Oxford researcher advancing non-invasive MEG speech decoding, brain foundation models, and open speech-BCI evaluation.
4
papers here
4
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
18
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Not yet scored: the underlying data is insufficient for a reliable output composite. This axis would measure publication volume and productivity over time. Raw counters exist — 21 works in the last three years, h-index 18 — but no score is derived.
Impact—
Not yet scored: citation-based impact data is insufficient in this snapshot. This axis would measure citation influence, field-normalised citation rates, and the uptake of benchmarks and metrics. No impact score should be inferred from the present record.
Focus0/100
The radar focus score is 0, grounded in an even OpenAlex field distribution: the four labelled fields each account for 0.25 — computational neuroscience / neural speech decoding, brain foundation models, machine-learning BCIs, and non-invasive speech decoding. Work spans methods, benchmarks, roadmap, and evaluation theory rather than one narrow line; topic counts (language neurobiology 25, phonetics/phonology 17, connectivity 15, EEG-BCI 12) reinforce this breadth.
Collaboration66/100
Collaboration scores 66, with 26 distinct co-authors recorded. Recent arXiv work is team-based: last-author roles on the semantic bottleneck, PNPL competition, and OVMI metric papers, and a middle authorship on the MEG foundation-model roadmap. Recurring co-authors include Dulhan Jayalath, Gilad D. Landau, Karim Jerbi, Philipp Thölke, and Catherine Duclos, suggesting an active network around MEG decoding and evaluation.
Activity90/100
Activity is the strongest dimension at 90. Works per year climbed from 1-2 in the late 2000s to 9 in 2019 and a peak of 11 in 2024, with 21 works in the last three years and sustained 5-7 outputs annually from 2019 through 2026. All four Evolve-reviewed papers appeared in 2026 alone.
Reproducibility69/100
Reproducibility scores 69, based on Evolve's review of 4 papers (reproMean 69). Readiness varies by output: the 2026 PNPL competition paper and the OVMI communication metric are rated high, the semantic bottleneck decoder medium — demonstrated on a single subject — and the MEG foundation-model roadmap low, being purely qualitative without new experiments.
Research summary
Oiwi Parker Jones is a researcher at the University of Oxford working at the interface of computational neuroscience and brain-computer interfaces, with a current centre of gravity in non-invasive speech decoding from MEG. The bibliometric snapshot (OpenAlex-resolved identity) records an h-index of 18, an i10-index of 27, 21 works in the last three years, and 26 distinct co-authors; the activity dimension scores 90, reflecting annual output that rose from 1-2 works in the late 2000s to peaks of 9 (2019) and 11 (2024). Topic counts show roots in the neurobiology of language and bilingualism (25) and phonetics and phonology research (17), now joined by functional brain connectivity (15) and EEG and brain-computer interfaces (12). All four papers reviewed by Evolve are 2026 arXiv outputs: a semantic-bottleneck approach to MEG-to-text decoding, a roadmap for MEG foundation models, the 2026 PNPL competition on LibriBrain100 word classification, and the OVMI communication measure for speech BCIs — three with last-author position. The field distribution splits evenly across four speech-BCI and computational-neuroscience labels (0.25 each), producing a focus score of 0 that signals deliberate breadth across methods, benchmarks, and evaluation theory rather than a single narrow line.
Collaboration pattern
Recent outputs are consistently team-based: Parker Jones is last author on three of the four reviewed 2026 papers and a middle author on the MEG foundation-model roadmap, within a network of 26 distinct co-authors. Recurring collaborators include Dulhan Jayalath, Gilad D. Landau, Hamza Abdelhedi, Benjamin Ballyk, Fouad Lbakali, and Vanessa Hadid, together with Karim Jerbi, Philipp Thölke, Catherine Duclos, Oumayma Gharbi, Annalisa Pascarella, and Yorguin Mantilla-Ramos.
Research keywords
Highlights
Semantic Bottleneck for Non-Invasive Speech Decoding
Roadmap for MEG Foundation Models
2026 PNPL Competition on LibriBrain100
OVMI: A Common Communication Measure for Speech BCIs
Timeline
- 2026Publishes 'A Common Measure of Communication for Speech Brain-Computer Interfaces', deriving OVMI, a coverage-weighted mutual-information metric for speech BCIs (last author).
- 2026Releases the 2026 PNPL competition on LibriBrain100, defining word-classification and efficient cross-subject generalisation tasks for MEG (last author).
- 2026Co-authors 'A Roadmap for MEG Foundation Models', a qualitative agenda for self-supervised, source-space MEG modelling (middle author).
- 2026Last-author preprint 'The Semantic Bottleneck' demonstrates MEG-to-text decoding via semantic embeddings, outperforming BrainECHO on sentence metrics.
Papers in the Evolve catalog
The Semantic Bottleneck: Leveraging Semantic Representations for Non-Invasive Speech Decoding
A Roadmap for MEG Foundation Models
The 2026 PNPL Competition: Word Classification and Efficient Cross-Subject Generalisation in LibriBrain100
A Common Measure of Communication for Speech Brain-Computer Interfaces
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometric figures (h-index 18, i10-index 27, per-year counts, topics, co-authors) come from an OpenAlex-resolved identity and may misattribute name variants.
- Output and impact radar dimensions are unscored; no citation-based impact estimate is available in this snapshot.
- All four reviewed papers are 2026 arXiv preprints; peer-review status and published versions are not reflected.
- The focus score of 0 reflects an even 0.25/0.25/0.25/0.25 split across four field labels and may penalise deliberately broad portfolios rather than signal lack of direction.