| dc.description.abstract | Visual attention has been shown to modulate perceptual behavior and neuronal activity. Early psychophysical studies on attention showed that the subjects detected a target presented at the attended location better (i.e., increase in detection accuracy) and faster (i.e., reduction in reaction time) compared to unattended location. Electrophysiological studies in macaques have shown that attention increases the mean and reduces the variability of firing rate and spike count correlation between neurons. In addition, attention also modulates the power of local field potential (LFP) at different frequency bands like alpha (8-12 Hz), gamma (30-80 Hz) and high-gamma (80-200 Hz). Furthermore, the phase-relationship in the gamma-band is also modulated by attention and is hypothesized to play an important role in inter-areal communication and information transfer. Behavioral performance like detection accuracy is shown to depend on the power and phase alpha oscillation and reaction time is shown to vary with the phase of the gamma-band. These strong links of attention and behavioral performance with different neural measures, motivated us to study the effectiveness of these neural measures in decoding attention and behavioral performance.
In this thesis, we compared the ability of spikes, LFP power and phase-based measures to decode the locus of attention and behavioral outcome under different cueing conditions. To achieve this objective, we analyzed the neural data recorded simultaneously using two 48 channels microelectrode arrays from visual area V4 of both the hemispheres of two male rhesus macaques (Macaca mulatta) while they performed an orientation change detection task with valid, invalid and neutral cueing conditions.
For our first aim, we studied the effect of neutral cuing condition on LFP and compared it with valid and invalid conditions. We also compared the ability of single and multi-channel LFP power in different frequency bands and spikes in discriminating attentional conditions. We showed that the effect of neutral cue on LFP is intermediate to the attend and unattended condition, similar to spiking activity. We also found that single-channel LFP in high-gamma followed by gamma band discriminated the attention condition better than spikes. Surprisingly, when we considered multi-channel LFP and spikes to decode attention, spikes did not outperform the LFP. Together, these results not only highlight the usefulness of LFP in decoding attention but also provide insights into the spatial scale over which attention signals operate.
Our second aim was to examine the ability of single-channel LFP power and phase in discriminating the behavioral outcome and compare it with the attentional discriminability. We found that while high-frequency LFP power in high-gamma (122-198 Hz) and gamma band (42-78 Hz) discriminated attention the best, behavioral outcome was discriminated well by low frequency LFP power in alpha band (8-12 Hz) and steady-state visually evoked potentials (SSVEP; 20 Hz). These results indicate that LFP power carries information about different cognitive variables in separate frequency bands and this provides important clues about the mechanisms underlying attention and behavioral state.
For our third aim, we developed a method to obtain the single trial estimates of pairwise measures like correlation and phase consistency (PPC) which are traditionally computed across trials and examined their ability to discriminate attention and behavioral outcome. Our single trial measures of correlation and PPC showed attention and behavioral modulations similar to their mean-trial counterparts, validating our estimation method. Although the pairwise measures did not perform better than LFP power in discriminating attention or behavior, PPC in the gamma band discriminated the attention condition on par with firing rate. Therefore, despite the proposed functional importance of the phase-based measures in information transfer and communication, they were a suboptimal choice to discriminate attention or behavioral outcome. | en_US |
