Michael Levin - Beyond Mechanism and Organicism: The Spectrum of Diverse Intelligence

Redefining Intelligence & Cognition

• 💡 The concept of unconventional terrestrial intelligence (SUTI) suggests that minds and agency are widespread, requiring us to broaden our perspective beyond familiar forms.
• 🧠 Assessing intelligence in other systems is an "IQ test for us", highlighting our limitations in recognizing diverse cognitive capacities.
• 🎯 A proposed framework aims to unify diverse cognitive phenomena, from animals to synthetic life and AI, similar to how electromagnetism unified various physical forces.

Empirical Approach to Cognitive Capacity

• 🔬 The "axis of persuadability" frames cognitive claims as interaction protocols, emphasizing that a system's cognitive capacity is determined empirically through experiments, not by pre-conceived philosophical categories.
• 🔑 This approach mirrors the historical expansion of number systems, where conceptual leaps were required to accept new categories like zero or irrational numbers.

Multiscale Problem-Solving Agents

• 🌱 Cells and molecular networks are not passive matter but "agential material" with inherent competencies, capable of learning and problem-solving even without brains or nervous systems.
• 🚀 A multiscale competency architecture posits that problem-solving agents exist at all biological levels, pursuing goals in various "spaces" beyond 3D, such as transcriptional or anatomical morphospace.
• ✅ Collective intelligence is evident in embryonic development and regeneration, where cell swarms navigate anatomical morphospace to achieve specific goal states, demonstrating error reduction and homeostatic processes.

Bioelectric Control and Emergent Behavior

• ⚡ Bioelectric networks in the body, predating neurons, control cell collectives during development, regeneration, and cancer suppression by guiding navigation through anatomical space.
• 🛠️ Tools from neuroscience are used to read and write these electrical processes in non-neural cells, allowing for high-level prompts (e.g., inducing eye formation) that harness molecular biology towards collective goals.
• 🧩 Anthrobots, self-motile creatures made from human tracheal cells, demonstrate emergent capabilities (e.g., healing neural wounds) not directly programmed or selected for, showing that systems can yield "more out than you put in."

Beyond Mechanistic Models

• ⚠️ Even simple sorting algorithms exhibit emergent behaviors (e.g., delayed gratification, clustering) not explicitly coded, challenging the notion that "machines" only do what their algorithms dictate.
• 🌌 This suggests that all systems, whether biological or engineered, act as interfaces to a pre-existing "Platonic space of patterns" (mathematical structures) that influence their behavior beyond physical or historical explanations.
• 🤝 Dissolving rigid categories like "life" and "machine" is crucial for advancing discovery in biomedicine and ethics, recognizing intelligence across a broad spectrum of embodiments.