Spatial attention modulates steady state VEPs in retinotopic human visual cortex

Introduction Given the complexity of our visual environment, the ability to selectively attend to certain locations, while ignoring others, is crucial for reducing the amount of visual information to manageable levels and for optimizing performance. It is commonly accepted that selective, top-down, attention influences responses in higher level visual areas. But, while fMRI literature on humans show robust attentional modulation in lower visual areas, particularly V1, this is practically absent in the primate electrophysiological literature. One explanation for this could be the difference in methods. Since fMRI measures a blood oxygen-level dependent (BOLD) signal, and thus metabolic activity, the imaging results could be due to non-spike-related activity, for example increased attentionally-driven synaptic activity from feedback connections. Buracas and Boynton [1] showed that attentional effects on the BOLD signal can be modeled by a purely additive mechanism. They hypothesized that their data could be explained by non-spike-related activity or by a DC increase in the firing rate of whole populations of neurons. Either of these mechanisms might be too small to be detected in single cell electrophysiological measurements, but could, nevertheless, generate significant changes in metabolic activity. In order to examine the nature of the attentional modulation on neural responses in human visual cortex, we measured source-imaged visually evoked potentials (VEPs) using a combination of highdensity EEG and fMRI retinotopic mapping. We found that dynamic neural responses were modulated by attention in all the visual areas we examined, including V1. Since EEG measures electrical currents, this modulation must be due to neural (axonal/spike) activity, and not just a metabolic change. Further, since the modulation is an amplitude modulation of the temporal frequency corresponding to the attended visual stimulus, it suggests that attention is acting to modulate this signal in a multiplicative manner in addition to any DC offsets that it might also be generating.


Introduction
Given the complexity of our visual environment, the ability to selectively attend to certain locations, while ignoring others, is crucial for reducing the amount of visual information to manageable levels and for optimizing performance.It is commonly accepted that selective, top-down, attention influences responses in higher level visual areas.But, while fMRI literature on humans show robust attentional modulation in lower visual areas, particularly V1, this is practically absent in the primate electrophysiological literature.One explanation for this could be the difference in methods.Since fMRI measures a blood oxygen-level dependent (BOLD) signal, and thus metabolic activity, the imaging results could be due to non-spike-related activity, for example increased attentionally-driven synaptic activity from feedback connections.Buracas and Boynton [1] showed that attentional effects on the BOLD signal can be modeled by a purely additive mechanism.They hypothesized that their data could be explained by non-spike-related activity or by a DC increase in the firing rate of whole populations of neurons.Either of these mechanisms might be too small to be detected in single cell electrophysiological measurements, but could, nevertheless, generate significant changes in metabolic activity.
In order to examine the nature of the attentional modulation on neural responses in human visual cortex, we measured source-imaged visually evoked potentials (VEPs) using a combination of highdensity EEG and fMRI retinotopic mapping.We found that dynamic neural responses were modulated by attention in all the visual areas we examined, including V1.Since EEG measures electrical currents, this modulation must be due to neural (axonal/spike) activity, and not just a metabolic change.Further, since the modulation is an amplitude modulation of the temporal frequency corresponding to the attended visual stimulus, it suggests that attention is acting to modulate this signal in a multiplicative manner in addition to any DC offsets that it might also be generating.

Methods and analysis
The experimental design is shown in Figure 1.Subjects either attended one of two frequency-tagged contrast-modulating gratings, while maintaining fixation at the fixation point (conditions 1 and 2), or attended a demanding letter task at the fixation point, thus ignoring both gratings (condition 3).
EEG was collected with a whole-head, 128-channel EEG system, and the three-dimensional locations of all electrodes were recorded.Signals were 0.1 Hz high-pass-filtered and 50.0 Hz (Bessel) low-passfiltered, and digitized at 600 Hz.The 30 second trials were binned in 60 0.5 second bins and bins with artifacts (e.g.eye blink) were removed.Distributed source reconstructions were made by using a leadfield-weighted minimum norm estimate [2].Cortical regions of interest were defined using fMRI retinotopic mapping and functional localizer scans.By combining these definitions with the sourceimaged EEG data, we extracted the mean cortical current density timecourses from visual areas V1, V3a, hV4 and MT+.
We performed spectral analysis on individual bins to compute the amplitudes of the responses corresponding to each of the frequency-tagged gratings in each visual area.Finally, we averaged these data across seven observers to generate a group mean.Both in V1 and in MT+, the 10Hz component, corresponding to VEP from the left grating is higher, when it is attended (A, C).Similarly, the 12Hz, right grating component, is higher when the right grating is attended.Similar effects are seen in V3a and V4 (not shown).Error bars are 1 SEM, n=7.
Targets were short, low amplitude contrast increments and subjects indicated detection of a target by a button press within a time window of one second after a target had appeared.The contrast was determined for each subject before the experiment in order to ensure a correct response rate of ~75%.This detection rate ensured that the task was engaging enough to keep high attention levels and guaranteed a significant number of missed target presentations, allowing us to perform additional signal-detection analysis on the VEPs.

Results
In conditions 1 and 2 we found that the allocation of spatial attention to a target increases the amplitude of the frequency component corresponding to that target in all the studied areas, including V1 (Figure 2).We saw no significant systematic effect of attention on signal latency.Interestingly, signal levels of the 'ignored' gratings in conditions 1 and 2 were generally lower than those in condition 3 suggesting that attentional selection is more effective for well-separated targets or, perhaps, more necessary when those targets share common spatial features such as shape and spatial frequency.

Discussion
We have shown that population-averaged electrical responses in all the studied areas exhibit attentional modulation.These signals must be the result of dynamic neural processing involving axonal or spike currents.Thus, the fMRI results cannot be explained by a purely metabolic effect.Rather, the signal increases due to attention indicate that they result from a modulation of the ongoing firing rate, and suggest the presence of a multiplicative gain mechanism in addition to the additive function reported by Buracas and Boynton [1] and similar to that suggested by Li, et al [3].
The discrepancies between primate and Human could be due to neural differences in the different spieces, and that attention does not modulate neural signals in primates in as early levels of visual processing as in Humans.But a more likely explanation is that EEG and fMRI measures averages over the entire cell population and electrode recordings target specific subgroups of cell populations or laminar structures.Other neuronal populations will then respond more similar to the Human results.
The current experimental paradigm cannot determine whether the observed multiplicative gain is mainly a contrast or response gain.Future studies can address this.

Figure 1 :
Figure 1: Experimental layout.Two gratings are displayed below and to the left/right of a fixation point in the middle of the screen.During a 30 second trial, both gratings are ON/OFF flickering with a temporal frequency of 10 and 12 Hz, to allow for frequency tagging of the resulting EEG signals.The fixation point, contains a small arrow indicating the side to attend to (left in this figure), as well as an attention demanding letter task.The experiment contains three different trial types: attend left or right while maintaining fixation, or attend to the letter task at the fixation point.Black text in the figure is not shown to the subject.

Figure 2 :
Figure 2: Spatial attention increases steady state VEPs of the attended stimulus.A, B: VEPs in V1.C, D: VEPs in MT+.Blue bars are while the subjects attend to the right grating, and the red bars are while the subjects attend to the left grating.Both in V1 and in MT+, the 10Hz component, corresponding to VEP from the left grating is higher, when it is attended (A, C).Similarly, the 12Hz, right grating component, is higher when the right grating is attended.Similar effects are seen in V3a and V4 (not shown).Error bars are 1 SEM, n=7.