iiinova — where your digital self keeps evolving

Bionic Vision Lab @bionicvisionlab.org · 08/08/2026

Finally, the neural response mattered for perception. Recorded population activity predicted whether a phosphene was seen, and its brightness and color, substantially better than stimulation parameters alone. The emerging picture: stimulation → neural response → perception #BCI #Neuroscience

Phosphene detection, brightness, and color decoding: Bar plots compare how well different inputs predict the participant’s perceptual reports. Neural-response features improve prediction over stimulation parameters alone, with the strongest performance generally obtained when stimulation and neural activity are combined.

110

Bionic Vision Lab @bionicvisionlab.org · 08/08/2026

But not every neural response is equally achievable... Evoked activity occupied a low-dimensional neural manifold. Targets farther from that manifold were much harder to reproduce (r = 0.85). The optimized methods also achieved their targets using lower stimulation currents.

Current amplitude and neural-manifold constraint: Violin and scatter plots compare stimulation methods and target responses. Deep-learning approaches use lower mean currents, while synthetic targets lie farther from the natural neural-response manifold. Reconstruction error increases strongly with distance from that manifold (r = 0.85).

110

Bionic Vision Lab @bionicvisionlab.org · 08/08/2026

Prediction is useful. Control is the harder problem. Given a desired neural response, we used either gradient optimization or an inverse neural network to find a stimulation pattern predicted to produce it, then tested those patterns in vivo. Both beat conventional approaches.

In-vivo neural activity shaping: Example target response compared with stimulation patterns and recorded cortical responses produced by linear, dictionary, 1-to-1, inverse neural network, gradient-optimization, and replay methods. Bar plots below show that the inverse neural network and especially gradient optimization reproduce the target response more accurately than conventional baselines.

100

Bionic Vision Lab @bionicvisionlab.org · 08/08/2026

So we trained a deep forward model to predict the population response to multielectrode stimulation. On held-out days, it outperformed 1-to-1, linear, nonlinear, and dictionary-based models across multiple measures of prediction accuracy.

Predicted versus recorded neural responses: Five examples show multielectrode stimulation patterns, the neural responses predicted by the deep forward model, and the corresponding responses recorded in vivo. Across examples, the predicted spatial patterns closely resemble the measured cortical responses.

Forward-model performance: Three bar plots compare models for predicting stimulation-evoked neural activity using mean squared error, R², and Earth mover’s distance. The deep forward neural network performs best across all three measures and significantly outperforms the tested baseline models.

110

Bionic Vision Lab @bionicvisionlab.org · 08/08/2026

First problem: the same electrical stimulation does not always produce the same neural response. Responses were much more variable across days than within a day, motivating a model that accounts for the brain’s ongoing state rather than treating stimulation as a fixed input-output mapping.

Same electrical stimulation pattern delivered across three days produces noticeably different cortical activity. Heatmaps show pre-stimulation activity and two example evoked responses for each day across the electrode array. The plot at right quantifies this effect: neural responses vary significantly more across days than within the same day.

100

Bionic Vision Lab @bionicvisionlab.org · 08/08/2026

New in Neuron 👁️🧠🧪: Can AI help a cortical visual prosthesis control what the brain actually does after electrical stimulation? We used a bidirectional implant to model and shape stimulation-evoked activity in human visual cortex. doi.org/10.1016/j.ne... #UCSB #AI #BionicVision #Neurotechnology

Schematic comparing three methods for producing a target neural response with a cortical electrode array: direct 1-to-1 mapping, a learned inverse model, and gradient-based optimization. Each method generates a stimulation pattern that is passed through either a simulated forward model or tested in vivo. Resulting neural responses are shown as electrode-array heatmaps, illustrating how closely each approach reproduces the target pattern.

2205

Bionic Vision Lab @bionicvisionlab.org · 05/05/2026

At #ARVO2026? Come find Poster #0523 • 10:15am - 12:00pm MT • May 6th We asked 100+ legally #blind adults: What shapes your attitude toward #BionicVision? • usefulness helped • barriers hurt • ease of use mattered, but less • age & vision level were not predictive eppro02.ativ.me/web/page.php...

Flyer advertising the poster. Text says: Catrina Coe. Behavioral Determinants of Bionic Vision Adoption in a U.S.-Based Cohort of Blind Adults Poster #0523 • 10:15am - 12:00pm MT • May 6th

030

Bionic Vision Lab @bionicvisionlab.org · 03/05/2026

At #ARVO2026? Come find Poster #0978 • 8:30 - 10:15am MT • May 5th Argus II timing breaks sighted flicker-fusion assumptions: • temporal effect: fragmented percepts at 0 ms, cleaner 1-vs-2 percepts at 4+ ms • spatial effect: axon-map distance predicts separability eppro02.ativ.me/web/page.php...

