Trees remember without a brain. So can a memory system.

A spruce that grew from a cold seed buds earlier, decades later. The DNA is the same; how it is read is not. What that taught us about our own memory, including the bug where we got it backwards.

1 Nov 2026 · 2 min read

Trees have no brain and still remember. Not as a metaphor: as a measurable change in how they respond to the next season.

What a tree keeps

The same DNA, read differently. Two Norway spruce trees with identical DNA can behave differently for years. The one whose seed developed at 18 °C starts growing much earlier in spring than its twin from a seed that developed at 28 °C 1.

The mechanism is epigenetic: marks on the DNA that control how tightly it is packed, “turned off” or “turned on”. The instructions are not rewritten. Which of them get used changes 1 2.

What triggers such memories in conifers:

signalwhat the tree remembers
temperature during seed developmentwhen to start growing in spring
radiation (the trees near Chernobyl)growth form: bushes instead of trees at high doses
defence chemicals applied in autumna stronger defence against insects the next spring

All from the same review of conifer epigenetics 1.

And the scars of a population. Plants that lived through human-caused population crashes still carry the trace generations later, as reduced genetic diversity 3. Memory at the level of the species.

Why this is fast

Natural selection rewrites the instructions. In long-lived trees that takes generations and hundreds of years. Epigenetic memory adjusts within one lifetime, without touching the code 1.

That is exactly the property a working memory system needs: adapt to what happened this week without retraining anything.

What we built from it

Our organic memory, phyto, keeps about 7,000 facts. The facts do not change when they are used. What changes is a score on top of them, an epigenetic score: how often a fact was reinforced, how recently, under what stress.

treephyto
DNA, unchangedthe fact itself, unchanged
epigenetic marksthe epigenetic score
priming by stresshormone-like signals that raise or lower facts
what is expressed this seasonthe aperture: what opens for this situation

Where we got it backwards

Our first real test found a bug that the tree would never make. The epigenetic score was added to the situation’s relevance instead of gating it. A heavily reinforced fact won every query, whatever was asked: five sticky facts held 38% of every answer.

A tree’s memory does not fire in every situation. The autumn defence primes the response to insects. It does not make the tree defend itself against the spring. Memory should change how a signal is answered, not replace the signal. The fix is to gate.

How far the analogy goes

Plants are computers in a narrower, literal sense too: a living plant’s physiology has been used as a physical reservoir for computation 4. We do not claim our software behaves like a tree. The design rule we took is narrower and testable:

Keep the facts fixed. Let use change how they are read. Never let use replace the question.

References
  1. P. Krokene, M. H. Mageroy. Tree Memories: How Can Trees Remember Without a Brain?. Frontiers for Young Minds 12:1400253 (2024)saved pagecorpus: ~/phyto
  2. C. G. Fossdal, P. Krokene, J. E. Olsen, et al.. Epigenetic stress memory in gymnosperms. Plant Physiology 195, 1117–1133 (2024)not in corpus
  3. McGill University Newsroom. Plants retain a 'genetic memory' of past population crashes, study shows (2026)saved pagecorpus: ~/phyto
  4. O. Pieters, T. De Swaef, M. Stock, F. wyffels. Leveraging plant physiological dynamics using physical reservoir computing. Scientific Reports 12, 12594 (2022)originalcorpus: ~/phyto
Our records

The measurements in this article come from these records in the running system.

  • ~/memory/2026-09-29-MEASURED-phyto-vs-crystal-first-real-ab.md — the epigenetic score that was added instead of gated
Read next