SOLO transition structures and the resolution of adaptation
How fine a change in demonstrated reasoning can a tutor respond to?
From broad categories to controlled contrasts
The SOLO taxonomy concerns the structure of observed learning outcomes. Here it supplies a starting vocabulary for designing contrasts, rather than a license to infer a learner’s complete understanding from one explanation. See Biggs and Collis.
The project proposes to compare explanations that identify no relevant structure, one relevant element, several separate elements, or relationships between them. The intermediate 1A, 2A, and 3A labels need explicit operational definitions within the study.
A working idea: adaptation resolution
Suppose a tutor responds differently to a prestructural explanation and a relational explanation. That leaves open whether it responds to finer distinctions in between.
I am using resolution of adaptation as a working term for the granularity of demonstrated-reasoning contrasts to which a tutor shows a meaningful behavioral response under a specified protocol. It is not yet a validated scalar property of a model.
The distance function is intentionally unspecified here. A lexical distance would answer a different question from an annotation of scaffolding or diagnostic strategy.
The difficult confounds
Longer explanations often contain more information. More connected explanations can also be more correct. If either varies with the reasoning condition, a response difference may not isolate the intended contrast.
The structure versus correctness note makes this separation explicit. The solution-leakage note explains why withholding a repair does not settle the adaptation question.
What to specify next
Construct paired explanations, audit the task information available in each, define annotation dimensions, and examine agreement before calling a contrast meaningful. Broad labels should not hide unresolved construct validity.
References & reading
Connected thinking.
Related reading 4
When relational-looking output no longer implies relational understanding
Adaptive AI tutoringWhy solution leakage is not enough to measure adaptation
Adaptive AI tutoringStructure is not the same as factual correctness
Adaptive AI tutoringSame Error, Different Learner
Adaptive AI tutoringLinked from 3
Backlinks from declared content relationships.
NoteWhy solution leakage is not enough to measure adaptation ↗NoteStructure is not the same as factual correctness ↗ProjectSame Error, Different Learner ↗