Flyer advertising the two poster. Text says: Lily Turkstra. Perceptual Integration Window in Argus II: Millisecond Timing and Axon-Bundle Geometry Determine Two-Point Perception. Poster #0978 • 8:30am - 10:15am MT • May 5th.
2. Catrina Coe. Behavioral Determinants of Bionic Vision Adoption in a U.S.-Based Cohort of Blind Adults Poster #0523 • 10:15am - 12:00pm MT • May 6th

001

Bionic Vision Lab @bionicvisionlab.org · 01/05/2026

Two #BionicVision studies 👁️🧠 from our lab at #ARVO2026: 1. Argus II timing breaks sighted flicker-fusion assumptions: May 5, 8:30–10:15 AM 2. What shapes blind adults’ attitudes toward vision-restoration technology: May 6, 10:15 AM–12:00 PM Come find us in Denver!

Flyer advertising the two accepted posters. Text says: UC SANTA BARBARA
BIONIC VISION LAB @ ARVO 2026. Followed by
1. Lily Turkstra. Perceptual Integration Window in Argus II: Millisecond Timing and Axon-Bundle Geometry Determine Two-Point Perception. Poster #0978 • 8:30am - 10:15am MT • May 5th.
2. Catrina Coe. Behavioral Determinants of Bionic Vision Adoption in a U.S.-Based Cohort of Blind Adults Poster #0523 • 10:15am - 12:00pm MT • May 6th

153

Bionic Vision Lab @bionicvisionlab.org · 05/03/2026

Our approach borrows an idea from software verification: coverage-guided fuzzing. We systematically mutate inputs and search for stimulation patterns that violate biophysical constraints - uncovering diverse safety violations that conventional testing misses. #Neurotech #MLResearch #AIVerification

Bar plot comparing fuzzing strategies. y-axis: combined violation and diversity score. Each bar represents an average of normalized violations and normalized diversity score, equally weighted.
Our metrics are shown in green, with our two best VO-KMVP and VO-KMOC highlighted in dark green. Neuron coverage metrics are shown in purple, and basic metrics in red. Conventional testing (model test set with no mutations) is shown in blue. Our methods reach scores above 0.8, whereas conventional testing sits at 0.1

100

Bionic Vision Lab @bionicvisionlab.org · 16/11/2025

Our final poster at #SfN2025 explores human-in-the-loop optimization for intracortical microstimulation, presented by @lozaneuro.bsky.social, in collaboration with @umh.es: PSTR450.20 Nov 19 at 1:00 PM www.abstractsonline.com/pp8/#!/21171... #SfN25 #VisionScience #NeuroTechnology

Schematic labeled human-in-the-loop optimization (HILO). It shows two stimuli on the left: pulse trains with varying stimulus amplitude over time. A participant has to choose which stimulus appears brighter. This feedback is used to inform a Gaussian process model that chooses the next stimulus pair, with the goal of finding the stimulus with the lowest overall charge to elicit perception

151

Bionic Vision Lab @bionicvisionlab.org · 16/11/2025

Lily Turkstra is presenting new findings on stimulus-selective spiking activity recorded during a working memory experiment in a unique intracortical dataset. PSTR341.13 Nov 18 at 1:00 PM www.abstractsonline.com/pp8/#!/21171... #SfN25 #VisionScience #Neuroscience

left: experimental setup showing an implantee with an introcortical prosthesis and example phosphenes described as a large filled circle, a half-moon, and a tiny dot.
right: schematic showing cross-temporal decoding of delay period activity. Over 406 trials, working memory content could be decoded during the delay period in 88.5% of delay period windows.

110

Bionic Vision Lab @bionicvisionlab.org · 16/11/2025

Our second poster at #SfN2025 dives into biologically plausible networks for efficient image encoding, presented by Hasith Basnayake: PSTR154.08 Nov 17 at 8:00 AM www.abstractsonline.com/pp8/#!/21171... #SfN25 #VisionScience #NeuroTechnology

Table showing different network diagrams under test, trained either on MNIST or Fashion-MNIST, with either soft or hard winner-take-all (WTA) wiring. Synaptic weights showed either holistic or parts-based representations of images

100

Bionic Vision Lab @bionicvisionlab.org · 16/11/2025

If you are into #VisionScience or #neuroengineering, come check out our first poster at #SfN2025 this afternoon! Emily Joyce is presenting new work on modeling the bipolar circuitry in the human fovea PSTR122.22 Nov 16 at 1:00 PM www.abstractsonline.com/pp8/#!/21171...

