The ability to record neural signals simultaneously from many sites with chronically implanted electrodes represents a difficult technical challenge which, if surmounted, could greatly enhance our understanding of the activity of networks of neurons, and particularly of learning-related changes in them. Most multi-electrode implants for chronic neural recordings use arrays of electrodes rigidly fixed in place for the duration of the implant. These arrays are implanted surgically and have so far been used mainly in cortical areas. We have developed an alternative method using custom-designed screw micro-drives and recording chambers along with conventional microelectrodes to record from head-fixed or unrestrained monkeys. Using a custom-built grid with precisely machined holes (1mm center-to-center), we have been able to record single-unit and LFP activity from cortical and sub-cortical structures accessible through a large (~1600mm2) craniotomy over the frontal or parietal lobes. Recordings can be obtained from multiple structures with a minimal track spacing of 1-2mm. By adjusting the depth of each electrode (upward or downward) over the course of an implant, we can improve the isolation of units and record from new units at multiple depths along each track. Implants can be removed and reconfigured to target different regions or different tracks within the same region. We have used the method to implant repeatedly a total of 11 monkeys. We have recorded from as many as 126 electrodes implanted bilaterally, and for up to 9 months at a time in subsets of PFC, M1, SMA/pre-SMA, CMA, ACC, PM, FEF, SEF, parietal cortex, amygdala, nucleus accumbens, caudate nucleus, putamen, thalamus and globus pallidus. Following initial stabilization of the implants (< 3 weeks), we typically recorded well-isolated single units from 30-60% of electrodes in each session. Different subsets of electrodes had units in different sessions, but some electrodes had extraordinarily stable unit activity across weeks of recordings. In a typical implant, every electrode had single unit activity in at least one session. We are currently exploring new techniques to improve the daily yield. Our main effort is geared toward reducing the presumed gliosis associated with chronic implants. Our method has a number of advantages in this respect. First, the relative sparseness of our implants reduces the risk posed to neighboring electrodes from a localized inflammatory response. Second, once implanted, individual electrodes can be moved in a controlled and gradual manner, in order to improve stability and to minimize tissue damage around the tips. Third, if necessary, implants can be extracted and the monkey can be re-implanted with fresh electrodes in a matter of weeks. Fourth, the method can be adapted to work with many types of electrodes, enabling flexibility in battling gliosis. Ultimately, our method facilitates the search for changes within and among brain regions across multiple recording sessions, and promises to be a powerful tool for studying the neural basis of learning.
This work supported by: NEI, NINDS (Javits), NPF