Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Cognition spaces: natural, artificial, and hybrid

Ricard Solé, Luis F Seoane, Jordi Pla-Mauri, Michael Timothy Bennett, Michael E. Hochberg, Michael Levin

TL;DR

The paper advocates a cognition spaces framework to compare cognitive processes across natural, artificial, and hybrid substrates, moving beyond substrate-based definitions. It defines three morphospaces—basal cognition, neural cognition, and human-AI hybrid cognition—each with distinct axes and occupancy patterns that reveal evolutionary and design constraints. By analyzing voids and hybrid opportunities, the work highlights how development, embodiment, and interaction shape cognitive possibilities and identifies hybrid cognition as a promising frontier for richer, scalable cognition. The framework has implications for designing resilient human-AI systems, understanding collective cognition, and probing new forms of complexity, while cautioning about alignment, autonomy, and dependency risks.

Abstract

Cognitive processes are realized across an extraordinary range of natural, artificial, and hybrid systems, yet there is no unified framework for comparing their forms, limits, and unrealized possibilities. Here, we propose a cognition space approach that replaces narrow, substrate-dependent definitions with a comparative representation based on organizational and informational dimensions. Within this framework, cognition is treated as a graded capacity to sense, process, and act upon information, allowing systems as diverse as cells, brains, artificial agents, and human-AI collectives to be analyzed within a common conceptual landscape. We introduce and examine three cognition spaces -- basal aneural, neural, and human-AI hybrid -- and show that their occupation is highly uneven, with clusters of realized systems separated by large unoccupied regions. We argue that these voids are not accidental but reflect evolutionary contingencies, physical constraints, and design limitations. By focusing on the structure of cognition spaces rather than on categorical definitions, this approach clarifies the diversity of existing cognitive systems and highlights hybrid cognition as a promising frontier for exploring novel forms of complexity beyond those produced by biological evolution.

Cognition spaces: natural, artificial, and hybrid

TL;DR

The paper advocates a cognition spaces framework to compare cognitive processes across natural, artificial, and hybrid substrates, moving beyond substrate-based definitions. It defines three morphospaces—basal cognition, neural cognition, and human-AI hybrid cognition—each with distinct axes and occupancy patterns that reveal evolutionary and design constraints. By analyzing voids and hybrid opportunities, the work highlights how development, embodiment, and interaction shape cognitive possibilities and identifies hybrid cognition as a promising frontier for richer, scalable cognition. The framework has implications for designing resilient human-AI systems, understanding collective cognition, and probing new forms of complexity, while cautioning about alignment, autonomy, and dependency risks.

Abstract

Cognitive processes are realized across an extraordinary range of natural, artificial, and hybrid systems, yet there is no unified framework for comparing their forms, limits, and unrealized possibilities. Here, we propose a cognition space approach that replaces narrow, substrate-dependent definitions with a comparative representation based on organizational and informational dimensions. Within this framework, cognition is treated as a graded capacity to sense, process, and act upon information, allowing systems as diverse as cells, brains, artificial agents, and human-AI collectives to be analyzed within a common conceptual landscape. We introduce and examine three cognition spaces -- basal aneural, neural, and human-AI hybrid -- and show that their occupation is highly uneven, with clusters of realized systems separated by large unoccupied regions. We argue that these voids are not accidental but reflect evolutionary contingencies, physical constraints, and design limitations. By focusing on the structure of cognition spaces rather than on categorical definitions, this approach clarifies the diversity of existing cognitive systems and highlights hybrid cognition as a promising frontier for exploring novel forms of complexity beyond those produced by biological evolution.
Paper Structure (5 sections, 7 equations, 4 figures)

This paper contains 5 sections, 7 equations, 4 figures.

Table of Contents

  1. Introduction
  2. Basal cognition space
  3. Neural cognition space
  4. A space of human--AI dyads
  5. Discussion

Figures (4)

  • Figure 1: Natural, artificial, and hybrid cognition across organizational scales. Representative systems are organized along three columns---natural, artificial, and hybrid---and levels of organizational complexity: basal cognition, biocomputation, swarm intelligence, ecosystems, and humans/AI. Examples in (a--c) illustrate basal cognitive and path-finding processes in homogeneous media (a), simulated particle swarms (b), and hybrid settings (c) as defined by human-designed graphs. In (d--f), different systems performing computations are shown: (d) an amoeba, (e) Lenia, and (f) a microfluidic device with spatially segregated cell strains. In (g--i), we display collective intelligence examples from social insects (g), robotic swarms (h), and biohybrid insect--robot systems (i). Finally, when dealing with higher cognition and AI (j--l), highlight human brains (j), neuromorphic computers (k), and human--AI dyads (l).
  • Figure 2: A Morphospace of basal cognition. A space of living (or hybrid). Here, biological, artificial, and hybrid systems are positioned within a three-dimensional space defined by spatial complexity, computational complexity, and developmental complexity. These axes capture (in an aggregated manner) the diversity and complexity of decision making, multicellular organization, and the role played by development. The diagram spans systems from single cells, bacterial colonies, and protozoa to multicellular organisms, organs, and engineered living systems such as organoids, xenobots, and neurally interfaced constructs. The dark shaded region at the bottom left highlights regimes of basal cognition based on synthetic systems. Illustrative sketches anchor representative systems along each axis, emphasizing the diversity of natural and engineered forms of minimal cognition and the presence of large voids.
  • Figure 3: A Morphospace of neural cognitive complexity. A conceptual space of cognitive systems spanned by agency, computational complexity, and agent--agent interactions. Biological systems (dark markers) span a wide range of agency values, from minimal organisms to humans. Artificial systems (light markers) cluster at high computational complexity but low agency. The shaded region highlights a large gap along the agency axis, motivating the search for hybrid systems combining artificial computation with biological constraints.
  • Figure 4: A morphospace of human--AI pairwise interactions, illustrated with representative systems. A conceptual space of cognitive agencies spanning biological, artificial, and hybrid systems. The axes represent human cognitive complexity, artificial cognitive complexity, and the degree of human--AI exchange. Two major subsets are indicated within the cube: embodied social agents and disembodied social agents. The figure emphasizes how different forms of cognition and interaction occupy distinct regions of this space. Some specific examples are indicated, involving diverse case studies: deep hybrid agencies between humans and AI agents; companion-type pet robot (Sony Aibo); affective social robot (Kismet); Large Language Models (LLM); disembodied AGI; early text-based chatbot (ELIZA). Affective social robots (Kismet, iCub, Aibo, Paro) occupy the region of low to moderate AI cognition but high human cognitive engagement. Disembodied social agents (ELIZA, Watson, Alexa) cluster in an area of moderate AI cognition and moderate to high user engagement.