Working definition
Plasticity is a persistent change in connection efficacy or system state caused by activity or intervention; learning is an operational performance change under a defined task and evaluation protocol. The two may be related but are not synonyms. Hardware updates, software optimization, and biological adaptation require different measurements and cannot share one unqualified learning claim.
Mechanism
- Observe activity and an eligibility condition.
- Apply a local, global, engineered, or biological update.
- Test persistence and held-out task consequences.
Measurements
- Pre/post performance
- Weight or response change
- Retention and transfer
Reproducibility controls
- Version hardware, software, firmware, and analysis code.
- Declare dataset, preprocessing, random seeds, and measurement boundary.
- Report repeated runs, variation, exclusions, and failed trials.
Limits and failure modes
- Adaptation can reflect drift or damage.
- Task improvement does not imply general learning.
Mathematical connection
Formal structure without substrate erasure
Detecting persistent adaptation
Test whether performance or response changes after an intervention beyond expected drift.
Inputs
- Timestamped outcomes
- Intervention time
- Drift model
Outputs
- Candidate change point
- Effect estimate
- False-alarm control
Limit: A change point cannot by itself distinguish learning from damage, maturation, fatigue, controller changes, or analysis flexibility.
Technical and governance sources
- [1]Taking Neuromorphic Computing to the Next Level with Loihi 2 · Intel Labs
Establishes: An official description of the Loihi 2 research chip, its programmable neuron models, event-based communication, on-chip learning support, and the Lava software framework used to construct neuromorphic applications.
Boundary: This is a vendor technical brief about a research platform. Performance and efficiency results remain workload-, configuration-, measurement-boundary-, and comparison-dependent and do not establish equivalence to biological intelligence.
- [2]In vitro neurons learn and exhibit sentience when embodied in a simulated game-world · Neuron
Establishes: A primary experiment coupling human- and rodent-derived neuronal cultures on high-density multielectrode arrays to a closed-loop simulated Pong environment and reporting task-related electrophysiological adaptation.
Boundary: The observed closed-loop behavior is a bounded experimental result. The paper title’s use of sentience is not accepted here as proof of consciousness, subjective experience, general intelligence, or deployable biological computing.
- [3]Interlaboratory reproducibility of microelectrode array measurements of spontaneous activity in networks of cultured neurons · Journal of Biomolecular Screening
Establishes: An interlaboratory study examining whether specified culture and multielectrode-array procedures can produce comparable spontaneous neural-network measurements across sites.
Boundary: Reproducibility under the studied protocol does not generalize to every cell source, maturation state, array, stimulation regime, analysis pipeline, learning task, or claimed computational capability.