Utah array · Grey Matter

The Utah array is a small bed of rigid silicon needles, each tipped with an electrode, pressed into the cortex to record single neurons, and it is the device behind most of what is known about brain-computer interfaces in people, from the first cursor control in 2004 to the speech decoders of 2023.

Utah array. The Utah array is a small bed of rigid silicon needles, each tipped with an electrode, pressed into the cortex to record single neurons, and it is the device behind most of what is known about brain-computer interfaces in people, from the first cursor control in 2004 to the speech decoders of 2023.

It was invented by Richard Normann at the University of Utah around 1989 and is made by Blackrock Neurotech. A 4 by 4 mm base carries a 10 by 10 grid of needles 1.0 to 1.5 mm long, of which 96 to 128 are wired, and a pneumatic inserter drives it into the cortex in a fraction of a second. Each tip records the spikes of the few neurons within reach and the local field potential, through a connector fixed to the skull and, in most studies, a cable to the recording system.

It made the field. The BrainGate trials, which began in 2004, used it to let people with tetraplegia move cursors, type and control robotic arms, and the arrays have been studied in dozens of people for a total of more than 30,000 implant days, some for over eight years.

Its limits are the needles. Rigid silicon in soft, moving tissue provokes a foreign-body response, and recordings slowly lose neurons over months to years; a percutaneous connector also carries a risk of infection.

It is cleared for short use. Its clinical version has US clearance for recording for 30 days or less; permanent use remains investigational.

The Utah array is the reference every new implant is measured against.

Newer devices compete on channel count, flexibility and wirelessness, while the array keeps two decades of human data.

Questions: How much power can an implant spend before it warms the brain by a degree? Measurements and models of a powered Utah array found that the surrounding tissue warms by about 0.029 °C for every milliwatt the device dissipates. With designers keeping the rise within about 1 °C, that leaves roughly 35 mW for everything inside the head, from amplifiers to the radio. The figure depends on the device's size and on blood flow around it, but it sets the order of magnitude: tens of milliwatts, which is why every added channel has to be cheaper in power than the last. What happens to a person's brain implant when the study that placed it ends? It depends on what was agreed, and for many early trials too little was. Utah arrays have stayed implanted in some participants for more than eight years, under research protocols whose funding, staff and equipment have their own lifespans, and a working implant may be removed, left in place unused, or kept running only while a lab supports it. Reviews of neural device research ask sponsors to plan before implantation for continued access to a device that helps, for maintenance and for removal, so that the end of a study does not decide alone what happens to a part of someone's body. Why has one small array from Utah carried most human brain-computer interface research? For most of two decades it was the only high-channel microelectrode array with regulatory clearance for recording single neurons in people (its clinical version is cleared for short-term use), and Blackrock Neurotech has supplied it to research groups since the BrainGate trials began in 2004. Its 4 by 4 mm grid of 96 to 128 needles reads the spikes that fine decoding needs, and its arrays have been studied in dozens of people for more than 30,000 implant days in total. Most landmark results, from cursor and robotic arm control to the speech decoders of 2023, were obtained with it, which is why newer devices are compared against it.