TL;DR
Intracortical microstimulation (ICMS) doesn't just "turn on" neurons; its effects are deeply influenced by a neuron's current state and the history of recent stimulation. This study reveals that ICMS in awake primates triggers a brief excitatory spike followed by a long inhibitory period (up to 150ms), with the success of stimulation depending on the neuron's spontaneous firing rate and the timing of its last natural spike.
What Problem Does This Paper Address?
Intracortical microstimulation is a cornerstone of modern neuroscience, used to map brain functions, treat neurological disorders, and provide sensory feedback in brain-computer interfaces (BCIs). However, most of our "rules" for how ICMS works come from anesthetized animals.
In an awake, behaving brain, neurons are constantly firing and receiving inputs. There is a significant gap in our understanding of how this background activity—the "internal state" of the cortex—interacts with electrical pulses. If we want to build precise neuroprosthetics, we need to know if a pulse delivered at 10:00 AM has the same effect as one at 10:01 AM, and how the local network's inhibitory circuits might "mute" or "amplify" those signals.
Methods
The researchers worked with three macaque monkeys (Macaca nemestrina) implanted with 96-channel Utah microelectrode arrays in the primary motor cortex (M1).
- Stimulation: They delivered single-pulse ICMS (biphasic, 200µs phase width) at various amplitudes (5–50 µA) and frequencies (2–20 Hz).
- Recording: While stimulating one channel, they recorded single-unit activity (individual neurons) from all other channels on the array.
- Novelty: Unlike many studies that look at averaged responses over an entire session, this team analyzed responses on a trial-by-trial basis. They specifically looked at how the timing of a neuron's spontaneous spikes (the ones the monkey's brain generated on its own) affected the probability of the electrode evoking a spike.
Datasets
The study utilized original experimental data from three male pigtail macaques.
- Primary Dataset: 148 distinct neural units recorded during tonic (regular) stimulation.
- Secondary Dataset: 18 distinct units recorded during Poisson-distributed (randomly timed) stimulation to better isolate the effects of spike timing.
- Comparison: The researchers compared "real" spike distributions against "shuffled" versions to ensure that correlations between neurons weren't just statistical artifacts of the network's overall activity level.
Key Insights and Ideas
- The "Reset" Effect: ICMS typically causes a spike followed by a "silence" (inhibition). The researchers found that if you hit a neuron with a second pulse during its inhibitory period, you can still evoke a spike, but the "inhibition clock" essentially resets. This suggests that the inhibitory period isn't a hard refractory state but a dynamic suppression.
- State-Dependency: A neuron’s "readiness" to respond to ICMS is not constant. If a neuron had fired spontaneously just before the stimulus, it was much less likely to respond to the electrical pulse. This confirms that intrinsic membrane properties (like after-hyperpolarization) play a massive role in whether ICMS "takes."
- Excitation and Inhibition are Decoupled: On a single-pulse basis, whether or not a neuron spiked didn't change the duration of the inhibition that followed. This implies that the excitation (direct or monosynaptic) and the inhibition (likely via GABAergic interneurons) are driven by different circuit mechanisms.
- Frequency-Dependent Fatigue: When stimulating at higher frequencies (e.g., 20 Hz), the probability of evoking a spike often decreased over time, while the duration of inhibition changed. This suggests that repetitive stimulation induces short-term plasticity or "fatigue" in the local circuit.
Key Results
- Latency: Evoked spikes occurred very quickly, typically between 1.5 and 10 ms after the pulse.
- Inhibition: The excitatory spike was followed by a period of suppressed firing lasting 5 to 150 ms.
- Distance & Amplitude: As expected, units closer to the electrode (<1 mm) had a higher probability of spiking and shorter latencies. The relationship between current amplitude and response followed a sigmoidal (S-shaped) curve, with a sharp "threshold" for activation.
- Correlation with Firing Rate: In 51% of units, the probability of evoking a spike was positively correlated with the neuron's spontaneous firing rate. Conversely, the duration of inhibition was often negatively correlated with the firing rate.
Limitations and Future Work
The authors acknowledge that using extracellular recordings makes it difficult to definitively identify cell types (e.g., distinguishing excitatory pyramidal cells from inhibitory interneurons), though they attempted to use spike-width analysis as a proxy.
Future research could use optogenetics to target specific cell types or pharmacological blocks to confirm exactly which GABA receptors (A vs. B) are responsible for the long inhibitory periods observed. Additionally, since this study focused on the motor cortex, it remains to be seen if these "rules of engagement" for ICMS hold true in sensory or associative areas of the brain.
Summary updated