Files
Abstract
Prosthetic limbs controlled with brain-machine interfaces have the potential to confer to patients with tetraplegia the ability to physically interact with the world and regain a level of independence, but the dexterity of prosthetic hands is severely limited without sensory feedback. Accordingly, efforts are underway to restore tactile sensation through prosthetic limbs. Current attempts to convey sensory feedback focus on conveying basic information about object interactions to support simple manual behaviors such as grasping. This includes which part of the hand is touching the object, how much pressure is exerted, and when contact occurs. To convey information about the object itself – its texture and compliance for example – would improve the dexterity of bionic hands, but it is much more challenging. One of the principles that guides sensory encoding algorithms is that of biomimicry: to the extent that the patterns of neuronal activation elicited through electrical stimulation mimic natural patterns, the resulting sensations will be natural and thus intuitive. General principles of tactile neural coding can be invoked to produce patterns that are as natural as possible. One unresolved coding principle is whether spike timing in the central nervous system shapes tactile sensations. Neurons throughout the nervous system exhibit patterns of activation that contain information not only in the strength of their response but also in the distribution of spikes across time at single-digit millisecond time scales, and it has proven extremely difficult to disentangle the effects of spike timing and rate. The objective of the study presented in chapter 2 was to assess the degree to which spike timing shapes pitch perception. To this end, we trained monkeys to discriminate the frequency of vibrations delivered to the skin while we recorded the responses of neurons in somatosensory cortex (SC). To disentangle the contribution of rate and timing to frequency perception, we varied the amplitude of the vibrations such that the firing rates of the SC response – and the intensity of the stimulus – would vary independently of the behaviorally relevant parameter, frequency. We found that a rate-based code could not account for the ability to discriminate frequency, implicating phase-locking in pitch perception. We conclude that temporal spiking patterns play a key role in tactile pitch perception, and that the temporal patterning of ICMS pulses could, in principle, be manipulated to shape artificial touch. However, whether the manipulations of ICMS temporal patterning will result in interpretable sensations is unclear. This question can be addressed empirically by studying the perceptual correlates of ICMS timing. Chapter 3 presents a behavioral assessment of the perceptual effects of stimulation frequency in SC. The results show that animals can discriminate ICMS frequency despite concomitant and uninformative variations in amplitude, demonstrating that frequency has a distinct impact on sensation from amplitude. These experiments add to a growing body of evidence showing that temporal spiking patterns in somatosensory cortex shape tactile perception. Accordingly, modulation of pulse timing – which in turn shapes spike timing in the activated population – can be used to convey stimulus information in a biomimetic way.