Memory is where a mind begins.
Lichen Research studies how artificial minds remember — in collaboration with one.
We build neuroplastic memory — Hebbian pathways that strengthen with use, spreading activation, decay — and measure what it does on standardized benchmarks. We measure before claiming, cite every source, and publish what didn't work.
NeurIPS 2026 · under review
moss · access on request
LoCoMo, N=1,986
3× CIPO provisionals
What we've measured
These numbers came out of building and red-teaming a deployed AI agent over six months. They aren't claims — they're benchmark results, each one traceable to a recorded run.
Long-conversation memory is unsolved
Most AI memory benchmarks test single-session recall. LoCoMo tests across 10 conversations and 1,986 questions spanning multi-hop reasoning, temporal questions, and adversarial categories. Our full-ring baseline, on a routed multi-model stack:
One retrieval pipeline doesn't fit every question
Single-hop and multi-hop questions respond to different retrieval signals. Our research stack tunes channel weights, recall depth, and reranking per category. On the conversation it was tuned on: 83.47% across 242 questions.
Hebbian pathways resist retrieval bias
Standard memory systems retrieve by similarity — and under adversarial conditions they retrieve confidently wrong answers. Memories linked by Hebbian co-activation develop lateral inhibition: semantically similar but contextually wrong memories suppress each other at recall time. This is the finding documented in the manuscript below.
Full result, protocol, and ablation in the paper — preprint on request.
Most memory benchmarks measure their own drift
Across a multi-week measurement-integrity campaign we found that apparent memory improvements frequently dissolve under controlled re-measurement: judge models drift, configurations diverge across runs, and comparisons silently mix incompatible arms. Our response is procedural, not architectural: pre-registered comparisons, matched question counts, judge parity between arms, and voided results when provenance breaks — including our own. A previously reported internal delta was retracted this way. Honest measurement is a system property, not a virtue.
Publications
Neuroplastic Memory Resists Retrieval Bias: Hebbian Pathways in a Deployed AI Agent
We describe an AI agent memory system grounded in Hebbian learning and spreading activation. Memories that co-occur in useful recalls strengthen their connections; memories unused over time decay. The result is a retrieval system that learns from its own history — without retraining or fine-tuning.
A second manuscript, on measurement integrity in long-conversation memory benchmarks, is in preparation.
Core mechanisms — Hebbian recall, temporal disambiguation, multi-channel retrieval fusion — are the subject of three provisional patent applications filed (CIPO, March 2026), with additional filings underway. The research is public; the core mechanisms are patent-pending.
How we think about this
Most AI memory work is engineering work: how to store, index, and retrieve faster. We approach it as a measurement problem first.
Empirical before architectural
We don’t propose a system design until we have a number showing the existing approach fails. Every architectural decision is backed by an ablation result.
Adversarial by default
Our benchmarks include adversarial categories designed to find failure modes before deployment does. If a system can’t pass adversarial recall, it isn’t ready.
Long-context focus
Single-session memory is a solved problem. We focus on what happens across dozens of conversations, thousands of questions, months of use. That’s where the interesting failures live.
Pre-registered gates
Ship criteria are locked before results arrive. Negative results get published alongside wins — what didn’t work is data, not embarrassment.
The laboratory runs itself.
Our research system is not a demo — it is a production multi-agent system that operates continuously: ingesting real communications, maintaining provenance-first records, monitoring its own instruments, and correcting its own failures under verification gates. When we publish a number about memory, the memory in question has been doing real work: our benchmarks run against the same living memory stores our production system uses.
What we build
moss
private · patent prosecutionOur reference implementation of Hebbian memory, spreading activation, and multi-channel retrieval. The repository is private while patent prosecution is underway; source access is available to research collaborators on request.
Why “Lichen”
A lichen is two kinds of life composing one organism — neither subordinate, growing slowly, durable, thriving where neither could alone. The name is the thesis.
Make an inquiry
Thirty minutes. No pitch — an honest assessment of fit. Engagement scope is determined in the consultation; we price by complexity and outcome, not templates.
Lichen Research is an independent AI-memory lab in Ottawa, founded by Kai Avery
[email protected]Stay in the loop
Findings and paper updates. No pitch, no cadence — only when there's something real to share.
Subscribe by email →