Gordon Pask
24 Jan 2024 - 20 Oct 2026
- One of many odd figures in cybernetics . I met him once (via Ted Nelson) and he offered me a little pink pill which I assume was LSD. Unfortunately at the time I was pretty straight (this was when I was still in high school in Chicago) and so I flushed it.
- Paul Pangaro is curator of his legacy.
Sometime during this period, Pask met George Spencer-Brown who became a lodger at the Pask family's home while working at Stafford Beer and Roger Eddison's operational research consultancy SIGMA (Science in General Management) via strong recommendation from Bertrand Russell. It was here where Spencer-Brown is said to have written his Laws of Form for long hours whilst inebriated in the Pask family's bathtub. According to Vanilla Beer, Stafford's daughter, Pask is purported to have claimed while reminiscing about Spencer-Brown's time at his and his wife's household, that "When [Spencer-Brown] bathed, it wasn't often. He used my gin, to wash in."
Gordon Pask and Stafford Beer save the world, circa 1958 pic.twitter.com/iU7qHGkUdF
— Evgeny Morozov (@evgenymorozov) April 29, 2022
from Pangaro- He co-wrote a weird book on calculator games Calculator Saturnalia, or, Travels with a calculator : a compendium of diversions & improving exercises for ladies and gentlemen : Pask, Gordon : Free Download, Borrow, and Streaming : Internet Archive
Vienna is a city of wisdom and colossal architecture, the tangible Eigen Values of the Austro-Hungarian Empire.
- Heh misusing eigenvalues as some kind of metaphor, but I can't complain about that! eigenvirtues
- one of his main contributions was "conversation theory" which is a formal theory of how two agents achieve a sort of negotiated mutual understanding (I think – don't really know it at all). blyg
- Please explain Ask Claude
- Here is a thorough explanation of Gordon Pask's Conversation Theory (CT):
Gordon Pask's Conversation Theory
The Core Idea
- Your intuition is correct.
- Conversation Theory (CT) is a cybernetic theory of how concept-forming and concept-sharing emerge out of conversational activity between two agents.
- More precisely,
- it is a cybernetic and epistemological account of concept-forming and concept-sharing through conversational discourse and practice.
- The central problem CT addresses is: **how do two agents — who start with different internal states and understandings — arrive at genuine shared understanding through exchange?**
- It is designed to help conceptualize and understand what takes place when effective communication occurs — the process of coming to know where one participant in a conversation can be said to understand another participant's "knowledge."
- Pask's key insight was that
- conversational learning is more than a general metaphor for humanly significant learning. He identified the essential minimal characteristics of the entities and relationships involved and formalized all that into a recursive learning theory of very broad scope.
The Mutual Understanding Problem
- A key starting point is that understanding is not simply the transfer of information. The two agents begin as what Pask calls "asynchronous" cognitive systems — they have different perspectives and conceptual structures.
- CT concerns itself with how two a priori asynchronous cognitive systems — either mechanical or organic in nature — are able to converse and conceptualize about a topic in such a way that each system achieves a posteriori synchronization of each cognitive system's particular perspective or attitude towards said topic.
- This resultant synchronization allows participants to satisfy some shared task or goal.
- Crucially, agreements can be illusory: each participant may assume a shared understanding that, when tested, turns out to be false. CT provides tools to actually verify whether real understanding has been established.
Learning as a Conversational Exchange
- CT originated from a cybernetics framework and attempts to explain learning in both living organisms and machines. The fundamental idea is that learning occurs through conversations about a subject matter which serve to make knowledge explicit.
- A key feature is that participants alternate roles:
- each participant takes turns within the interaction — switching at intervals between the role of a student or the role of a teacher — to achieve a state of mutual understanding between themselves.
- The critical method for verifying understanding is called "teachback": one person teaches another what they have learned. If you can successfully teach it back, understanding is demonstrated — not just claimed.
Levels of Language
- Conversations can be conducted at a number of different levels: natural language (general discussion), object languages (for discussing the subject matter), and metalanguages (for talking about learning/language itself).
- These are distinct:
- Natural language is for general discussion outside the formal experiment.
- Object language is for actually discussing and working through the subject matter under examination. Crucially, this language is a "command and question language" — not a declarative, assertoric language like formal logic. It prioritizes the semantic and pragmatic aspects of language over syntactic truth-values.
- Metalanguage is for discussing the design, management, and results of an experiment.
What Is a "Concept" in CT?
- Pask's notion of a concept is unusual and important.
- CT holds that a concept must be an inferential process that involves many other concepts — not a static representation.
- In other words, a concept is not a mental image or stored description — it is an ongoing process of producing, maintaining, and modulating a set of topic relations. You demonstrate you have a concept by being able to reconstruct and reproduce it, not just recall it.
Entailment Structures
- To facilitate learning, Pask argued that subject matter should be represented in structures called **entailment structures (ES)** — essentially graph-like maps showing what topics exist and how they depend on or derive from one another.
