Electrodes implanted in the brain typically record two types of brain signals: the microscopic signal of the activity of individual neurons (recorded as short-duration electrical “spikes”) and the macroscopic local field potential (LFP), which is generated by the summed activity of thousands of neurons. In general, neural spiking is thought to convey the “content” of the brain’s representation of the external world while the LFP signal reveals the “context” in which this representation is encoded. For example, during exploration of an environment, cells in the rodent hippocampus, a temporal lobe structure implicated in long-term memory, discharge selectively at particular places in an environment. Thus, the ongoing activity of many (tens to hundreds) simultaneously recorded hippocampal “place cells” can be used to precisely “decode” where the animal is located in a given environment.

Conversely, when the animal is actively exploring an environment, the hippocampal LFP signal oscillates at the “theta” (5 to 10 Hz) frequency, reflecting the macroscopic structure of activity. This cyclic activity structure reflected in the LFP has been found to organize the firing of individual place cells active in nearby parts of the environment into discrete information “packets” and is thought to thus organize place cell coding of location. In this vein, changes in the internal brain regime, for instance those found between sleep and waking, are reflected in the hippocampal LFP. Importantly though, since the LFP reflects the mixed activity of thousands of place cells coding for disparate areas of space, the LFP signal was thought to “average away” the precise information content of location found in the activity of individual place cells.

In a recent paper published in the journal Science, Dr. Gautam Agarwal and colleagues find that, consistent with current theory, the LFP recorded at individual sites within the hippocampus does not provide information content about an animal’s location. Surprisingly, however, they report that subtle heterogeneities between the LFP signals recorded at anatomically distributed sites in the hippocampus can be used to “decode” an animal’s location and that this LFP-based decoding can be as precise as the well-established place cell–based decoding method.

In order to arrive at this surprising result, the authors relied on a method derived from radio signal transmission in which the part of a signal common to many channels (i.e., many anatomically distributed hippocampal sites) is treated as the “carrier” of a “signal” encoded in the subtle local deviations away from the global “carrier” frequency. These results have many implications, including in furthering the understanding of how the hippocampus performs the anatomically distributed computations involved in memory. Moreover, since the LFP signal (also called EEG or ECoG when recorded superficially to the brain) can typically be recorded in a less invasive manner than is required for the recording of individual neurons, these findings may have important implications for the future development of brain–machine interfaces, such as those used in brain-controlled limb prosthesis.

—Andres Grosmark, PhD

Read the paper “Spatially distributed local fields in the hippocampus encode rat position” in Science, published May 9, 2014.