High-frequency oscillations in distributed neural networks reveal the dynamics of human decision making

We examine the relative timing of numerous brain regions involved in human decisions that are based on external criteria, learned information, personal preferences, or unconstrained internal considerations. Using magnetoencephalography (MEG) and advanced signal analysis techniques, we were able to non-invasively reconstruct oscillations of distributed neural networks in the high-gamma frequency band (60-150 Hz). The time course of the observed neural activity suggested that two-alternative forced choice tasks are processed in four overlapping stages: processing of sensory input, option evaluation, intention formation, and action execution. Visual areas are activated first, and show recurring activations throughout the entire decision process. The temporo-occipital junction and the intraparietal sulcus are active during evaluation of external values of the options, 250-500 ms after stimulus presentation. Simultaneously, personal preference is mediated by cortical midline structures. Subsequently, the posterior parietal and superior occipital cortices appear to encode intention, with different subregions being responsible for different types of choice. The cerebellum and inferior parietal cortex are recruited for internal generation of decisions and actions, when all options have the same value. Action execution was accompanied by activation peaks in the contralateral motor cortex. These results suggest that high-gamma oscillations as recorded by MEG allow a reliable reconstruction of decision processes with excellent spatiotemporal resolution.