10.2 EFP modulation is correlated to behavior and spatial attention
10.2 EFP modulation is correlated to behavior and spatial attention
The results of Chapter 8 revealed a modest attentional modulation of the trial-averaged broadband γ-activity and a correlation between RT and the latency of evoked EFP responses.
Differences between fast and slow RT were found for both the EFP latency and the attentional modulation of the EFP. Behavioral choices and neuronal responses in V1 co-variate [Palmer et al. 2007]. The present findings emphasize the potential usage of EFPs for the prediction of perception and behavioral performance from meso-scale recordings. The found attentional modulation, however, was small and the experiment was performed with only one monkey. The experiment needs to be repeated with additional subjects to verify the present results. Furthermore, for future investigations of the attentional modulation of EFPs recorded from V1, it is necessary to use an alternative task design which potentially leads to higher attentional effects (as discussed in Chapter 8.4). This could enable a precise decoding of the focus of covert spatial attention using EFPs from V1. Fortunately, for the repetition of the current and the execution of future experiments, the non-binary positive reinforcement training, described in Chapter 9, will be helpful for a fast and reliable training of macaques.
Using this training strategy, the time between the start of the monkey training and the first signal recording from task-performing monkeys can be shortened, while the motivation and the task confidence of the monkeys is kept high.
10.3 Optimization of information decoding
To further improve the decoding performance, the placement of stimuli into areas of dense electrode coverage is suggested by Chapters 5 and 6. Stimulation in foveal and close-to foveal regions paired with high-density recordings in corresponding V1 areas could further increase stimulus differentiation because of the large cortical area representing the central part of the visual field [Tootell et al. 1988]. Within the analysis of most electrophysiological studies in this thesis, optimal-selective electrodes were selected for a given stimulus to decode visual information. Alternatively, the visual stimulation could have been adapted to fit
10.3 Optimization of information decoding
optimally into the RFs of single or multiple electrodes. This optimization of the visual stimulation for a pre-defined set of electrodes could lead to an improvement in the decoding of visual information and should therefore be tested in future studies that are focused on investigating the benchmarks of information decoding from EFPs originating from V1.
10.4 Transfer to the human brain
An important field of research that uses superficial extracortical recordings is the reestablishment of communication and motor control of severely paralyzed patients via BCIs.
A direct transfer of the present findings to human BCI applications is, however, beyond the scope of this thesis. Three major reasons challenge the direct translation of the present results to clinical applications.
1. The location of V1 on the brain’s surface is different in humans compared to macaques [Van Essen et al. 2001]. Only a small part of V1 representing the fovea, the central degrees and the vertical meridian of the visual field is exposed to the cortical surface [Horton & Hoyt 1991]. The rest of V1 and therefore the representation of the largest part of the visual field is concealed in the calcarine sulcus [Horton & Hoyt 1991]. Due to the restricted epidural access to predominantly the foveal parts of the visual field in human V1, investigations need to be expanded by studying EFPs recorded from regions in V1 that represent the foveal region.
2. The size of the brain, the cortical surface area and the cortical magnification differs between humans and non-human primates [Hubel & Wiesel 1974b; Schira et al. 2007;
Van Essen et al. 2001]. As a result, the spatial resolution of the same array differs between humans and non-human primates. Due to the fact that the size of the human brain is larger compared to the macaque brain, using the same array as in the present studies for human EFP recordings should lead to a denser sampling of the cortical surface. Nevertheless, investigations on humans are needed to enable an evaluation of the functional spatial resolution of the used electrode density in humans.
3. Patients, in contrast to the monkeys in the present studies, are usually not head fixated in clinical applications. Due to the retinotopic organization of V1, slight head and eye
10.4 Transfer to the human brain
movements could decrease the decoding performance because the cortical activity is changed by the displacement of the visual scene. In addition, the monkeys were highly motivated and trained. Patients are usually not in good health which could lead to a non-optimal performance and issues that are not present in non-human primate studies. In the circumstance that, for example, cLIS patients use visual BCIs, issues could arise from the moistination of the eyes which needs to be artificially maintained because these patients are unable to make eye blinks [Bauer et al. 1979].
Despite these challenges that impede a direct translation of the present findings to the human brain, they demonstrate that EFPs represent the differentiated neuronal activity evoked by stimulus attributes in high resolution and that EFPs can be correlated to attention and behavior. Even though the dura mater attenuates the neuronal signal recorded with high-density arrays [Bundy et al. 2014], the attentional modulation of neuronal populations from V1 was found to be represented in EFPs. According to the similarity of the macaque and the human visual system [Newsome & Stein-Aviles 1999], these findings provide evidence for the applicability of EFPs for future BCIs. In case that further investigations fail to precisely decode the spatial location of covert visual attention from V1, the usage of high-density EFPs from visual areas that have shown a stronger attentional modulation in comparison to V1, for example from V2 or V4 [Buffalo et al. 2010; Luck et al. 1997; Mehta et al. 2000], could enable this decoding approach. Due to the beneficial properties of EFPs, the present findings suggest further investigations of EFPs with regard to their application in BCIs.
10.5 Outlook
Altogether, high-density epidural recordings from V1 reflect the cortical activity beneath the electrodes accurately and therefore visual information can be precisely extracted from high-density EFPs. Visual BCIs could benefit from epidural recording from V1 because of V1’s richness of extractable information. V1 was, however, targeted because of its beneficial properties for the objective of this thesis. Due to the high specificity of EFPs, the present findings indicate that high-density epidural arrays promise meso-scale mildly-invasive superficial extracortical recordings not only from V1 but also from other cortical areas.
Unfortunately, EFPs are rarely used (for a detailed tabular list of studies based on
high-10.5 Outlook
density ECoG and high-density EFP, see [Wang et al. 2017]). Until now, subdural recordings dominate in clinical application [Jayakar et al. 2016; Kovac et al. 2017; Reif et al. 2016;
Sperling 1997] and BCI research [Slutzky & Flint 2017]. The lack of long-term studies investigating the stability of epidural recordings (as discussed in Chapter 3.2.1) as well as the limited knowledge about its specificity impedes a common usage of EFPs for long-term clinical applications. Thus, epidural recordings could be an equivalent alternative to subdural and intracortical recordings if the knowledge about epidural recordings is expanded. This thesis contributes to the literature which emphasizes the usage of epidural recordings because of its high specificity. Epidural recordings expand the variety of neurophysiological methods rather than replacing other options. My thesis demonstrates the beneficial properties of EFPs and constitutes reasons to consider epidural recordings as mildly-invasive, alternative neurophysiological method for research and also for clinical application.