Abstract
At least three neurofeedback techniques can produce measurable behavioral and neurophysiological changes. These involve modulating slow-cortical potentials (SCP), fMRI BOLD signals, or amplitude (or power) in narrow EEG bands. We think SCP neurofeedback may work indirectly by conditioning task-specific attentional focus, whereas fMRI neurofeedback directly conditions regional brain activity during task performance. However, the object of conditioning in EEG-band training is unclear because there is no overt task to perform and the electrophysiological and behavioral changes can occur rapidly. In a previous double-blind, placebo controlled experiment, we trained participants to increase EEG amplitude in narrow bands (C3-SMR, C4-SMR, C3-Beta, sham) using auditory rewards and 30-minute training sessions over five consecutive days. We observed widespread changes in ERP/ERSP patterns for reward stimuli during training. However, we did not find significant conditioning of trained bands within or across sessions. We also observed improved performance of a hemispheric attention task in the C3-Beta group after five sessions. To explain these and related results we are developing a four-part model: 1) Introspective tasks such as mind wandering during rest or autobiographical memory, engage a Switching Network which activates an internally-directed Default Mode Network and deactivates externally-directed problem-solving networks. 2) The DMN activates a "self-control" system which links rewards to internal states of arousal, motor activation, attentional focus, or cognitive engagement. 3) The internal states co-vary with synchrony in fundamental EEG "atoms," which we estimate with multi-way parallel factor analysis of high-density EEG recordings. Each atom represents one oscillatory EEG mode in the 1-40 Hz range, with a unique spectral envelope, topographical power distribution, and pattern of connectivity. 4) When the self-control system is engaged, neurofeedback rewards condition internal state modulations or "responses" which cause EEG atom synchrony levels to change accordingly. Thus EEG-band training conditions the self-control system, which can then manage internal states and global network connectivity.
Go back