Individual versus task differences in slow potential generators.

Average slow potentials (SPs) can be computed from any voluntary task, minimally involving attention to anticipated stimuli. Their topography when recorded by large electrode arrays even during simple tasks is complex, multifocal, and its generators appear to be equally multifocal and highly variable across subjects. Various sources of noise of course contaminate such averages and must contribute to the topographic complexity. Here, we report a study in which the non-averaged SP band (0 to 1 Hz) was analyzed by independent components (ICA), from 256 channel recordings of 18 subjects, during four task conditions (resting, visual attention, CPT, and Stroop). We intended to verify whether the replicable SP generators (between two separate day sessions) modeled as current density reconstruction on structural MRI sets were individual-specific, and if putative task-related differences were systematic across subjects. Typically, 3 ICA components (out of 10) explained SPs in each task and subject, and their combined generators were highly variable across subjects: although some occipito-temporal and medial temporal areas contained generators in most subjects; the overall patterns were obviously variable, with no single area common to all 18 subjects. Linear regression modeling to compare combined generators (from all ICA components) between tasks and sessions showed significantly higher correlations between the four tasks than between sessions for each task. Moreover, it was clear that no common task-specific areas could be seen across subjects. Those results represent one more instance in which individual case analyses favor the hypothesis of individual-specific patterns of cortical activity, regardless of task conditions. We discuss this hypothesis with respect to results from the beta band, from individual-case fMRI studies, and its corroboration by functional neurosurgery and the neuropsychology of focal lesions.

Higher similarity in beta topography between tasks than subjects.

We have recently provided evidence for highly idiosyncratic topographic distributions of beta oscillations (as well as slow potentials) across individuals. More recently, by emphasizing the analysis of similarity instead of differences across tasks, we concluded that differences between an attention task and quiet resting may be negligible or at least unsystematic across subjects. Due to the possibility that individual differences could be due to noise in a wide sense or some inherent instability of beta activity, we designed a replication study to explicitly test whether pairs of individuals matched for head size and shape would still present less similar beta topography than each individual between sessions or tasks. We used independent component analysis (ICA) for an exhaustive decomposition of beta activity in a visual attention task and in quiet resting, recorded by 256-channel EEG in 20 subjects, on two separate days. We evaluated whether each ICA component obtained in one task and in one given individual could be explained by a linear regression model based on the topographic patterns of the complementary task (correlation between one component with a linear combination of components from complementary conditions), of the same task in a second session and of a matched individual. Results again showed a high topographic similarity between conditions, as previously seen between reasoning and simple visual attention beta correlates. From an overall number of 16 components representing brain activity obtained for the tasks (out of 60 originally computed where the remaining were considered noise), over 92% could satisfactorily be explained by the complementary task. Although the similarity between sessions was significantly smaller than between tasks on each day, the similarity between sessions was statistically higher than that between subjects in a highly significant way. We discuss the possible biases of group spatial averaging and the emphasis on differences as opposed to similarities, and noise in a wide sense, as the main causes of hardly replicable findings on task-related forms of activity and the inconclusive state of a universal functional mapping of cortical association areas.

Individual topographic variability is inherent to cortical physiology but task-related differences may be noise.

The observation of highly variable sets of association neocortical areas across individuals, containing the estimated generators of Slow Potentials (SPs) and beta oscillations, lead to the persistence in individual analyses. This brought to notice an unexpected within individual topographic similarity between task conditions, despite our original interest in task-related differences. A recent related work explored the quantification of the similarity in beta topography between largely differing tasks. In this article, we used Independent Component Analysis (ICA) for the decomposition of beta activity from a visual attention task, and compared it with quiet resting, recorded by 128-channel EEG in 62 subjects. We statistically tested whether each ICA component obtained in one condition could be explained by a linear regression model based on the topographic patterns from the other condition, in each individual. Results were coherent with the previous report, showing a high topographic similarity between conditions. From an average of 12 beta component maps obtained for each task, over 80% were satisfactorily explained by the complementary task. Once more, the component maps including those considered unexplained, putatively "task-specific", had their scalp distribution and estimated cortical sources highly variable across subjects. These findings are discussed along with other studies based on individual data and the present fMRI results, reinforcing the increasingly accepted view that individual variability in sets of active neocortical association areas is not noise, but intrinsic to cortical physiology. Actual 'noise', mainly stemming from group "brain averaging" and the emphasis on statistical differences as opposed to similarities, may explain the overall hardship in replication of the vast literature on supposed task-specific forms of activity, and the ever inconclusive status of a universal functional mapping of cortical association areas. A new hypothesis, that individuals may use the same idiosyncratic sets of areas, at least by their fraction of activity in the sub-delta and beta range, in various non-sensory-motor forms of conscious activities, is a corollary of the discussed variability.

Lack of systematic topographic difference between attention and reasoning beta correlates.

Based on previous evidence for individual-specific sets of cortical areas active during simple attention tasks, in this work we intended to perform within individual comparisons of task-induced beta oscillations between visual attention and a reasoning task. Since beta induced oscillations are not time-locked to task events and were first observed by Fourier transforms, in order to analyze the cortical topography of attention induced beta activity, we have previously computed corrected-latency averages based on spontaneous peaks of band-pass filtered epochs. We then used Independent Component Analysis (ICA) only to single out the significant portion of averaged data, above noise levels. In the present work ICA served as the main, exhaustive means for decomposing beta activity in both tasks, using 128-channel EEG data from 24 subjects. Given the previous observed similarity between tasks by visual inspection and by simple descriptive statistics, we now intended another approach: to quantify how much each ICA component obtained in one task could be explained by a linear combination of the topographic patterns from the other task in each individual. Our hypothesis was that the major psychological difference between tasks would not be reflected as important topographic differences within individuals. Results confirmed the high topographic similarity between attention and reasoning beta correlates in that few components in each individual were not satisfactorily explained by the complementary task, and if those could be considered "task-specific", their scalp distribution and estimated cortical sources were not common across subjects. These findings, along with those from fMRI studies preserving individual data and conventional neuropsychological and neurosurgical observations, are discussed in support of a new functional localization hypothesis: individuals use largely different sets of cortical association areas to perform a given task, but those individual sets do not change importantly across tasks that differ in major psychological processes.