Are mental properties supervenient on brain properties?

Joshua T Vogelstein, R. Jacob Vogelstein, Carey E. Priebe

Abstract

The "mind-brain supervenience" conjecture suggests that all mental properties are derived from the physical properties of the brain. To address the question of whether the mind supervenes on the brain, we frame a supervenience hypothesis in rigorous statistical terms. Specifically, we propose a modified version of supervenience (called $\epsilon$-supervenience) that is amenable to experimental investigation and statistical analysis. To illustrate this approach, we perform a thought experiment that illustrates how the probabilistic theory of pattern recognition can be used to make a one-sided determination of $\epsilon$-supervenience. The physical property of the brain employed in this analysis is the graph describing brain connectivity (i.e., the brain-graph or connectome). $\epsilon$-supervenience allows us to determine whether a particular mental property can be inferred from one's connectome to within any given positive misclassification rate, regardless of the relationship between the two. This may provide further motivation for cross-disciplinary research between neuroscientists and statisticians.

Introduction

The relationship between the mind and the brain has long been a central debate in analytical philosophy. At the heart of this discussion is the concept of supervenience: the claim that mental properties ($M$) are entirely dependent on physical brain properties ($B$). In essence, supervenience posits that there can be no change in the mind without a corresponding change in the brain. While this idea is foundational to modern neuroscience, it has historically remained a philosophical conjecture, lacking a framework for empirical validation.

This paper, published in Scientific Reports, bridges the gap between abstract logic and empirical science by translating supervenience into a testable statistical hypothesis. By leveraging pattern recognition theory and connectomics, the authors provide a roadmap for investigating the physical substrate of the mind.

Defining $\epsilon$-Supervenience

In classical philosophy, supervenience is often framed as a deterministic "if/then" relationship. However, biological data is inherently noisy, and scientific measurements are probabilistic. To reconcile these views, the authors redefine supervenience through the lens of classification error.

They equate supervenience to the existence of a perfect classifier. If the mind truly supervenes on the brain, there must be a function that maps every brain state to a mental state with zero error. The minimum possible error rate for any such mapping is known as the Bayes error ($L^*$). Under strict supervenience, $L^* = 0$.

Because it is mathematically impossible to prove that an error rate is exactly zero using finite data, the authors introduce $\epsilon$-supervenience. This relaxed definition suggests that a mental property supervenes on a brain property if the Bayes error is less than some small, pre-defined threshold $\epsilon$. This transformation turns a metaphysical claim into a hypothesis that can be tested using machine learning algorithms.

Methodology: Connectomes and Classifiers

To test for $\epsilon$-supervenience, the authors propose using universally consistent classifiers, such as the $k_n$-Nearest Neighbor ($k_n$-NN) algorithm. These algorithms are powerful because they are guaranteed to converge to the optimal Bayes error as more data is collected, regardless of how complex the relationship between the brain and mind might be.

For the physical property ($B$), the study focuses on the connectome—a graph-based map of neural connections. By representing the brain as a graph, researchers can use mathematical tools like the Frobenius norm to measure the "distance" between different brain states, providing a structured space for the classifier to operate.

Experimental Illustration: The C. elegans Model

The authors demonstrated their framework using a synthetic dataset derived from the known connectome of the nematode Caenorhabditis elegans, which consists of 279 somatic neurons.

In this simulation:

  • Physical Property ($B$): Variations of the C. elegans brain-graph with added stochastic noise.
  • Mental Property ($M$): A simulated "odor-evoked behavior" categorized into two classes (present or absent).

The goal was to determine how many samples are required to statistically confirm supervenience. The results showed that with a threshold of $\epsilon = 0.1$, the researchers could confirm supervenience with 99% confidence using a sample size of $n=360$. At this size, the classifier achieved an error rate of approximately 0.057, successfully rejecting the null hypothesis that the true error was $\ge 0.1$.

Philosophical and Practical Implications

A significant finding of the paper is the one-sided falsifiability of the supervenience test. While the framework can prove that a mental property supervenes on a brain property (within error $\epsilon$), it cannot strictly disprove supervenience. If a classifier fails to meet the $\epsilon$ threshold, it could mean that supervenience is false, or it could simply mean that the researcher hasn't collected enough data or hasn't measured the correct physical properties.

Furthermore, the authors note that while this approach works for a 279-neuron worm, scaling it to the 86 billion neurons of the human brain presents a monumental data challenge. However, the framework provides a rigorous way to test which specific brain features (e.g., structural connectivity vs. functional activity) are the true substrates of specific mental states.

Conclusion

By providing a rigorous statistical definition for a classical philosophical problem, this work offers a new way for neuroscientists and statisticians to collaborate. It moves the conversation from "Does the mind depend on the brain?" to "With what degree of certainty can we predict mental states from neural architecture?"

Summary updated

Abstract

The "mind-brain supervenience" conjecture suggests that all mental properties are derived from the physical properties of the brain. To address the question of whether the mind supervenes on the brain, we frame a supervenience hypothesis in rigorous statistical terms. Specifically, we propose a modified version of supervenience (called ε-supervenience) that is amenable to experimental investigation and statistical analysis. To illustrate this approach, we perform a thought experiment that illustrates how the probabilistic theory of pattern recognition can be used to make a one-sided determination of ε-supervenience. The physical property of the brain employed in this analysis is the graph describing brain connectivity (i.e., the brain-graph or connectome). ε-supervenience allows us to determine whether a particular mental property can be inferred from one's connectome to within any given positive misclassification rate, regardless of the relationship between the two. This may provide further motivation for cross-disciplinary research between neuroscientists and statisticians.
BibTeX
@inproceedings{vogelstein2011are,
  author = {Joshua T Vogelstein and R. Jacob Vogelstein and Carey E. Priebe},
  title = {{Are mental properties supervenient on brain properties?}},
  booktitle = {Scientific},
  year = {2011},
  doi = {10.1038/srep00100},
}
Figure