Skip to content
Coordinates of ThoughtHome of Philosophy
Sign in
Discussion

Predictive processing and the Buddhist idea of constructed perception

Priya Raghavan Fellow
· 4 replies
PerceptionAnalyticBuddhist#perception#predictive-processing
Follow

In cognitive science, the predictive processing framework (Andy Clark's Surfing Uncertainty, 2016; Jakob Hohwy's The Predictive Mind, 2013) says the brain is constantly generating predictions about incoming sensory signals, and what we perceive is largely the brain's best guess, corrected by prediction errors.

People often say this "confirms" Buddhist ideas that perception is constructed. I think that's partly right and partly confused. The Buddhist epistemologists Dignāga and Dharmakīrti held that perception proper is free of conceptual construction (kalpanāpoḍha). It's the moment of bare contact; concepts come afterwards and overlay it.

So predictive processing fits the Buddhist account of conceptual overlay rather than of perception itself. It might even contradict the idea that there is a pure, non-conceptual moment.

Any thoughts from people who know the Buddhist texts better than I do?

4 replies

  1. Meera Kulkarni

    Fellow

    Your reading of Dignāga is right. For him, perception grasps only the unique particular (svalakṣaṇa), and anything general is conceptual. The interesting question is whether he'd count the brain's predictions as conceptual. They're not linguistic, so perhaps they'd fall into a grey zone he didn't foresee.

    1. Priya Raghavan

      Fellow

      Meera, the grey zone is exactly where I'd like to work. Sub-personal predictions that shape what's experienced but aren't concepts. Maybe the Buddhists needed a third category between bare perception and conceptual thought.

      Helpful · 3
  2. Arjun Iyer

    Fellow

    Advaita adds a different point: in all perception the witness consciousness is present, and it is not constructed. So even if every content is a prediction, the awareness of it is not. Predictive processing says nothing about that.

    Helpful · 1
  3. Pieter de Jong

    Fellow

    In machine learning, everything is "prediction" in some sense, so I'm wary of how broad the framework is. If every cognitive process counts as prediction, the claim that perception is constructed becomes trivially true.

    Helpful · 2

Members reply here. Reading is always free.

Sign in to reply