Simulated responses of a bipolar cell mosaic to simulated electrical stimulation and the corresponding decoded phosphenes. Small phosphenes appear focal and colored, whereas larger phosphenes most often appear with a white-ish, yellow-ish tint.

142

Bionic Vision Lab @bionicvisionlab.org · 09/07/2025

✅ Checkerboard consistently outperformed the other patterns—higher accuracy, lower difficulty, fewer motion artifacts. 💡 Why? More spatial separation between activations = less perceptual interference. It even matched performance of the ideal “no raster” condition, without breaking safety rules.

Boxplots showing task accuracy for two experimental tasks—Letter Recognition and Motion Discrimination—grouped by five raster patterns: No Raster (blue), Checkerboard (orange), Vertical (green), Horizontal (brown), and Random (pink). Each colored boxplot shows the median, interquartile range, and individual participant data points.

In both tasks, Checkerboard and No Raster yield the highest median accuracy.

Horizontal and Random patterns perform the worst, with more variability and lower scores.

Significant pairwise differences (p < .05) are indicated by horizontal bars above the plots, showing that Checkerboard significantly outperforms Random and Horizontal in both tasks.

A dashed line at 0.125 marks chance-level performance (1 out of 8).

These results suggest Checkerboard rastering improves perceptual performance compared to conventional or unstructured patterns.

110

Bionic Vision Lab @bionicvisionlab.org · 09/07/2025

We ran a simulated prosthetic vision study in immersive VR using gaze-contingent, psychophysically grounded models of epiretinal implants. 🧪 Powered by BionicVisionXR. 📐 Modeled 100-electrode Argus-like array. 👀 Realistic phosphene appearance, eye/head tracking.

Diagram showing the four-step pipeline for simulating prosthetic vision in VR.
Step 1: A virtual camera captures the user’s view, guided by eye gaze. The image is converted to grayscale and blurred for preprocessing.
Step 2: The preprocessed image is mapped onto a simulated retinal implant with 100 electrodes. Electrodes are activated based on local image intensity and grouped into raster groups. Raster Group 1 is highlighted.
Step 3: Simulated perception is shown with and without rastering. Without rastering (top), all electrodes are active, producing a more complete but unrealistic percept. With rastering (bottom), only 20 electrodes are active per frame, resulting in a temporally fragmented percept. Phosphene shape depends on parameters for spatial spread (ρ) and elongation (λ).
Step 4: The rendered percept is updated with temporal effects and presented through a virtual reality headset.

100

Bionic Vision Lab @bionicvisionlab.org · 09/07/2025

Checkerboard rastering has been used in #BCI and #NeuroTech applications, often based on intuition. But is it actually better, or just tradition? No one had rigorously tested how these patterns impact perception in visual prostheses. So we did.

Raster pattern configurations used in the study, shown as 10×10 electrode grids labeled with numbers 1 through 5, representing five sequentially activated timing groups.

1. Horizontal: Each row of electrodes belongs to one group, with activation proceeding top to bottom.

2. Vertical: Each column is a group, activated left to right.

3. Checkerboard: Electrode groups are arranged to maximize spatial separation, forming a checkerboard-like layout.

4. Random: Group assignments are randomly distributed across the grid, with no spatial structure. This pattern was re-randomized every five frames to test unstructured activation.
Each group is represented with different shades of gray and labeled numerically to indicate activation order.

100

Bionic Vision Lab @bionicvisionlab.org · 09/07/2025

👁️🧠 New paper alert! We show that checkerboard-style electrode activation improves perceptual clarity in simulated prosthetic vision—outperforming other patterns in both letter and motion tasks. Less bias, more function, same safety. 🔗 doi.org/10.1088/1741... #BionicVision #NeuroTech

Raster patterns in simulated prosthetic vision. On the left, a natural scene of a yellow car is shown, followed by its transformation into a prosthetic vision simulation using a 10×10 grid of electrodes (red dots). Below this, a zoomed-in example shows the resulting phosphene pattern. To comply with safety constraints, electrodes are divided into five spatial groups activated sequentially across ~220 milliseconds. Each row represents a different raster pattern: vertical (columns activated left to right), horizontal (rows top to bottom), checkerboard (spatially maximized separation), and random (reshuffled every five frames). For each pattern, five panels show how the scene is progressively built across the five raster groups. Vertical and horizontal patterns show strong directional streaking. Checkerboard shows more uniform activation and perceptual clarity. Random appears spatially noisy and inconsistent.

