Research

Cognitive – Translational Brain Dynamics Laboratory: —RESEARCH AREAS
Our main scientific focus is on selective attention and sensory/motor integration. We work on neurophysiological studies, and on directly integrating neurophysiological research. The studies entail sampling laminar profiles of synaptic activity as indexed by current source density (CSD) analysis of local field potentials (LFPs) and the associated neuronal firing using multiple multielectrode arrays positioned in different brain areas. Our parallel studies using ElectroCorticoGraphic (ECoG) recordings in surgical epilepsy patients provide an ideal vehicle for bridging the gap between studies; ECoG measures can be directly compared to invasive LFP recordings, and can serve as a bridge between activity measures at the micro-scale and the more macroscopic scale of noninvasive EEG and fMRI measures. Through its connections to ongoing clinical studies at CUMC and NKI, the findings of our studies have numerous points of relevance to the developing understanding of the brain mechanisms of neurological/cognitive disorders.

Integrative-translational studies:To optimize the understanding of the human brain, it is necessary to develop rock-solid model systems that closely approximate the human, yet allow application of high-resolution cellular and molecular techniques. We have developed behavioral paradigms and we adapted human studies to using paradigms [e.g., (Schroeder et al., 1991; Mehta et al., 2000)]. Using an experimental model, we pioneered both multielectrode technologies such as the linear array multielectrode, and analytic procedures such as the combination of field potential, current source density (CSD) and multi-unit action potential (MUA) analyses that allow controlled systematic study of synaptic processes and related neuronal firing patterns over distributed neuron populations. We have demonstrated that we can distinguish ascending and descending influences, functionally, in laminar activity profile [e.g., (Schroeder et al., 1998)], and that specific cellular processes in particular cortical layers can be connected to network dynamics (Besle et al., 2011).

Neurogenesis of (local) field potentials (FPs): Field potentials are critical tools in the systematic and mechanistic study of brain operation, and in the process of applying findings toward understanding the physiology of the human brain. We address the fundamental mechanistic understanding of FPs, as described above, by relating FPs to their specific physiological underpinnings and by connecting them to the dynamics of ongoing neuronal activity [e.g., (Shah et al., 2004)]. A major controversy surrounding the study of FPs stemmed from the assumption that because an FP recorded within the brain, it is by definition a local FP or LFP. Motivated by early findings [e.g., (Schroeder et al., 1992)] showing that due to volume conduction, FPs generated in the base of the thalamus could be recorded at the outer brain surface, we investigated mechanisms of volume conduction (Kajikawa and Schroeder, 2011), and subsequently, developed a volume conductor model to guide further use and interpretation of the FP (Kajikawa and Schroeder, 2015).

Combining physiological, anatomical and computational methods to functionally dissect brain circuitry: Prior to laminar electrophysiological studies (above), sensory and cognitive studies entailed recordings from neurons whose laminar locations (and connectivity patterns) were unknown. In partnership with anatomists we have combined anatomy with physiology studies [e.g., (Hackett et al., 2007; Falchier et al., 2009)]. We used these methods to define the interaction of extralemniscal “modulatory” (Layer 1) inputs with lemniscal “driving” (Layer 4) inputs to promote multisensory integration at the level of primary auditory (A1) cortex (Lakatos et al., 2007). Early on (Tenke et al., 1993), we developed a computational model of the laminar activation profile in V1, and later, conducted a series of studies with Mingzhou Ding’s lab to use Granger causality analysis to understand laminar patterns of attentional dynamics at numerous stages of the visual pathways (Bollimunta et al., 2011). We are collaborating with the Kopell lab at BU, to develop an iterative exchange between a cellular/computational model and laminar electrophysiology in order to explore and represent the neuron ensemble processes that generate rhythmic entrainment in A1.

Fundamental significance of neuronal dynamics: The view that neuronal oscillations are not incidental, but rather, are instrumental (literally, essential tools) in most brain operations reflects an ongoing paradigm shift in Neuroscience. Contrasting with widespread focus on gamma (30-60 Hz) and higher frequencies, our work over the last 10 years has shown that delta (1-4 Hz) range oscillatory phase regulates amplitude in higher (theta/gamma/high gamma) frequency oscillations to underpin attentional selection in local neuron populations (Lakatos et al., 2005; 2007; 2008), and at a network level as measured with ECoG [(Zion Golumbic et al., 2013); see also Besle et al., 2011]. Combined with stimulus selectivities in local neurons low frequency dynamics form compound (e.g., spectrotemporal) filters (Lakatos et al., 2013). We have devised a broad conceptual framework to describe and predict the dynamic engagement of neuronal dynamics specific frequency ranges by the nature of task demands (Schroeder and Lakatos, 2009).

Active Sensing as an emerging paradigm and perspective: A second, and related shift in our research paradigm stems from the growing recognition most sensory input actively acquired (Schroeder et al., 2010). “Active Sensing” is a term borrowed from robotics, which refers to use of a sensor or detector device that requires input energy from a source other than that which is being sensed. While biological sensors like the eyes and fingers have been thought of mainly as passive, transducing input energy into a neuronal code, closer examination of the manner in which humans and other animals gather data from the environment suggests that overall, it is more of an active process in which data are acquired using a motor sampling routine. Attention is an essential component of Active Sensing. The fact that many of the motor and attentional influences that power Active Sensing are inherently rhythmic, immediately draws a strong mechanistic connection between motor/attentional control of brain rhythms and active sensing (Morillon et al., 2015). While the auditory system has been proposed to be mainly a passive receiver that operates differently from all other motor driven sensory sampling recent findings suggest, that while data sampling is often covert (attentional) auditory processing can be strongly enhanced by motor behavior, and thus, fits very well into the Active Sensing scheme (Morillon et al., 2014). In key respects, this Auditory Active Sensing framework extends to speech processing [Zion-Golumbic et al., 2013, cited above; (Morillon and Schroeder, 2015)], in a way that is reminiscent of the much-debated “Motor Theory of Speech Perception.”