Skip to content

22 Through the Fractured Looking Glass Sandra D. Mitchell

Key Themes

  • The differences between theories about phenomena and causal models of those phenomena
  • The partiality of all models – every model we use is only a partial representation of the world because it is, of necessity, a) abstract and b) idealized.
  • The epistemological danger of a standard assumption in inquiry that there is one right model of nature – one true representation of what nature is “doing” to produce the phenomena that interest us (Mitchell calls this the “mirror-of-nature” approach).
  • The epistemological value of using different strategies for detection and measurement that imply diverse models complex phenomena that can either a) be contradictory, b) be mutually reinforcing, or c) provide integrated, mutually correcting insight into the phenomenon being modeled.

Philosophy of Science and Pluralism

Mitchell advocates for epistemic diversity in scientific practice, mirroring the diversity of questions and methods in nature. She emphasizes:

  • Rejecting rigid dichotomies in scientific methods and explanations.
  • Valuing multiple, interacting perspectives to uncover deeper insights.

Model Partiality and Integration

  • Partial Representations: Scientific models simplify reality by selectively focusing on specific aspects, leaving out others.
  • Perspectival Selection: Every model encodes a unique perspective, shaped by its goals and methods.
  • Integration Over Reduction: Rather than reducing all models into a single one, science benefits from integrating diverse models to address complex phenomena.

Case Studies in Model Diversity

  1. London Underground Maps: Illustrates how different models serve different purposes. The 1908 map prioritizes geographic accuracy, while the 1933 map optimizes navigation. Both are valid but serve distinct goals.
  2. Protein Structure Analysis:
    • Techniques like X-ray crystallography and NMR spectroscopy provide divergent, yet complementary, models of protein structures.
    • Integration through "joint refinement" reconciles differences, yielding more accurate predictions.

Challenges of Convergence and Divergence

  • Converging Models: Agreement between models strengthens the confidence in their predictions.
  • Diverging Models: Even when models disagree, their integration can refine understanding by addressing systematic biases and highlighting blind spots.

Philosophical Implications

Mitchell argues against a singular, objective "truth" in scientific representations. Instead:

  • The plurality of perspectives enriches our understanding of complex systems.
  • Integration does not eliminate diversity but uses it to create mutually informative, more reliable representations.