# SIG-MRG 2026-07-16 — Session Notes

Participants:
  - Jeremiasz Rzenno | CET
  - Kei 🗝 | UTC-4
  - Mark Elder
  - rtk947 | UTC+1
  - Sachin
  - Vansh Kumar UTC-8
  - Wip | UTC+1
  - Yatú | UTC−7
  - AH
  - ayseayse__
  - Aaron Z Lewis | UTC-4

---

## 📖 The Reading

An excerpt from a book by **André Leroi-Gourhan** (rendered phonetically in the transcript as "Leroy Gouran/Gourin/Gwan"), beginning one page into a chapter titled **"Freeing Up Memory"** (participants reference page ~232). The discussion centers on his three levels of memory (automatic, mechanical, and lucid) and the analogy between the nervous system and a machine. The session is part of a "memory research group."

## 🧭 Overview

The session opened with a check-in round in which participants shared personal examples of "mechanical" habits that are hard to teach verbally (chicken sexing, driving, touch typing, hanging laundry, organizing a tech bag). Participants then read the excerpt silently for ~30 minutes before discussing Leroi-Gourhan's levels of memory, the tension between individual freedom and its dependence on collective/societal memory, and how his machine-to-nervous-system analogy holds up against contemporary technology and enactivist/embodied-cognition frameworks. The conversation closed with planning for the next session.

## 💡 Key Points & Themes

- **Opening theme — tacit, hard-to-teach skills:**
  - **Mark:** Introduced "chicken sexing" — experts can sort chicks by sex but can't verbally explain how; learners pick it up by observation. Teaching sometimes happens through observing rather than talking.
  - **Sachin:** Driving on "automatic mode"; sports (e.g., a tennis stroke) go better when done mechanically rather than over-thought.
  - **Vansh:** Touch typing as near-direct translation of thought to screen — hard to build a curriculum for beyond "practice."
  - **Wip (Will):** Hanging laundry — better to describe the *purpose* (get it dry) and leave the method open, since there are many valid ways.
  - **Yatú:** A multi-compartment tech bag where item placement is trained into memory; notable that the "knowledge" includes remembering where everything is.

- **Kei's framing:** Beneath individual habits (how you store things, clean a sink) lies a deeper *societal/cultural memory* (e.g., cleaning to reduce germs). The text compares the human situation to different classes of animals.

- **Leroi-Gourhan vs. Zerzin (Mark's contribution):** Mark contrasted Leroi-Gourhan with anarcho-primitivist thinker **John Zerzan**, whom he read as a teenager.
  - Leroi-Gourhan: "freedom of behavior is attainable only at the level of symbols"; intelligence is the ability to project symbolic sequences → language as an instrument of liberation.
  - Zerzan: with language "we have already accommodated ourselves to a world of unfreedom"; symbolic mediation closes off direct experience.
  - Mark's synthesis: same argument, opposite conclusions. Zerzan sees inherited symbolic programs as limiting our possibilities; Leroi-Gourhan sees those same programs as *increasing* our scope of possibilities. Mark aligned himself closer to Leroi-Gourhan.

- **The sinusoidal curve of lucidity (Mark):** Troughs = mechanical series; peaks = adjustments to those series. Freedom resides in the peaks of lucidity, where one can consciously adjust otherwise automatic behavior.

- **Human vs. earthworm distinction (rtk947):** The distinguishing human trait is the ability to *conceptualize* and ask "let's do this instead," then re-externalize that choice back into culture/cultural memory (e.g., building an electric toothbrush that then reshapes behavior). He suspects there is more continuity between earthworm and human behavior than the text implies.

- **Kei's central discussion prompt:** Leroi-Gourhan's three levels — automatic memory (earthworm/nervous system as simple machine, born knowing what to do, can't "choose a script"), mechanical memory, and lucid consciousness. Kei questioned whether the machine analogy still holds now that machines adapt to feedback (e.g., his electric toothbrush that reports which mouth areas are under-brushed, forming a "synaptic" feedback loop). He invoked **Michael Levin's** work on electrically reprogramming planarians/worms to be reborn with multiple heads (heritable across generations) as evidence that some "liberation" may be available to simpler creatures.

- **The core paradox (Kei):** "The individual's possibilities for comparison and liberation rest upon a potential memory whose entire contents belong to society." Individual epistemic freedom is only achievable via the collective mass of human memory — which arguably makes the human situation not so different from worms, potentially undermining the liberation thesis (and possibly vindicating Zerzan).

- **Materialist reading of freedom (Kei):** Freedom/novelty/invention can only be generated from a wealth of representational/material possibilities. This fits Leroi-Gourhan's methodology as an archaeologist/paleontologist — inverting the usual approach by tracing how objects *pass through* social and genetic organisms over time rather than how they arise. The symbol derives meaning from the pattern in which it's embedded — "the figure emerges from the ground." Kei tied this to Leroi-Gourhan's influence on **Stiegler**.

