Loading...

Postulate is the best way to take and share notes for classes, research, and other learning.

More info

Source notes

Why Cognitive Science Needs Philosophy and Vice Versa

Profile picture of Samson ZhangSamson Zhang

Paul Thagard (U of Waterloo) in Topics in Cognitive Science 1 (2009), DOI: 10.1111/j.1756-8765.2009.01016

Feb 1, 2022Last updated Feb 3, 2022

Summary

Cogsci needs philosophy to 1) define methodology as a frontier science and 2) resist faulty normative extrapolations of cogsci like Bayesianism and maximizing utility functions. As an almost separate point, Thagard briefly argues that the stacked subdisciplines of cogsci, i.e. social => psychological => neurological => molecular, are best seen as interacting with each other rather than being a uni-directional chain in one direction or another.


Takeaways

  • Good overview of different disciplines of cogsci and how they interact
  • Interesting pushback against lesswrong-type rationalism defined by Bayesianism and utility functions. I don't think it's damning; the questions that Thagard proposes should deepen and strengthen this body of thought rather than weaken it if truly taken into consideration. But it also pokes important holes in the direct link between biology and normative philosophy that many in the tech-ish crowd rely on.
  • Also pushback against science-rejecting social sciences like post-modernism, which I personally align with through media studies. I agree with Thagard that this kind of philosophy is much flimsier, but I think its theories describe social systems and challenge existing understandings of social systems in ways that "naturalistic philosophy" fail to

Notes

Philosophy helps cogsci with generality and normativity

  • Generality: the nature of theories, explanations, methodology; relations between sub-disciplines
    • Cogsci is a frontier science => what is an acceptable theory? What is causation? These questions don't have pre-defined answers like in established sciences; philosophy must provide them
    • Cogsci is multi-disciplinary => do molecular conditions define neural, psychological, and social conditions (in a chain)? Or social conditions psychological, neural, and molecular? Thagard: they all impact each other; other approaches are reductionist, but easy to slip into if not aware of past philosophical doctrines that outline them


  • Normativity: what should be done. Cogsci often implicitly informs philosophy, but this connection is not examined. A dunk on tech culture rationalism
    • Bayesian model used in cogsci frequently => implication that Bayesian thinking should be the norm. But things like probability are undefined, and it's not nearly universally best to use Bayesian reasoning (high-level decisions, scientific discovery, etc. vs. situations where a large quantity of data is already available)
    • Utility functions => what is utility? Are needs reducible to a single metric, or should they be considered separately? Naturally it's not even clear if categories of needs are maximized or satisfied, so normative projections of cogsci research is sus

Philosophy provides the foundations for science, but science provides the methods to extend it

  • Philosophical reasoning builds a chain of logic that is only as strong as its weakest link
  • Science builds a cable with many thin threads supporting each other -- they can break and change with time much more robustly (Pierce, 1958)

Comments (loading...)

Sign in to comment

LGCS 11: Intro to Cognitive Science

Notes from Spring '22 Pomona class LGCS 11 with Prof. Zirnstein