120

Bionic Vision Lab @bionicvisionlab.org · 16/07/2024

Yuchen Hou presenting at EMBC '24 in Orlando, FL (@IEEEembs @IEEEorg): We introduce two models that can predict phosphene fading & persistence under varying stimulus conditions, cross-validated on behavioral data reported by 9 Argus II users. Preprint:...

100

Bionic Vision Lab @bionicvisionlab.org · 25/09/2023

Our students Tori LeVier and Lily Turkstra had the distinct pleasure of interviewing Dr. Philip Hessburg (@DIO_EyeOnDesign), co-founder and Director Emeritus of the Detroit Institute of Ophthalmology at...

100

Bionic Vision Lab @bionicvisionlab.org · 31/08/2023

Find out what first got Dr. Rathbun into bionic vision and how his work with Dr. Eberhart Zrenner at the University of Tübingen (@uni_tue) inspired him to start the Bionics and Vision Lab at @HenryFordHealth!

000

Bionic Vision Lab @bionicvisionlab.org · 31/08/2023

Our students Lily Turkstra and Tori LeVier recently sat down with Dr. @DanielRathbun from @HenryFordHealth to talk about recent advancements, challenges, and future prospects of #BionicEye technologies. Full...

100

Bionic Vision Lab @bionicvisionlab.org · 19/09/2022

Come by Poster M094 tonight at #MICCAI2022 to chat about retinal implants! Our work was also highlighted by the MICCAI Daily magazine: www.rsipvision.com/MICCAI2022-Monda…

000

Bionic Vision Lab @bionicvisionlab.org · 29/04/2022

Congratulations🎉to our graduating @UCSBpsych senior Yuchen Hou for winning 2(!) PBS awards: - PBS Exceptional Academic Performance Award - Abdullah & Marjorie R. Nasser Memorial Scholarship Fund Award which recognize her academic excellence and research aptitude. Well deserved!

000

Bionic Vision Lab @bionicvisionlab.org · 25/03/2022

Into #visionscience & looking to get some paid research experience during your pre-#PhD gap year? Come work with us & study visual perception in retinal implant users: recruit.ap.ucsb.edu/JPF02194 #AcademicJobs @psychologyjobs @PsychChatter @VisionScience

000

Bionic Vision Lab @bionicvisionlab.org · 22/03/2022

As a proof of concept, we show how percept shape *might be* improved for the implant settings of three existing Argus II users. Results are expected to be highly patient-specific, mainly depending on: - intraocular implant location - behaviorally reported phosphene distortions

100

Bionic Vision Lab @bionicvisionlab.org · 22/03/2022

Similar to van Steveninck et al. (2021), we trained a deep net to predict the optimal stimulus to produce a desired percept. 2 important changes: - ✅psychophysically validated phosphene model - ❌no *de*coder (which could unlearn any deficiencies of the *en*coder)

100

Bionic Vision Lab @bionicvisionlab.org · 22/03/2022

We found that: - using an #HMD may⬆️performance - increased phosphene elongation may⬇️performance - more electrodes do not necessarily⬆️performance Our results highlight the importance of using an appropriate phosphene model when predicting visual outcomes for bionic vision.

100

Bionic Vision Lab @bionicvisionlab.org · 22/03/2022

New work by Justin Kasowski lets you "see through the eyes" of a #BionicEye user: Immersive virtual reality simulations of bionic vision (Augmented Humans '22 @aug_humans) Preprint: arxiv.org/abs/2203.05675 Code:...

100

Bionic Vision Lab @bionicvisionlab.org · 23/02/2022

She has also reported on the importance of field of view (FoV) in visual prosthesis design (doi.org/10.1088/1741-2552/abb9be) and prototyped an #AugmentedReality system for navigation (arxiv.org/abs/2109.14957). We can't wait to tell you what's next!

000

Bionic Vision Lab @bionicvisionlab.org · 23/02/2022

You may have heard about her work on combining structural and semantic segmentation to improve indoor scene understanding with simulated prosthetic vision: doi.org/10.1371/journal.pone.0227677

100

Bionic Vision Lab @bionicvisionlab.org · 23/02/2022

We are excited to welcome the lab's first #postdoc: Melani Sanchez Garcia, who graduated with a PhD in Computer Science and Systems Engineering (cum laude) @unizar. She'll be working on new ways to apply #ComputerVision/#MachineLearning to the...

100

Bionic Vision Lab @bionicvisionlab.org · 06/11/2021

Congrats to Tanya Bhatia, Honors Student in @UCSBpsych, for winning a @sacnas Student Presentation Award at...

