New Technology / Science
Follow science breakthroughs, applied research, frontier experiments and emerging technologies through curated structured summaries.
![Abstraction & Idealization: AI's Plato Problem [Mazviita Chirimuuta]](https://i.ytimg.com/vi/yq318DIwPqw/hqdefault.jpg)
Abstraction & Idealization: AI's Plato Problem [Mazviita Chirimuuta]
Topic
Philosophy of Neuroscience and AI
Key insights
- The relationship between neuroscience and the philosophy of mind raises questions about how much we can infer about the mind from neuroscience data, as generalizing lab results to real-world cognition is complex due to intricate interactions
- Mazviita Chirimuutas book, *The Brain Abstracted*, explores the implications of computational models in neuroscience, discussing how these models simulate brain functions and claim to replicate the functions of brain cells
- Abstraction and idealization are critical in scientific modeling; abstraction ignores certain details while idealization attributes false properties, which can mislead our understanding of complex systems
- Chirimuuta argues that both abstraction and idealization can lead to false representations in science, presenting a cleaner view of reality that does not reflect the complexities of the systems being modeled
- The relationship between neuroscience and the philosophy of mind highlights the challenges of generalizing lab results to real-world cognition due to complex interactions. Mazviita Chirimuuta's book, *The Brain Abstracted*, discusses how computational models in neuroscience can misrepresent brain functions through abstraction and idealization.
- Francois Chollets kaleidoscope hypothesis suggests that the universe is composed of neat mathematical rules, which AI researchers aim to decompose from the complex appearances of reality. This perspective aligns with Platonic philosophy, contrasting the messy world of appearances with an underlying stable truth
Perspectives
Analysis of the philosophical implications of neuroscience and AI.
Mazviita Chirimuuta's Perspective
- Challenges the mechanistic understanding of cognition
- Critiques the oversimplification in scientific models
- Emphasizes the importance of complexity in real-world cognition
- Argues against equating brain functions with computational models
- Highlights the role of human finitude in knowledge acquisition
- Questions the validity of AIs understanding without embodiment
AI Researchers' Perspective
- Propose that the universe operates on mathematical rules
- Assume that AI can replicate human cognition through computational models
- Argue for the potential of AI to achieve understanding through sensory-motor engagement
- Suggest that simplifications in models can yield valid insights
- Claim that technological advancements are disconnected from real-world constraints
Neutral / Shared
- Discusses the historical context of reflex theory in neuroscience
- Explores the relationship between philosophy and technology
- Considers the implications of digital engagement on human cognition
Metrics
publication_year
2024
year of publication for *The Brain Abstracted*
The publication year indicates the relevance of the ideas presented in the current discourse.
it came out in 2024
start_year
2018
year when writing of the book began
This timeline shows the long-term development of the ideas discussed in the book.
I think officially I started writing it maybe 2018
other
1943 year
the year McCulloch and Pitts published their landmark paper
This year marks a significant milestone in the development of neural networks.
McCollough and Pits, in then 1943, sort of landmark paper of interpreting neuronal cells as logic gates
energy_efficiency
more effective cognition than artificial neural networks, which are expensive to operate
comparison of biological and artificial cognition
Understanding this efficiency can inform the development of more sustainable AI systems.
we do a lot more with a very limited energy budget running our brains every day than his artificial neural networks are really really expensive to run.
other
all done online
obtaining a driving license
This highlights the shift from physical to virtual landscapes in daily life.
even to get a driving license it's all done online
other
more controlling pressures
social media compared to physical interactions
This suggests that digital engagement can significantly influence social dynamics.
there's almost more controlling pressures in the social media world than there is in our physical world
social_interaction
a lot less time looking at people's faces than they used to
decline in face-to-face interactions among children
This decline may affect children's future social skills and relationships.
young children nowadays spend a lot of less time looking at people's faces than they used to.
Key entities
Timeline highlights
00:00–05:00
The relationship between neuroscience and the philosophy of mind highlights the challenges of generalizing lab results to real-world cognition due to complex interactions. Mazviita Chirimuuta's book, *The Brain Abstracted*, discusses how computational models in neuroscience can misrepresent brain functions through abstraction and idealization.
- The relationship between neuroscience and the philosophy of mind raises questions about how much we can infer about the mind from neuroscience data, as generalizing lab results to real-world cognition is complex due to intricate interactions
- Mazviita Chirimuutas book, *The Brain Abstracted*, explores the implications of computational models in neuroscience, discussing how these models simulate brain functions and claim to replicate the functions of brain cells
- Abstraction and idealization are critical in scientific modeling; abstraction ignores certain details while idealization attributes false properties, which can mislead our understanding of complex systems
- Chirimuuta argues that both abstraction and idealization can lead to false representations in science, presenting a cleaner view of reality that does not reflect the complexities of the systems being modeled
05:00–10:00
Francois Chollet's kaleidoscope hypothesis posits that the universe is governed by mathematical rules that AI researchers strive to uncover. This perspective highlights the tension between the complexity of reality and the simplifications necessary for scientific modeling.
