# Algorithms on psychedelics: Can computation get high? | Michael Levin and Lex Fridman

Source: https://www.youtube.com/watch?v=mDaFzC93Sfs
Recap page: https://rapidrecap.app/video/mDaFzC93Sfs
Generated: 2025-12-05T21:39:01.151+00:00

---
## Quick Overview

The discussion between Lex Fridman and Michael Levin primarily concludes that while computational systems can model biological processes like morphogenesis through algorithms, the current scientific understanding lacks the necessary intuition or generalized rules to definitively predict where to stop applying these models, underscoring the gap between computational simulation and true biological understanding, especially when anthropomorphizing phenomena like consciousness or psychedelic effects.

**Key Points:**
- Michael Levin argues that the complexity of biological systems, like those studied by Jagadish Chandra Bose concerning plant sensitivity, suggests that intelligence emerges from inherent computational mechanisms, not necessarily from complex structures like brains.
- Levin distinguishes between linear goals (like achieving a specific temperature) and non-linear, goal-directed behaviors, suggesting that the latter requires a different kind of computational framework to model.
- The concept of 'anthropomorphizing' complex phenomena (like assigning consciousness to algorithms or psychedelic effects) is a category error because the underlying computational principles are being misapplied or over-generalized.
- Levin references the work of Jagadish Chandra Bose (1858-1937), who showed plants respond to anesthetics, demonstrating responsiveness beyond just animals.
- The inherent intelligence in biological systems is powerful, but current science lacks the intuition to know precisely where to stop applying computational models or when the model's boundary has been reached.
- Levin asserts that the mathematical framework they use, derived from sorting algorithms, can model basal intelligence and predict outcomes in chaotic systems, even if the intuition about the system's limits is missing.

![Screenshot at 07:42: Michael Levin displays an image comparing spider web construction under the influence of various substances \(Normal, LSD, Marijuana, Sleeping Pills, Caffeine, Benzedrine\) to illustrate how different inputs cause predictable, yet structurally varied, outcomes in a self-organizing system, linking this concept to algorithmic behavior.](https://ss.rapidrecap.app/screens/mDaFzC93Sfs/00-07-42.png)

**Context:** This segment features an interview between Lex Fridman and Michael Levin, likely discussing Levin's research on computational morphogenesis, basal intelligence, and the potential for complex behaviors to arise from simple, self-organizing computational rules. The conversation centers on the philosophical and scientific implications of applying algorithmic models to biological phenomena, contrasting linear goal-setting with emergent, complex system behavior, and touching upon the historical context of early 20th-century research into plant sensitivity.

## Detailed Analysis

Michael Levin explains to Lex Fridman that setting goals for computational systems needs to be precise, distinguishing between linear goals and those requiring dynamic, complex behavior, which he relates to the concept of 'cognitive light cones' that define immediate vs. long-term outcomes. Levin critiques the tendency to anthropomorphize, warning against applying concepts like consciousness or the effects of psychedelics to algorithms without the necessary computational framework, calling this a 'category error' (02:13). He cites the historical work of Jagadish Chandra Bose (00:54-01:00) who demonstrated that plants respond to anesthetics, challenging the exclusivity of responsiveness to animals. Levin argues that the inherent intelligence in systems like sorting algorithms, when analyzed through the lens of morphogenesis, shows they can model complex biological phenomena (02:04). He illustrates this by showing how external inputs (like drugs on spiders building webs, 07:42) cause predictable structural changes. The core issue, according to Levin, is that while the science can model these systems, current intuition is insufficient to know where to draw the line or stop the process, leading to a potentially short-sighted approach if one assumes current knowledge is complete (08:22).

### Goal Setting & Computation

- Goals must be precisely defined, distinguishing between linear pursuit and complex, non-linear outcomes
- The concept of a cognitive light cone separates immediate from long-term effects
- Anthropomorphizing results in category errors when applying concepts like consciousness to computation.

### Historical Context & Biological Intelligence

- Reference to Jagadish Chandra Bose (1858-1937) and his work showing plants respond to anesthetics, suggesting intelligence is substrate-independent
- The inherent intelligence in biological systems is powerful but hard to map precisely.

### Algorithmic Modeling

- Sorting algorithms serve as a minimal model for basal intelligence and morphogenesis
- Experiments show that applying external variables (like anesthetics or drugs on spider webs) predictably alters system output.

### Scientific Limitations

- Current scientific intuition is inadequate for determining the boundary conditions or stopping points for these models
- Relying solely on existing knowledge without further experimentation is short-sighted.

![Screenshot at 00:02: Lex Fridman Podcast intro screen showing an Earth horizon view from space.](https://ss.rapidrecap.app/screens/mDaFzC93Sfs/00-00-02.png)
![Screenshot at 00:03: Lex Fridman sitting at a desk with a microphone, looking down at papers.](https://ss.rapidrecap.app/screens/mDaFzC93Sfs/00-00-03.png)
![Screenshot at 00:16: Michael Levin speaking animatedly into a microphone, gesturing with his hands in a home-like studio setting.](https://ss.rapidrecap.app/screens/mDaFzC93Sfs/00-00-16.png)
![Screenshot at 01:44: Lex Fridman writing on a document while listening to the guest.](https://ss.rapidrecap.app/screens/mDaFzC93Sfs/00-01-44.png)
![Screenshot at 07:42: Graphic displaying spider web construction under the influence of various substances \(Normal, LSD, Marijuana, etc.\) to illustrate systemic response to external inputs.](https://ss.rapidrecap.app/screens/mDaFzC93Sfs/00-07-42.png)