Speed cells in the medial entorhinal cortex

EMILIO KROPFF; ERIC CARMICHAEL; EDVARD I MOSER; MAY-BRITT MOSER

Chicago

Conferencia; Society for Neuroscience Annual Meeting 2015; 2015

Grid cells in the medial entorhinal cortex (MEC) and surrounding parahippocampal areas are unique in their spatial code(Hafting et al, 2005). Unlike other place-modulated neurons, their population firing pattern not only repeats periodicallywithin a given environment, but also seems to apply equally to all explored environments (Fyhn et al, 2007), reflecting theuniformity of space despite the unevenness of contextual details. This property makes grid cells ideal candidates for apath integration-based representation of space (McNaughton et al, 2006). In such a scheme, running speed is integratedacross short time windows to obtain the instantaneous displacement of the animal,which in conjunction with headdirection input is used to update the representation of the animal´s position. Any path integration mechanism thus requiresrunning speed as a major input.However, while speed has been reported to correlate marginally with entorhinal thetafrequency (Jeewajee et al, 2008) and firing rate of grid cells (Sargolini et al, 2006; Wills et al, 2012), the existence andnature of a reliable and locally available speed signal has remained unclear.We found that running speed is represented in the firing rate of a ubiquitous but functionally dedicated population of MECneurons (~15%) with low levels of overlap with other MEC populations, such as grid, head direction and border cells.Speed cells are characterized by a positive, linear response to running speed, with a great variability in slope andy-intercept. This response is context-invariant, so that running speed can be decoded from speed cell activity even acrossrooms or in darkness without loss of accuracy. Speed cells and grid cells share a similar prospective bias of around 50 to80 ms. In both cell types, prospective activity is highly modulated by theta rhythm, suggesting that path integration couldtake place on a theta cycle basis (Navratilova et al, 2012). In the hippocampus, speed cells were also identified in asmaller proportion (~10%) and with a higher overlap with the place cell population. These cells, however, had aretrospective behavior, while place cells had no temporal bias, a combination that seems to be incompatible with pathintegration. Taken together, these observations point to MEC speed cells as a key component of the dynamicrepresentation of self-location.