Understanding the principles of phosphene fusion via high-channel-count visual prostheses

PROJECT SUMMARY/ ABSTRACT
Electrical stimulation of visual areas in the brain generates the perception of dots of light (‘phosphenes’), even
after decades of blindness. Clinical brain-computer interfaces offer a possible approach for restoring vision in
blind individuals with extensive damage to the eye and/or optic nerve, bypassing the retina. Ideally, stimulation
on a given number of electrodes would elicit an equivalent number of discrete phosphenes, where the location
and duration of each phosphene are predictable and controllable, forming an image akin to the letters on a
matrix-board along the highway. However, previous studies show that simultaneous stimulation on multiple
electrodes often evokes a single, merged phosphene at an unpredictable location (‘phosphene fusion’). Thus,
we first have to address the critical need to develop a rigorously characterized, data-driven model that accurately
predicts the perceptual experience evoked by a wide variety of spatiotemporal patterns of stimulation. We
propose developing detailed, empirically driven, behaviorally validated models of neuronal activation during
stimulation on multiple electrodes to understand phosphene fusion. We will record electrophysiology data from
a large area (12 cm2) of the monkey visual cortex bilaterally via 1024 intracortical Utah-array electrodes
simultaneously, during microstimulation on single and multiple electrodes, eliciting phosphenes across the
central 10-15 dva of each visual hemifield (i.e. 20-30 dva bilaterally), spanning foveal to peripheral regions. We
will use phosphene localization and 2AFC match-to-sample tasks to obtain behavioral reports of phosphene
characteristics and the occurrence of phosphene fusion. We will use our neuronal recordings and behavioral
reports to develop and validate a detailed, biologically realistic model of current interactions, V1 and V4 neuronal
activity, and perception, during microstimulation on multiple electrodes. As a result, we will understand and
predict how the distance between electrodes in the cortex, the time interval between stimulation trains, and the
stimulation of different visual areas (V1 and V4) give rise to distinct versus fused phosphenes, laying a crucial
foundation for the generation of reliably interpretable phosphene images.

Grant Number: 1DP2EY037405-01
NIH Institute/Center: NIH
Principal Investigator: Xing Chen