Почему коннектом червя не дает психику | Владимир Алипов
Vladimir Alipov discusses the limitations of connectome-based brain modeling, using the OpenWorm project as a case study. He explains that while a complete connectome of C. elegans has been mapped, simulating complex behavior requires much more than static connection maps — it requires dynamic modeling of synaptic plasticity and realistic neuron models. The talk concludes that connectome simulations alone cannot produce psyche or consciousness.
Summary
The talk begins with a brief discussion of Cortical Labs and their work with biological cell cultures for neuroscience research, noting that many neurobiologists are skeptical of cell cultures but that Cortical Labs may be revitalizing the field by demonstrating reliable learning in vitro. Alipov also explains the methods for obtaining neurons for culture, including harvesting from embryos, inducing adult cells into stem cells, and even growing personalized brain organoids from a specific patient's cells — raising ethical questions about exploiting someone's cellular properties.
The main focus shifts to connectome modeling, specifically the OpenWorm project, which attempts to simulate the complete nervous system of the nematode C. elegans (302 neurons) inside a virtual body in a virtual environment. Alipov explains that three models are needed for such a simulation: a brain model, a body model, and an environment model. While OpenWorm has achieved some success in reproducing simple behaviors like crawling, it has failed to simulate more complex behaviors such as social interaction, mating, and associative learning.
Alipov identifies several reasons for these failures: the neuron models used are overly simplified (not even Hodgkin-Huxley level), the model omits modulatory synaptic effects, and crucially, it treats the connectome as a static structure rather than a dynamic one. He emphasizes that synaptic connections change continuously as an organism interacts with its environment — a process called synaptic plasticity — and that this dynamic nature is essential for reproducing real behavior. The talk concludes that the OpenWorm simulation has no psyche, soul, or consciousness, and that a static connectome snapshot is fundamentally insufficient to model the mind.
Key Insights
- Alipov argues that OpenWorm's simulation fails to reproduce complex worm behavior partly because it uses an overly simplified neuron model — not even Hodgkin-Huxley level — and omits modulatory synaptic effects entirely.
- Alipov claims that a static connectome snapshot is insufficient for modeling behavior because synaptic connections continuously change as the organism interacts with its environment, making the brain a dynamic rather than static system.
- Alipov states that the OpenWorm simulation definitively has no psyche, soul, or consciousness, concluding that connectome modeling alone cannot produce subjective experience.
- Alipov explains that three separate models are required to test connectome-based behavior: a brain model, a body model, and an environmental model — all three must be sufficiently accurate.
- Alipov notes that human neurons can be grown in culture from any cell of a specific patient by inducing it into a stem cell and then into a neuroblast, raising ethical questions about exploiting the cellular properties of individuals without consent.
Topics
Full transcript available for MurmurCast members
Sign Up to Access