Co-written with Dr. Marissé Masis-Solano, M.D..
Imagine that you’re standing on one leg. To avoid falling down, you integrate multiple sources of information: visual motion of objects gives information about the velocity of your head, while proprioceptive input from the neck joint gives information about the head’s position relative to the body. Pressure sensations from the footpads and muscle activity in the lower limbs reflect lower body leaning. The loss of any these inputs — for example, closing your eyes — means you’re much more likely to fall down.
In this post, we’ll explore what happens when people lost perhaps the most important contributor to the sense of equilibrium: the vestibule. The vestibular apparatus is a sensory organ located in the inner ear. It consists of the semicircular canals, which measure angular acceleration, and the saccule and utricule, which measure linear acceleration and gravity. Together, they relay information about the head’s orientation in space, much like the gyros and accelerometers in a smartphone measure the device’s tilt.
When the head turns rapidly, the liquid in the 3 semi-circular canals stays in place due to its inertia, then slowly catches up, triggering hair cells within the ampulae of the canals. That leads to a rapid increase in firing rate in the vestibular nerve, which relays this information to the brain stem and cerebellum.
The first chapter of Doidge (2007) tells the fascinating story of people with bilateral vestibular damage (BVD). People with BVD sometimes call themselves wobblers: they can have normal equilibrium as long as their eyes are open and all motions are performed slowly. But when moving rapidly or when their eyes are closed, equilibrium is lost, resulting in a characteristic wobbly gait.
BVD is most often due to hair cell damage caused by ototoxic drugs. The aminoglycoside family of antibiotics in particular is associated with BVD (Roland 2003), which results in catastrophic loss of equilibrium. Gentamicin is the most frequent culprit, while amikacin and streptomycin can also damage both vestibular and cochlear cells. Balance can be affected in up to 15% of patients taking gentamicin (Fausti 1999, Fee 1980).
Equilibrium in people with BVD can be treated with sensory substitution, relaying information equivalent to that of the vestibule through another sense, and letting the brain’s plasticity take care of integrating the new information seamlessly. Doidge (2007) talks about a particular form of sensory substitution for people with BVD: BrainPort.
Sensory substitution for people with BVD
BrainPort came from a long line of development into sensory substitution for people suffering from blindness. In the early 70’s, the neuroscientist Paul Bach y Rita experimented with relaying visual information to blind people through their intact senses. He first built a chair with several actuators that could vibrate at multiple points on a person’s back. He connected a film camera to the actuators, relaying brightness information via vibration, and demonstrated basic recognition — blind people could identify the the orientation of lines and recognize large familiar objects after some experience with the chair.
Back-based systems are bulky; an ideal, miniaturized system would stimulate an area of a person’s body with a high density of touch receptors. As you can judge from the shape of the somatosensory homunculus, the tongue is highly over-represented in the brain, and it contains a very high number of touch receptors compared to its size. Furthermore, it comes with its own saline solution, saliva — it’s thus ideally suited to be directly stimulated through electric impulses with a miniature device. Thus was born BrainPort, a sensory substitution device which can relay visual information to a blind person though the tongue.
When Bach y Rita encountered people with BVD several years later, in the 2000s, he had a device on hand that he knew could help them — BrainPort. He attached a gyro to a plastic hard hat, filtered that information minimally, and relayed the information through BrainPort. When the patient tilted their head to the right, they got a jolt on the right side of the tongue; when they tilted their head forward, they felt a jolt at the front of the tongue, and so forth.
When the device was turned on, he saw an immediate and profound decrease in oscillatory movement in people with BVD. The mapping between vestibular and somatosensory information was sufficiently simple that very little training with the device was necessary.
Long-term effects of sensory substitution
Perhaps most intriguing is that after several months of training with the device, oscillatory movements decreased when the device was off. At first, improved equilibrium persisted for a few minutes after turning the device off; after having trained with the device for an hour a day for a few months, however, the improvement persisted first for a few hours, and then for a few days, and eventually where the improvement became persistent.
That might seem a little magical, but it seems to be driven by long-term plasticity. We know that following vestibular deinnervation, there’s behavioural compensation after an initial period of complete loss of equilibrium; subjects learn to compensate through proprioception, somatosensation, vision, and efference copy. Furthermore, what drives this compensation is plasticity, with multi-sensory neurons increasing their sensitivity to non-vestibular signal (Sadeghi et al. 2012).
To optimally fuse the information from several noisy sensors, more reliable signals should be integrated with a higher weight. BrainPort’s stickiness might work by giving the brain a reference signal against which it can measure the accuracy of other equilibrium sensors. Indeed, we know that single neurons in the parietal cortex perform this type of Bayesian sensor fusion (Fetsch et al. 2012), and might reweight the signals through plasticity using BrainPort as reference.
BrainPort could lead to long-term improvement in equilibrium by a second mechanism: by recalibrating the residual vestibular input. Not every hair cell dies after BVD; there’s residual signal, perhaps 2-5% of the original signal strength, that comes through the vestibular nerve.
The signals measured by the semi-circular canal are proportional to angular acceleration and jerk. To infer the head’s tilt, the signals must be integrated twice, which makes them vulnerable to drift. A small systematic error in the acceleration signal, due to an incorrect calibration, gets integrated over time, and eventually blows up to a large error. The brain now thinks the head is an entirely different position that it actually is.
Drift is not just a problem for the vestibule, it’s a general problem in angular accelerometers — it also afflicts solid-state gyros. To avoid this problem, in the vestibule as in electronics, angular acceleration must be integrated with another signal that doesn’t suffer from drift — linear acceleration. Unfortunately, BVD also affects the parts of the vestibule that respond to linear acceleration — the utricule and saccule. The external signal from Brainport can serve as a reference signal against which to calibrate all of these signals, decreasing drift over months of plasticity.
Practical sensory substitution methods
As impressive as BrainPort is for patients with BVD, this form of therapy hasn’t moved much beyond stage I clinical trials. The main reason Brainport hasn’t taken off is that walking around with a piece of electronic in one’s mouth looks silly, and makes talking difficult. The most commonly used therapy these days uses sensory stimulation via a belt with integrated haptics and balance board system.
These systems are, however, bulky and expensive. Because gentamicin is on the WHO list of essential medications, it’s available pretty much everywhere, including in rural hospitals in developing countries; that’s thousands of new people with BVD in developing countries every year.
Dr. Marissé Masis-Solano and I thus embarked on building a vestibular replacement system based on open-source electronics. Our goal was to make a system that would be easy to build for under a 100$. An IMU installed on the head would record its position, it would be beamed to a receiver device, and communicated via haptic motors installed in the palm. In the next few installments, we’ll show you how we’re building our solution.
Tyler M, Danilov Y, & Bach-Y-Rita P (2003). Closing an open-loop control system: vestibular substitution through the tongue. Journal of integrative neuroscience, 2 (2), 159-64 PMID: 15011268