Calendar

A fundamental goal in the neurobiology oflanguage is to understand how acoustic information in speech istransformed intomeaningful linguistic content. To achieve this goal, researchers have employed model-based methods to determine how acoustic, phonetic, or higher order linguistic cues map onto neural representations. However,such approaches are limited in that they requirea prioriknowledge of the features to be modeled. In this talk, I show how we employed data-drivencomputational methods on anextensive set of high-density intracranial recordings from 27 patients to reveal the existenceofa spatially-localized region of the pSTGthat specifically parses acousticonsets. By combining unsupervised methods with model-based approaches, we relate these results to previous work on phonetic feature and spectrotemporalrepresentations. Our findings demonstrate a fundamental organizational propertyof the human auditory cortexthat has been previously unrecognized.
Context-dependent grid cell activity in the human entorhinal cortex
Dr. Zoltan Nadasdy, Research Scientist
The spatially periodic activity in the entorhinal cortex, ahallmark of ‘grid cells’ in the rodent brain, is considered evidence for an internal coordinate system of the world around us. These cells in the entorhinal cortex (EC), together with the hippocampus, allow for localizing ourselves relative to our environment. The three defining features of grid cell activity, scale invariance, orientation anchoring to distant cues, and a hexagonal tessellation pattern, suggest a robust context independence, confirmed by extensive research on rodents. Recent primate and human data validated the prevalence of spatial periodicity in the EC, yet the environmental-dependency of grid patterns has not been studied on primates. Patients implanted with electrodes in the EC for localization of non-tractable seizures performing virtual navigation tasks enabled us for the first time to investigate the relationship between grid geometry and environmental features. We argue based on direct single cell electrophysiology that, in contrast with rodents, the spatially periodic activity of human EC is highly context-dependent. Grids appear to linearly scale with the size of environments, orient to corners, and display a broad range of tessellation patterns, including Cartesian patterns. These results suggest that neurons in human EC are endowed with an adaptive flexibility and larger dependency on visual input than those observed in rodents. The novel properties of grid cells, together with the place cell system, provide a more accurate understanding of human spatial navigation that may take us a step closer to be able to decode positional information directly from the brain. This may open new perspectives with regards to implantable neuronal interfaces for the treatments of neurological disorders to repair the torn spatio-temporal fabric of episodic memory.