Recording neuronal behavior during behavior with experimental microstimulation

J.W. Gnadt¹ and S.D. Echols²

¹Department of Neurobiology and Behavior, Stony Brook University, Stony Brook, New York, USA
²Riverbend Instruments, Inc., Birmingham, Alabama, USA

Over 100 years ago, Fritsch and Hitzig, Ferrier and other experimental biologists demonstrated the utility of using systematic electrical stimulation to reveal the working mechanisms of the nervous system in large vertebrates. Roughly 50 years later Lorente de No introduced the use of fine ‘needle’ electrodes to record neuronal action potentials from individual neurons in vivo. By the early 1970s, neurophysiologists like Evarts, Hyvärinen, Mountcastle and Asanuma had perfected techniques of chronic, in vivo neurophysiology and ‘micro’-stimulation to study neural function during highly trained behaviors in nonhuman primates. By 1990, Newsome and colleagues demonstrated that in vivo brain stimulation could be used to bias behavioral responses in monkeys behaviorally trained to report their introspective perception. However, because electrical stimulation pulses corrupt the neural recordings with large artifacts, neural action potentials that fall within the time course of the artifacts are completely obscured or badly corrupted, making them difficult to detect by conventional techniques. Until recently, this difficulty has presented an intractable limitation for the field of behavioral neurophysiology. The experimental electrophysiologist has been unable to reliably record neural activity simultaneously with experimental stimulation of the neural tissue.

Using an adaptive noise cancellation technique, we present a real-time method for recording neural activity during experimental microstimulation in actively behaving monkeys. This allows the simultaneous measurement of neural and animal behavior while using microstimulation to activate or perturb ongoing behaviors in intact, trained subjects. The adaptive filter uses a combination of one-shot and iterative learning to quickly cancel the spectral content of the corrupting artifact waveform that is time-locked to a triggered stimulation pulse, without interfering with spectral content of the underlying neural recording. Even during complete corruption of the neural recording during prolonged, repetitive microstimulation, we demonstrate accurate recovery of action potentials during stimulation-induced behaviors in real-time, regardless of the size, shape, or timing between action potentials and stimulation artifacts. Signal detection reliability for recovery of action potentials can achieve better than one error in 104. Because the noise cancellation is stimulus-event triggered, even stimulation artifacts with identical shapes as action potentials are accurately rejected.

Paper presented at Measuring Behavior 2005 , 5^th International Conference on Methods and Techniques in Behavioral Research, 30 August - 2 September 2005, Wageningen, The Netherlands.

© 2005 Noldus Information Technology bv