What’s wrong with modeling?

2022-09-02 · Source: computational biology blog · Field: Science & Research — Life Sciences & Biology, Mathematics & Computational Sciences, Research Methodology & Innovation · Depth: Advanced, quick

Summary

Computational neuroscience often employs bottom-up modeling, such as building network models of the cerebellar cortex that incorporate diverse cell types and synaptic connections. This approach aims to understand the functions of these biological components, utilizing methods from statistical physics, nonlinear dynamics, and machine learning, while being constrained by electrophysiological, transcriptomic, and connectomic data. However, a critique suggests that this prevalent bottom-up methodology, despite being considered "state of the art," inherently limits true functional understanding. It primarily synthesizes existing information, pre-selects model aspects, and often omits crucial biological details deemed "irrelevant" to the model, potentially overlooking significant functional mechanisms like whole-neuron learning or intrinsic excitability, which were historically dismissed.

Key takeaway

For AI researchers developing computational models of biological systems, you should critically evaluate the limitations of purely bottom-up approaches. Consider incorporating top-down reasoning, focusing on evolutionary advantages and logical requirements for neuronal operation, to avoid overlooking crucial functional mechanisms like intrinsic excitability that may be dismissed by data-driven constraints.

Key insights

Bottom-up neuroscience modeling can hinder functional understanding by pre-selecting data and overlooking crucial biological mechanisms.

Principles

• Model constraints can limit functional discovery.
• Evolutionary perspective aids functional understanding.

In practice

• Consider top-down modeling for functional insights.
• Integrate intrinsic excitability into learning models.

Topics

• Computational Neuroscience
• Neural Modeling
• Intrinsic Excitability
• Bottom-Up Modeling
• Top-Down Modeling

Best for: AI Researcher, Research Scientist, AI Scientist

Related on AIssential

• Evolution With Purpose: Hierarchy-Informed Optimization of Whole-Brain Models — Guiding evolutionary optimization with biological hierarchy improves whole-brain model generalization and behavioral predictability.
• Why Every Brain Metaphor in History Has Been Wrong [SPECIAL EDITION] — Scientific models are useful simplifications, not necessarily literal truths about underlying reality.
• This Fly is LIVING in the Matrix... — A fully simulated fruit fly, based on biological neural mapping, demonstrates complex behaviors without AI training.
• Uneven Evolution of Cognition Across Generations of Generative AI Models — Generative AI models exhibit uneven cognitive development, excelling in language but struggling with visual-perceptual reasoning.
• Models of the Mind: How physics, engineering, and mathematics have shaped our understanding of the brain — Mathematical models are crucial for explicit hypothesis testing and understanding the brain's complex, evolved systems.
• They Copied a Fly's Brain Into a Computer. It Started Walking. Nobody Taught It How — Biological brain wiring can inherently produce complex behaviors without explicit training or optimization.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by computational biology blog.