- **Enactivism / embodied cognition (ayseayse__ and rtk947):**
  - **rtk947:** The enactivist frame usefully establishes continuity between simple and rational life forms, but enactivism *struggles to make room for symbolic cognition* — the transition from structural coupling / autopoiesis to language is the hard problem. Breaking out of the collective-endowment paradox may require positing a basic human *capacity for freedom* (the "instinctual middle level").
  - **ayseayse__:** From an enactivist/embodied stance (she is a yoga teacher and data scientist), the concept of **structural coupling** and **autopoiesis** matters: an organism must maintain a boundary and a sense of significance/relevance to enact affordances and be "freer." This tradition runs *against* symbolism. She connected this to her AI-interpretability research on **symbol grounding** — observing how mathematical symbols (e.g., "2 + 2 = 4") are processed during inference, holding past/present/future in a temporal pattern. The pattern provides the grounding for the symbol, not vice versa — significance/relevance (boundaries/patterns) create the conditions for grounding.

- **Machines as assemblages (Jeremiasz):** If we take the machine analogy broadly (not as a reducible thing), it resembles **Levi Bryant's onticology/"ontocartography"** — machines as large assemblages whose interactions produce meaning and essence. He also invoked (tentatively **Stiegler** or another thinker) collections of patterns that stabilize within different temporalities into "gravitational" fields, with meaning always changing and patterns rearranging under external conditions.

## 🔀 Questions & Disagreements

- **Does symbolic substrate matter? (Sachin):** If "lived memory" (vs. static memory, from Kei's earlier "artificial memory" paper) can be continuously transmitted and changed in animals as well as humans, does that negate the need for a symbolic substrate? He was unsure and wanted to reread.
- **How much continuity is there between earthworm and human? (rtk947):** He thinks the text may understate the continuity.
- **Does the machine analogy still hold? (Kei):** Given adaptive/feedback-driven machines and Levin's reprogrammable worms, is the simple-machine-to-earthworm mapping still valid?
- **The liberation paradox (Kei):** If individual freedom depends entirely on socially-owned memory, is Leroi-Gourhan's liberation thesis undermined (and Zerzan partly vindicated)?
- **Enactivism vs. symbolism (ayseayse__, rtk947):** Whether an enactivist/autopoietic account (grounded in boundaries and structural coupling, and against symbolism) can accommodate symbolic cognition remains an open tension.
- **Zerzan vs. Leroi-Gourhan (Mark):** A genuine competing-interpretations framing — language as unfreedom vs. language as liberation. Mark noted his own reading of the sinusoidal-curve section was tentative ("I have no idea how accurate it is").

## 🔗 References Mentioned

- **André Leroi-Gourhan** — the author of the reading; chapter "Freeing Up Memory."
- **John Zerzan** — anarcho-primitivist thinker (contrasted with Leroi-Gourhan by Mark).
- **Michael Levin** — work on electrically reprogramming worms/planarians with heritable morphological changes.
- **Bernard Stiegler** — cited as influenced by Leroi-Gourhan (Jeremiasz and Kei).
- **Levi Bryant** — "ontocryptography"/onticology; machines as assemblages (Jeremiasz).
- **Gilbert Simondon** ("Simandan") — suggested as a possible future reading.
- **Kei's own "artificial memory" paper** — referenced by Sachin (lived vs. static memory).
- Concepts: enactivism, autopoiesis, structural coupling, affordances, symbol grounding, embodied cognition, interpretability in AI.

## ✅ Action Items & Next Time

- **Next session options (Kei):** Three branching paths — (1) the next chapter of Leroi-Gourhan on the *extensions of memory* (more straightforward, allows revisiting today's themes); (2) **Simondon** ("Simandan"); or (3) an embodied-cognition reading, with **ayseayse__** invited to suggest a text (Kei will locate specific excerpts).
- rtk947 expressed interest in reading the next chapter.
- The group appeared to settle on a direction (Kei: "We can do that then") and plans to revisit this chapter within the next session for continuity. Simondan kept on the list for later this year.
- The exact chosen reading is **not fully clear from the transcript** due to overlapping/garbled audio during the temp-check.

## ⭐ Memorable Quotes

- **Mark (paraphrasing Zerzan):** "With language we have already accommodated ourselves to a world of unfreedom."
- **Kei (quoting the text):** "The individual's possibilities for comparison and liberation rest upon a potential memory whose entire contents belong to society."
- **Mark:** "Teaching things isn't by talking about them, but by simply observing." (on chicken sexing)

*Note: Much of the transcript's middle portion has heavily interleaved audio (Kei's logistics announcements cut through Mark's and rtk947's remarks), so some attributions and the exact next-reading decision are reconstructed and may be imperfect.*