Our research attempts to identify, locate and characterize the neuro-cognitive stages used to encode and interpret events. Initially, we stimulated and recorded from electrodes that had been implanted into the human brain for seizure localization. These recordings included both action-potentials from single or a few neurons (units), as well as population synaptic potentials (iEEG- the intracranial electroencephalogram). Other methods were incorporated as they were developed: positron-emission tomography (PET), functional magnetic resonance imaging (fMRI.), magnetoencephalography (MEG), and current-source density (CSD). We have also tested cognitive and emotional function before and after surgical therapy. Early neuro-cognitive stages tend to be localized and material-specific. For example, a small basal occipito-temporal region was shown using fMRI, MEG and iEEG to selectively encode faces at ~170 ms after they are shown. Unit, synaptic and spectral measures suggest that the early encoding areas may extract specific information, distribute it widely, and then become relatively quiescent. Later stages tend to be highly distributed and integrative. For example, during a stage at ~280 ms, cingulo-parieto-frontal sites concerned with attention and visceral control integrate an orienting response to arousing stimuli. During another stage, at ~400 ms, ventral occipito-temporo-frontal areas integrate complex stimuli within the current cognitive context. This is usually paired with a stage of cognitive closure at ~600 ms. We hypothesize that the 400/600 ms stages represent the divergent/convergent aspects of event-encoding. Our data suggests that over this timeframe, the focus of cortical activity shifts from layers concerned with bottom-up processing to those concerned with associative and top-down processing. Using both macroscopic (combined fMRI and MEG) and microscopic (laminar local population synaptic activity and cell-firing) we have shown that widespread areas are active simultaneously in a sustained and interactive manner. The components are related to spontaneous rhythms such as alpha, theta, sleep spindles and the K Complex, and thus to fundamental network activities such as cortical up- and down-states. The orienting component at ~280ms corresponds to a scalp event related potential (ERP) termed the ‘P3a;’ the cognitive integration component at ~400 to the ‘N400,’ and the cognitive closure component at ~600 to the ‘P3b.’ These have been studied intensively with respect to normal cognition, development/aging, and neurological/psychiatric disorders. These studies can be re-interpreted based on our new knowledge of their neural bases. During the integrative stage at 400 ms, we found that a part of the brain necessary for memory formation (the medial temporal lobe) strongly alters its synaptic activity as a word or face becomes more familiar, and that its individual cells are specifically responsive to a particular word or face within a given context. Areas involved in event-encoding decrease their response with priming, whereas areas specifically involved in remote episodic memory increase their firing and synaptic activity. Brief disruption of the medial temporal lobe at this time impairs both the encoding and retrieval of stimuli for recent memory. Artificial activation can provoke profound feelings of recognition and intense recollections. We proposed that the hippocampus interacts with cortical sites, forming a trace integrating different aspects of events, thus allowing events to be reconstructed from their parts. The insights and techniques derived from these basic studies have been used to help understand the neural bases of human epilepsy, alcoholism, and dementia. For example, we have described the intracortical processes generating epileptiform phenomena, and their modulation in the seizure focus. Currently, local cortical laser Doppler and point spectroscopy are being combined with laminar physiological measurements to understand the neuronal correlates of the BOLD response during human cognition. Our long-term goal is to solve a nested series of inverse problems to permit the neuronal circuit dynamics supporting human cognition to be inferred from non-invasive imaging.