminability. Our results provide a clear dissociation between the neural mechanisms that regulate attentional focus and those that govern behavioral outcome.

Decoding of attentional state using local field potentials.

We review recent efforts to decode visual spatial attention from different types of brain signals, such as spikes and local field potentials (LFPs). Combining signals from more electrodes improves decoding, but the pattern of improvement varies considerably depending on the signal as well as the task (for example, decoding of sensory stimulus/motor intention versus location of attention). We argue that this pattern of results conveys important information not only about the usefulness of a particular brain signal for decoding attention, but also about the spatial scale over which attention operates in the brain. The spatial scale, in turn, likely depends on the extent of underlying mechanisms such as normalization, gain control via excitation-inhibition interactions, and neuromodulatory regulation of attention.

Decoding of Attentional State Using High-Frequency Local Field Potential Is As Accurate As Using Spikes.

Local field potentials (LFPs) in visual cortex are reliably modulated when the subject's focus of attention is cued into versus out of the receptive field of the recorded sites, similar to modulation of spikes. However, human psychophysics studies have used an additional attention condition, neutral cueing, for decades. The effect of neutral cueing on spikes was examined recently and found to be intermediate between cued and uncued conditions. However, whether LFPs are also precise enough to represent graded states of attention is unknown. We found in rhesus monkeys that LFPs during neutral cueing were also intermediate between cued and uncued conditions. For a single electrode, attention was more discriminable using high frequency (>30 Hz) LFP power than spikes, which is expected because LFP represents a population signal and therefore is expected to be less noisy than spikes. However, previous studies have shown that when multiple electrodes are used, spikes can outperform LFPs. Surprisingly, in our study, spikes did not outperform LFPs when discriminability was computed using multiple electrodes, even though the LFP activity was highly correlated across electrodes compared with spikes. These results constrain the spatial scale over which attention operates and highlight the usefulness of LFPs in studying attention.