MEART created drawings with neural control and showed learning when training stimulus was updated.

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Neuroscience and mechanistic interpretability have not yet made enough progress to identify neural correlates marking necessary and sufficient conditions of conscious experience in both brains and neural networks.claim0.767
  Paper explicitly identifies this as a current gap requiring alternative experimental approaches
• Pretraining stores latent patterns that coherent anchors can bind (or misbind) to targets.quote0.750
  Load-bearing quote capturing the core metaphor
• Human cognition evolved to detect agency in medium-sized objects at medium speeds in 3D space, limiting recognition of intelligence in unfamiliar substrates.claim0.745
• Consciousness in AI is best assessed by drawing on neuroscientific theories of consciousness.claim0.743
  Central methodological claim of the paper.
• Neural networks, trained with different objectives on different data and modalities, are converging to a shared statistical model of reality in their representation spaces.quote0.743
  The paper's central thesis statement, presented prominently after the abstract
• Neural cultures learn to control virtual and robotic bodies in closed-loop systemsfinding0.743
• The collective intelligence of tissues is sophisticated enough to be trainable via reinforcement learning for specific morphological outcomes.hypothesis0.743
  Ongoing test prediction: tissues can associate stimuli with rewards to modify anatomy.
• The experimental method of starting with a vision and adding color to create life in the centers has predictable results: the eleven invariants.claim0.742
  Assertion that the process yields a specific set of color qualities, listed in the chapter.