Glossary Updates16 new terms added to the glossaries · 12 November 2026, 21:23 CET
AI TechDocKnowledge
  • English (US)
  • English (UK)
  • Deutsch
Context cards of The art of thinking

The art of thinking · Context card

Frames, conceptual spaces and prototypes

How do frames, conceptual spaces and prototypes fit together?

By knowledge.aitechdoc.world

Reviewed

Review log and changes

The short answer

They describe three sides of a concept. A frame gives its structure — which attributes it has. A conceptual space gives the geometry of the values — how similar two instances are. A prototype gives the typical case and the defaults. The introduction to “Concepts, Frames and Cascades” treats them as connected rather than competing.

For: Information architects, terminologists and technical writers

Key points

  • Frames: the attributes of a concept and how they are connected (Barsalou).
  • Conceptual spaces: values as points in spaces of quality dimensions; similarity as distance; natural concepts as convex regions (Gärdenfors).
  • Prototypes: categories organised around typical members; membership is graded (Rosch).
  • They connect: frame attributes take values in conceptual spaces, and regions in those spaces behave much like prototypes.
  • Recent work refines typicality: modifiers can change other attributes, and typical members may be extremes rather than central cases.

The context

Three questions about one concept

Take the concept apple. Which attributes does an apple have — colour, taste, size, origin? That is the frame. How similar are a green and a yellow apple, and where does apple-coloured end? That is the conceptual space. What does a typical apple look like, and what may we assume about an apple nobody has described? That is the prototype.

How they connect

The introduction to Concepts, Frames and Cascades in Semantics, Cognition and Ontology (2021) notes that these notions are linked by integration rather than divided. A frame’s attribute names a dimension, the conceptual space gives that dimension its geometry, and the regions a concept occupies there have central, typical cases. Several chapters work at these joints:

  • Igor Douven builds a conceptual space for the types of implicature from experimental data and asks whether they are natural (convex) concepts.
  • Carla Umbach and Helmar Gust explain why similar is gradable and same is not, using attribute spaces of different granularity.
  • Corina Strößner, Annika Schuster and Gerhard Schurz model why “old coyotes howl” sounds less likely than “coyotes howl”: a modifier can change other attributes through constraints in the frame.
  • Robert van Rooij and Thomas Brochhagen argue that typical members are often extremes rather than averages.
  • Samuel D. Taylor and Peter R. Sutton weight features in category learning by their position in the frame.

Why it matters for models

Data models usually state attributes and types, sometimes value ranges, rarely similarity and almost never defaults. Concept theory suggests looking at all three: what is definitional, what is typical and may be overridden (as in inheritance), and how values can be compared. See the encyclopedia articles on conceptual spaces and prototypes and typicality.

Questions readers ask next

Do I need all three to model a concept?
No. They answer different questions. Most models need the structure; similarity and defaults become important where instances vary or where the model has to classify new cases.
Is a prototype the average member?
Often it is described that way, as a central point. Van Rooij and Brochhagen argue that the typical member can be an extreme — the clearest case rather than the average one.
What does convexity mean here?
That for any two members of a concept, everything between them in the space also belongs to it. Gärdenfors proposed it as the mark of a natural concept.

Sources

  1. Sebastian Löbner et al. (eds.), Concepts, Frames and Cascades in Semantics, Cognition and Ontology (open access, CC BY 4.0) — Springer, 1 January 2021
  2. Sebastian Löbner, Thomas Gamerschlag, Tobias Kalenscher, Markus Schrenk and Henk Zeevat, Introduction (in: Concepts, Frames and Cascades in Semantics, Cognition and Ontology, open access, CC BY 4.0) — Springer, 1 January 2021
  3. Igor Douven, Implicatures and Naturalness (in: Concepts, Frames and Cascades in Semantics, Cognition and Ontology, open access, CC BY 4.0) — Springer, 1 January 2021
  4. Carla Umbach and Helmar Gust, Grading Similarity (in: Concepts, Frames and Cascades in Semantics, Cognition and Ontology, open access, CC BY 4.0) — Springer, 1 January 2021
  5. Robert van Rooij and Thomas Brochhagen, Extremes are Typical. A Game Theoretical Derivation (in: Concepts, Frames and Cascades in Semantics, Cognition and Ontology, open access, CC BY 4.0) — Springer, 1 January 2021
  6. Corina Strößner, Annika Schuster and Gerhard Schurz, Modification and Default Inheritance (in: Concepts, Frames and Cascades in Semantics, Cognition and Ontology, open access, CC BY 4.0) — Springer, 1 January 2021
  7. Samuel D. Taylor and Peter R. Sutton, A Frame-Theoretic Model of Bayesian Category Learning (in: Concepts, Frames and Cascades in Semantics, Cognition and Ontology, open access, CC BY 4.0) — Springer, 1 January 2021

Review log and changes

Every context card is checked against its sources before it is published, and again whenever it changes; the date under the byline is the last review. Corrections (something was wrong) and additions (something was missing) are logged below with date and time (Berlin time). Typos, formatting and link fixes are not listed.

Reviewed

No corrections or additions since publication.