Basal Cognition

Which minimal systems sense, remember, learn and regulate behavior, and how can those capacities be tested? Covers simple neural and non-neural organisms, and controls that distinguish learning from adaptation, fatigue or selection.

Capacities and behavioral organization

  • Plant cognition criteria - Adams requires cognitive representations beyond information processing; Segundo-Ortin and Calvo defend flexible, anticipatory plant behavior as cognitive. The dispute concerns both mechanisms and what counts as cognition. Adams 2018, Segundo-Ortin and Calvo 2019.
  • Stentor avoidance hierarchy - Repeated stimulation elicits an ordered repertoire of bending, ciliary changes, contraction and detachment in Stentor roeseli, despite individual variation. Replicates Jennings with the correct species; the hierarchy differs from habituation. Dexter et al. 2019.
  • Basal cognitive capacities - Organizes cognition around sensing, memory, learning and regulation across simple organisms with and without nervous systems. Proposes a research framework grounded in biological functions, rather than starting from human mental categories. Lyon et al. 2021.
  • Operational agency - The Technological Approach to Mind Everywhere compares interventions for predicting and controlling systems, from rewiring to training. Its authors distinguish goal pursuit from consciousness and affirm compatibility with materialism. Levin 2022, Levin and Resnik 2026.

Learning and alternative explanations

  • Physarum habituation - Repeated exposure reduces avoidance of quinine or caffeine barriers. Response recovery after withdrawal and stimulus specificity support habituation in this slime-mould assay. Boisseau et al. 2016.
  • Evidence for single-cell learning - Reassesses historical learning experiments and reasons for their dismissal. Argues that associative learning merits renewed investigation with stronger controls; a review of disputed evidence, not a new replication. Gershman et al. 2021.
  • Cellular spacing effect - Four spaced chemical pulses induce more persistent gene-reporter expression than a duration-matched pulse in two human cell lines. Inhibiting signaling and transcription pathways disrupts this molecular spacing effect. Kukushkin et al. 2024.

Origins of capacities

  • Capacities without direct selection - E. coli selected in one environment gains traits useful elsewhere. Levin proposes that embodiments access pre-existing patterns and minds; latent capacities alone do not distinguish this ontology from evolutionary mechanisms. Karve and Wagner 2022, Levin 2026.
  • Pattern-guided exploration - In simulated mazes, combining fractal and sensory inputs improves trained agents’ coverage over sensory-only inputs. Replacing fractals with uniform noise yields no significant change, leaving a fractal-specific benefit unresolved. Ertle et al. 2025.

Open questions

  • Does a changed population response reflect changes within tracked cells, differences in survival and reproduction, or both?
  • Which capacities transfer across tasks and preparations, and which depend on a particular assay?
  • Which predictions distinguish a proposed source of novel capacities from inherited components, physical dynamics and external training?
  • Hyperorganism
  • Biological computation - Using an organism to compute and testing its learning capacity are different investigations.
  • Environmental memory - Bacteria can carry the effects of earlier growth in the medium, changing later competition without requiring memory inside individual cells.
  • Neural interfaces - Closed-loop laboratory tasks expose both biological activity and apparatus contributions.
  • Consciousness - Memory or learning does not by itself establish experience.
  • Morphogenesis - Changing regenerative outcomes establishes physiological control; interpreting that control as a trainable goal requires additional tests.
  • Engineered living constructs - Xenobots and Anthrobots test capacities of cells assembled outside their usual anatomy.