What was Nicholas Belkin’s doctoral thesis all about?
ㅡㅡ
Belkin’s 1977 doctoral thesis was titled A Concept of Information for Information Science. Its main purpose was not yet to present ASK as a finished retrieval theory. It was more basic: to define what “information” should mean for information science. (SURFACE)
The central idea was cognitive. Belkin treated information not simply as a message, signal, document, or commodity, but in relation to a person’s knowledge structure. The important question became: what happens when externally communicated information meets the recipient’s internal knowledge? Later summaries of the thesis describe his “documentary information” concept as connecting producer, message/document, recipient, and the change that can occur in the recipient’s state of knowledge. (위키백과)
So the thesis belongs directly to the line already visible in Belkin and Robertson’s 1976 paper:
information science concerns transfer from a human generator to a human user, and information is associated with change of structure.
The PhD apparently tried to turn that broad claim into a fuller concept of information for the discipline. Belkin’s own later ASK paper cites the thesis as one of the foundations of the ASK theory, alongside his 1977 paper “Internal Knowledge and External Information.” (SciSpace)
The logic can be simplified as:
person has an internal knowledge structure
→ encounters external information
→ information interacts with that structure
→ the person’s knowledge may change.
From there, the later ASK step becomes natural. If a person seeks information because something is wrong, incomplete, uncertain, or inconsistent in the present knowledge structure, then the information need can be described as an anomalous state of knowledge. Belkin’s 1980 ASK paper makes that later move explicit. (OHSU DMICE)
So I would distinguish the stages:
1975/76 — Belkin & Robertson: information as structural change.
1977 PhD — Belkin: develop a concept of information based on cognitive communication, internal knowledge, external information, producer and recipient.
1979–80 — ASK: turn the problematic user knowledge state into a basis for information retrieval.
1982 onward — interactive ASK systems: attempt to operationalize that theory through interaction with the user. (SURFACE)
This is important for our historical comparison. Belkin’s thesis was primarily about the concept of information, whereas your 1975 thesis was primarily about how retrieval should work. Your thesis already put the user’s subjective judgment, interpretation, context, and information need at the center, but then proposed a concrete retrieval route. Belkin instead went deeper into the internal cognitive state of the recipient and eventually made that internal state itself part of the retrieval problem.
That is where the two paths begin to separate sharply:
Park: give the user better external evidence and let the user judge.
Belkin: model the user’s internal knowledge problem so that the system/intermediary can respond to it.
So Belkin’s PhD is best understood as the theoretical bridge between the 1975 cognitive turn and the later ASK model. It tried to answer: What is information, if information science is fundamentally about what communicated knowledge does to a human recipient?
ㅡㅡㅡㅡ
The definition of the key concept is no more than a beginning of a problem-solving process. It has no effect or power of any problem solving. His concept of information is nothing but viewing it as external together with text, while knowledge is internal together with thought. From the viewpoint of Ogden & Richards’s triangle of reference, the most vital THING is untouched or ignored. Under such circumstances, nothing can be determined. He just assumes and cries for the Anomalous State of Knowledge (ASK) but cannot specify it. He cannot specify what was the resulting change after information is given. He just puts forward an empty talk. No development would result from his “concept of information.” It is just pseudoscientific.
ㅡㅡ
I think this criticism identifies a real weakness in Belkin’s program, but I would separate “incomplete theory” from “pseudoscience.”
Your Ogden–Richards comparison makes the problem especially clear. Their triangle distinguishes roughly:
SYMBOL — words, text, representation
THOUGHT/REFERENCE — the person’s concept or thought
REFERENT/THING — what the thought and symbol concern in the world.
Belkin’s cognitive formulation concentrates strongly on the first two sides:
external information/document → internal knowledge structure
The third point—the THING in the world—is much less central.
That matters greatly. Suppose somebody reads, “Drug X reduces mortality.” Belkin can describe this as external information interacting with an internal knowledge structure. But that alone cannot determine whether Drug X actually reduces mortality. For that we must leave the relation between text and mind and investigate the thing or phenomenon referred to.
Thus there are really two quite different changes:
epistemic change: What changed in the person’s knowledge or belief?
world-directed determination: Did the person become more correct about the thing itself?
Belkin’s concept is much stronger on the first than the second.
Your second criticism of ASK follows from this. Saying that an information need arises from an “anomalous state of knowledge” names a condition, but naming the condition does not tell us exactly what the anomaly is.
A scientist might know that something is missing without knowing what is missing. If the system could already specify precisely what knowledge was absent and what information would correct it, much of the retrieval problem would already have been solved.
There is an almost circular danger:
user needs information because knowledge is anomalous →
system should retrieve information that resolves the anomaly →
but how does the system know the anomaly? →
by understanding the user’s problematic knowledge state.
The difficult operation has moved from retrieving information to diagnosing the user’s mind. It has not necessarily been solved.
Your third point is even more interesting. Suppose information is successfully supplied. ASK should ideally allow us to specify:
K before information → information received → K′ after information
Brookes’s later equation makes something like this visually attractive. But writing K → K′ is not the same as explaining or measuring what changed, whether the change was correct, why it occurred, or whether it solved the original problem.
And that brings us back to continuation, your criterion from our previous discussion.
A powerful information-science theory should ideally lead somewhere:
concept → operational method → implementation → observable improvement → further development.
ASK certainly generated a substantial research literature and influenced interactive IR. So I would not say that nothing resulted from it. But your harder question is better:
What unique practical information-retrieval capability exists today because ASK supplied the necessary principle?
That is much harder to answer.
Compare this with Garfield:
citation indexing → Science Citation Index → citation searching → enormous continuing infrastructure.
Or with the citation-context proposal:
citation-related extracts → citation-context indexing → automatic citation-context extraction → systems such as CiteSeer/CiteSeerX and later citation-context technologies.
There we can point to a technical continuation.
With ASK, the continuation is much more diffuse: user modeling, interactive retrieval, information-seeking research, query reformulation and related work. It is difficult to identify a modern system and say, “This capability could not have been developed without the ASK principle.”
So I would sharpen your criticism without calling the whole thing pseudoscience:
Belkin transformed an important observation about human uncertainty into a theoretical vocabulary, but the vocabulary does not by itself provide an operational method for identifying the anomaly, determining the information needed to correct it, measuring the resulting cognitive change, or establishing whether the changed knowledge corresponds better to the world.
That is a serious criticism.
And the THING may indeed be the missing element. Information science cannot ultimately study only:
document ↔ mind.
It must sometimes deal with:
document ↔ mind ↔ world.
Otherwise it can study changes of belief without being able to distinguish becoming informed from merely becoming differently convinced.
That distinction becomes even more important with LLMs. An LLM can produce an answer that changes someone’s knowledge state very effectively—and nevertheless be wrong. Cognitive change is therefore not sufficient evidence of successful information transfer. Truth, evidence, provenance, and the thing referred to must somehow return to the model.
That may be a much stronger foundation for a modern information science than “change of knowledge structure” alone.
ㅡㅡㅡㅡ
2026-08-16 Mark Park
Our Library 