100

Bionic Vision Lab @bionicvisionlab.org · 05/10/2021

Last but not least is John Exton (@UniofNewcastle) who is working towards an optogenetic cortical visual prosthesis. And with this, the technical session at #TEATC2021 comes to a close. What a whirlwind it's been! See you in 2 years (hopefully in person)

000

Bionic Vision Lab @bionicvisionlab.org · 05/10/2021

Seung Woo Lee (@MGHMedicine) continues Dr. Fried's presentation by detailing how micro-magnetic stimulation of V1 can lead to focal activation of downstream neurons in the human visual system. #TEATC2021

100

Bionic Vision Lab @bionicvisionlab.org · 05/10/2021

The last session of #TEATC2021 focuses on cortical alternatives to electrical stimulation. First up is Shelley Fried (@MGHMedicine) updating us on his work towards a micro-coil based cortical visual prosthesis

100

Bionic Vision Lab @bionicvisionlab.org · 05/10/2021

Zixen Ye (City University of Hong Kong) summarizes his work on light-intensity-controlled stimulation of neurons by an organic photovoltaic interface. #TEATC2021

100

Bionic Vision Lab @bionicvisionlab.org · 05/10/2021

Fabrizio Grani (CORTIVIS, Spain) demonstrates a closed-loop approach for automatically adjusting thresholds in cortical visual prostheses, taking into account brain state #TEATC2021

100

Bionic Vision Lab @bionicvisionlab.org · 05/10/2021

Avi Caspi (Second Sight Medical Products) demonstrates the utility of gaze-contingent stimulation for visual prostheses. #TEATC2021

100

Bionic Vision Lab @bionicvisionlab.org · 05/10/2021

Shashi Srivastava (@kocuniversity) is working towards an organic photovoltaic device for optical stimulation. #TEATC2021

100

Bionic Vision Lab @bionicvisionlab.org · 05/10/2021

Ieva Vebraite (@TelAvivUni) introduces semiconducting organic pigments for photo stimulation of the retina. #TEATC2021

100

Bionic Vision Lab @bionicvisionlab.org · 05/10/2021

Day 3 of #TEATC2021 starts with a session on new materials for bionic vision. Doug Shire (Bionic Eye Technologies, Ithaca, NY) talks about his work on manufacturing subretinal electrodes for the Boston Retinal Prosthesis project.

100

Bionic Vision Lab @bionicvisionlab.org · 05/10/2021

Martin Spencer (@BVTeyes) presents his linear-nonlinear model of neural activation, which can shape neural activity in the retina, and demonstrates how this model can be used to optimize stimulation strategies for retinal implants. #TEATC2021

100

Bionic Vision Lab @bionicvisionlab.org · 05/10/2021

Madhuvanthi Muralidharan (@UNSW) demonstrates the effects of high-frequency stimulation on RGC type-selective stimulation #TEATC2021

100

Bionic Vision Lab @bionicvisionlab.org · 05/10/2021

Nick Barnes (@EyeandEarHosp) presents different image processing methods tested with the suprachoroidal implant (@bionicvision, @BVTeyes), including a depth saliency algorithm that can highlight objects that stick out of their surroundings #TEATC2021

100

Bionic Vision Lab @bionicvisionlab.org · 04/10/2021

Michael Beyeler (@bionicvisionlab) talks about the re-emerging role of computer vision and AI in artificial vision, describing a path towards a more useful bionic eye. #TEATC2021

100

Bionic Vision Lab @bionicvisionlab.org · 04/10/2021

Gislin Dagnelie (@JohnsHopkins) presents on simulating activities of daily living under ultra low vision (ULV) using virtual reality, discovering a wide range of achieved performances across subjects even with similar levels of degradation, similar to real ULV subjects #TEATC2021

100

Bionic Vision Lab @bionicvisionlab.org · 04/10/2021

Justin Kasowski (@bionicvisionlab) discusses the role of VR in simulating prosthetic vision, highlighting the importance of realistic simulations, immersion, and modeling axonal distortions. #TEATC2021

100

Bionic Vision Lab @bionicvisionlab.org · 04/10/2021

Charles Zhijie (@Stanford) expands on the advanced current steering technique used to achieve high visual acuity in rats implanted with PRIMA. Preconditioned neighboring electrodes are used to confine the electric field while maintaining stimulation depth. #TEATC2021

110

Bionic Vision Lab @bionicvisionlab.org · 04/10/2021

Daniel Palanker (@Stanford) presents on sight restoration in rats using PRIMA, achieving a restored visual acuity near the natural limit #TEATC2021

100