What is it like to be a bat? (1974) [pdf]

Original Article ↗ Hacker News Discussion ↗

Overall view of the paper

  • Widely treated as a landmark in philosophy of mind; some call it seminal and memorable.
  • Others find it overrated or even harmful, arguing its core point could be compressed to a short paragraph.
  • Some criticize it as relying on ambiguities around “to be,” or on language in general, while defenders say it raises a genuinely hard problem, not a mere semantic quibble.

Subjective experience vs imagination

  • Central tension: science gives shareable descriptions; subjective experience (what-it’s-like) is not obviously shareable.
  • Multiple comments stress the gap between “imagining being a bat as a human” and “what it is like for a bat to be a bat.”
  • Empathy and imagination are seen as partial, human-centric simulations that you can “step out of,” unlike an organism’s native experience.

Animal consciousness & what-it’s-like

  • Debate over whether we can meaningfully ascribe subjective experience to bats, dogs, octopuses, etc.
  • Some think only humans (or at least not all animals) are “persons” with genuine first‑person perspectives; others see consciousness as widespread, possibly “all the way down.”
  • There is dispute over whether experience can exist without a “point of view” or subject. This remains unresolved in the thread.

Qualia, gestalt, and the Hard Problem

  • Many frame the discussion in terms of qualia: the irreducible feel of experiences (pain, color, sonar).
  • Others prefer “gestalt” for the overall integrated perspective of a system.
  • Strong emphasis that full objective description (brain states, functions) doesn’t seem to yield the subjective feel.

Scientific and computational analogies

  • Comparisons to human echolocation, sonar screens, embeddings, and predictive processing.
  • Some argue studying bat brains could reveal the structure of echolocation experience; others counter that this still misses the bat’s own subjectivity.
  • Several note that description vs execution/recursion is key: running the process “from inside” has a cost and context we cannot share.

AI/LLMs and consciousness

  • Multiple anecdotes about chatting with language models that “feel” spooky or self‑aware, but commenters note they are trained to deny sentience.
  • Some propose that current LLMs lack recurrent self‑models and attentional autonomy, so their “self‑talk” is only a simulacrum.
  • Others suggest future systems with unified self‑representation might have genuine what‑it’s‑like experiences, which would be hard to detect and raise moral concerns.

Metaphysical positions

  • Positions discussed include strict materialism, dualism, panpsychism, and language-based accounts of consciousness.
  • No consensus; several comments stress that we lack criteria or proof methods for deciding which systems truly have subjective experience.

Miscellaneous side topics

  • Speculation about bats sensing each other’s “views,” octopus minds, plant experience, and analogies like thermostats, trust‑fund kids, and radio‑frequency “souls.”
  • Some speculative ideas are criticized as pseudo‑scientific or unfalsifiable.