- Francois Chollets kaleidoscope hypothesis suggests that the universe is composed of neat mathematical rules, which AI researchers aim to decompose from the complex appearances of reality. This perspective aligns with Platonic philosophy, contrasting the messy world of appearances with an underlying stable truth
- The assumption that mathematical representations reveal deeper truths about reality is prevalent in AI research and scientific inquiry. However, this view can overlook the cognitive limitations of scientists, who often simplify complex systems to make them more tractable
- Abstraction in science is often employed due to our finite capacity to comprehend complexity, highlighting the necessity of simplification in modeling strategies. This mundane explanation challenges the notion that abstraction accesses a higher level of reality
- The concept of denoising data to uncover real patterns is subjective, as scientists decisions about what constitutes signal versus noise can obscure significant patterns. This classification process can lead to the creation of patterns rather than merely revealing them
- Models like Newtons are recognized as idealizations that simplify reality. The historical context of reflex theory, exemplified by Pavlovs experiments, illustrates how scientific models can persist despite their limitations
10:00–15:00
Reflex theory, once dominant in neuroscience, has been criticized for oversimplifying brain functions by reducing them to conditioned reflexes. The computational theory emerged as an alternative framework, yet it too risks idealization and neglects the subjective experiences that differentiate conscious beings from machines.
- Reflex theory, which dominated neuroscience at the end of the 19th century, proposed that all brain functions could be explained through conditioned reflexes. However, Charles Sherrington acknowledged that the notion of a simple reflex is an idealization that likely doesnt exist in reality, illustrating how elegant models can mislead researchers
- The rise of the computational theory during World War II provided an alternative framework that relied on idealization, allowing cognitive science to draw inferences about consciousness based on behavior. This approach treats systems as black boxes, often overlooking the subjective experiences that differentiate conscious beings from machines
- A constructivist perspective on knowledge posits that we actively create knowledge rather than merely discovering it. This view contrasts with scientific realism and empiricism, suggesting that our understanding of the world is shaped by our interactions and interpretations
15:00–20:00
Haptic realism posits that knowledge is generated through active engagement and interaction rather than passive observation. This perspective challenges traditional views of knowledge acquisition, emphasizing the importance of physical manipulation in understanding phenomena.
- Haptic realism emphasizes that knowledge is produced through active engagement and interaction, contrasting with passive observation. This perspective challenges the spectator theory of knowledge, asserting that understanding requires physical manipulation and involvement with the subject matter
20:00–25:00
The enterprise of science operates under the idealization of progressively getting closer to the truth, but this can lead to cul-de-sacs where progress stalls. Nature's inexhaustible complexity means any representation will lack completeness, necessitating a different approach to understanding knowledge.
- The enterprise of science operates under the idealization of progressively getting closer to the truth, but this can lead to cul-de-sacs where progress stalls. Scientific realism suggests a singular nature, yet the speaker argues that natures inexhaustible complexity means any representation will lack completeness
- The concept of nature as protean illustrates its constant change, making it impossible to fully capture in one representation. While scientists can obtain true answers, nature will continue to shift and present new challenges
- Biology presents a multitude of particularities that can seem less satisfying than the unified theories in physics, necessitating a different approach to understanding knowledge. This historical view of cognition as machine-like, traced back to Descartes, has influenced comparisons between machine-like reflexes and biological processes
- The evolution of the computational framework in understanding cognition connects to earlier theories like reflex theory and cybernetics, embedding the idea of machine-like processes in cognitive science
25:00–30:00
The research explores how mechanistic bodily processes can inform the engineering of non-living systems that mimic biological principles. It critiques the assumption that neuronal cells can be equated to logic gates, emphasizing the complexity of biological systems beyond computational models.
- The core research idea suggests that mechanistic bodily processes can inform the engineering of non-living systems that capture biological principles, tracing back to McCulloch and Pitts 1943 interpretation of neuronal cells as logic gates, which led to neural networks
- Creating devices inspired by biology to analyze biological phenomena can lead to neglecting details that do not fit non-living machines, risking the conclusion that the brain is merely a computer
- Connectionists advocate for the functional equivalence of biological and artificial systems, proposing that replicating brain mechanisms in machines could produce similar behaviors, but neuronal cells should not be seen as uniquely cognitive compared to other cells