Assessing Generalization in Connectionist and Rule-Based Models Under the Learning Constraint
Bibliographic record
Abstract
Although it is commonly assumed that rule-based models generalize more effectively than do connectionist models, the comparison is often confounded by pitting hand-written rules against learned connections. Three case studies from cognitive development show that, under the constraint that both types of models learn their representations from equivalent examples, generalization is consistently superior in connectionist models. Generalization Problems A significant part of the ongoing debate between rulebased and connectionist modeling in psychology has focused on the ability to generalize. A common claim from supporters of the classical, symbolic approach is that rule-based models are superior because they generalize more effectively than do connectionist models (Ling & Marinov, 1993; Pinker, 1997; Marcus, 1998). Generalization is considered important by most modelers because it distinguishes understanding of a problem from mere memorization of solutions. The generalization ability of rules is often enhanced by the use of variables that can be bound to any number of objects or events. Consider the following rule, written in Common Lisp for a production system program. It generates correct responses on some Piagetian conservation of number problems: ((response more?x?y)
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.017 | 0.120 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.003 |
| Scholarly communication | 0.002 | 0.012 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.003 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".