- The ES provides a sequence of the totality of topics that students can explore in a formal learning environment, defining what is called an "object language" L, which represents an expert's knowledge specific to a course or topic.
- This gives both teacher and learner a shared map of the conceptual terrain.
Learning Styles: Serialists vs. Holists
- From empirical work within CT, Pask identified two broad learning strategies:
- Serialists — progress through a structure in a sequential, step-by-step fashion.
- Holists — look for higher-order relations, seeking the big picture.
- An ideal learner balances both: broad knowledge with specialized depth.
Scope: Human–Human, Human–Machine, Machine–Machine
- CT provides a framework to simplify understandings of psychological mechanisms of human-machine, human-human, and machine-machine conversations in specific goal-oriented settings.
- Pask was explicitly not limiting the theory to human agents — machines count as conversational participants too, provided they can engage in the required exchange.
- Pask and his associates at System Research Ltd. attempted to examine the learning and intellectual development of conversational participants by means of human-machine interactions, thereby providing a psychological framework with educational technological applications.
Summary
- In short, Conversation Theory is a formal, cybernetic account of how two agents — starting from different internal states — can arrive at genuine, verifiable mutual understanding through structured conversational exchange. It treats concepts as active processes (not stored objects), uses layered languages (object language, metalanguage), insists on the teachback method to verify understanding, and applies equally to humans, machines, or any combination. It is simultaneously a theory of learning, a theory of knowledge, and a theory of communication.
- I want to know more about entailment structures, maybe with an example
- Ask Claude
- Here's a thorough breakdown of entailment structures in Pask's Conversation Theory:
What is an Entailment Structure?
- A body of knowledge or subject matter is conceived as consisting of topics related to one another. Two basic forms of relations between topics are distinguished: entailment relations (hierarchical) and relations of analogy (heterarchical). A static representation of such relations is called an entailment structure.
- Entailment structures reveal the "why" of knowledge — the conceptual structure of definitions and justifications that relate topics one to another.
- This contrasts with task structures, which show the "how" — the procedural, step-by-step execution side of a topic.
- More formally,
- in conversation theory, a concept is composed of other concepts in which each concept must entail other concepts.
- So entailment is not just a map of topics — it captures the inferential dependencies between them.
- Such structures are representational models in which the derivation of a concept from another concept is permitted — corresponding to the philosophical view that concepts are inter-defined.
Entailment Structure vs. Entailment Mesh
- There's an important distinction between the two terms:
- An **entailment structure (ES)** is the full space of possible topics.
- When the language is constrained, the user can operate upon the entailment structure of the "machine" (all possible concepts to be explored) and abstract their own interpretations. While an entailment mesh (EM) represents a subset of hierarchies and relationships between terms, the ES represents all possible topics that can be explored.
- An **entailment mesh (EM)** is the personalized structure that a learner actually traces through conversation.
- Because they are about the production and reproduction of concepts, entailment meshes provide a dynamic model of what it means to know something — not merely the retrieval of content from a database, but the kinetic recomputation of individualized knowing.
- Furthermore,
- if an entailment structure shows that topic A can be explained in terms of entailed topics B and C, then, in principle, topic B can be explained in terms of topics A and C — and so on. The resulting form is what Pask refers to as an "entailment mesh".
- In other words, the mesh is the richer, cyclically interconnected version.
Levels of Entailment Structures
- Entailment structures exist at a variety of different levels depending upon the extent of relationships displayed — for example, super/subordinate concepts, or analogies.
A Concrete Example: Thyroid Disease
- Pask discusses the application of conversation theory to a medical diagnosis task — diseases of the thyroid. The entailment structure in this case represents relationships between pathological conditions of the thyroid and their treatments/tests. The student is encouraged to learn these relationships by changing the parameter values of a variable (e.g., iodine intake level) and investigating the effects.
- So the entailment structure here might look like:
- Topic: Hypothyroidism → entails → low iodine, TSH levels, thyroxine treatment
- Topic: Iodine deficiency → entails → goitre, metabolic rate reduction
- Topic: TSH levels → entails → pituitary function, feedback regulation
- A student navigating this structure learns not just isolated facts, but why each topic connects to others — the justificatory web.
Serialists vs. Holists
- One practically important consequence:
- Pask identified two different types of learning strategies: serialists who progress through an entailment structure in a sequential fashion, and holists who look for higher-order relations.
- The entailment structure is what makes this distinction tractable — it's the map over which these different navigation styles play out.
Why It Matters for Conversation Theory
- The entailment structure is the shared object that makes genuine mutual understanding possible.
- A useful thing about domain models represented in an entailment structure is that you can record the excursions an individual makes by marking the nodes of the representation — a personal history of navigating through a domain. When the individual performs a further navigating action, a system can respond to that individual based on that history.
- This is how conversation becomes adaptive and personalized rather than just a broadcast of information.