INQUIRING LINE

Why does comparison reasoning generalize better than composition reasoning?

This explores why tasks that ask a model to compare or contrast two things tend to hold up outside training data, while tasks that chain several reasoning steps together (composition) tend to break — and the corpus speaks more directly to why composition fails than to why comparison is robust, so the answer is partly read off that asymmetry.


This explores why comparison-style reasoning survives distribution shift better than multi-step compositional reasoning. The honest starting point: the collection has a lot to say about why composition collapses, and comparison's relative durability falls out of that as the mirror image. The single sharpest finding is that transformers don't actually compose — they reduce compositional reasoning to memorizing computation subgraphs from training and matching against them, then fail drastically on novel combinations Do transformers actually learn systematic compositional reasoning?. The killer detail is that errors *compound across reasoning steps*. Composition is a chain, and every link is a fresh chance for an unfamiliar instance to derail the whole thing.

That compounding is the heart of it. Reasoning failures, it turns out, aren't triggered by abstract task complexity but by *instance-level unfamiliarity* — models fit patterns tied to specific examples rather than general algorithms, so a chain only succeeds when each of its steps resembles something seen before Do language models fail at reasoning due to complexity or novelty?. A comparison is typically one shallow operation over two items; a composition is many operations stacked. If each step has some independent chance of hitting an unfamiliar pattern, the probability of getting the *whole chain* right decays multiplicatively with depth — which is exactly the exponential drop-off in success that shows up when reasoning models are pushed to deeper problems Why do reasoning LLMs fail at deeper problem solving?. Comparison generalizes better not because it's a smarter kind of thinking, but because it's *shorter*: fewer links, fewer places to break.

The same pattern explains why chain-of-thought, the workhorse of composition, is so fragile. CoT reproduces the *form* of reasoning through pattern-matching against familiar schemata rather than performing genuine inference, so it degrades predictably the moment the task, length, or format drifts from training Does chain-of-thought reasoning reveal genuine inference or pattern matching? Does chain-of-thought reasoning actually generalize beyond training data?. A model imitating reasoning form can fake a short comparison convincingly; a long composition exposes the imitation because the fluent-but-invalid logic has to stay coherent across many dependent steps What makes chain-of-thought reasoning actually work?.

There's a wrinkle worth knowing: some of what looks like a 'composition can't generalize' wall is actually an *execution* ceiling, not a reasoning one. Models often know the algorithm but can't run it across many steps in text-only generation — give them tools and they cross the supposed cliff Are reasoning model collapses really failures of reasoning?. That reframes the question: composition may not be inherently less generalizable than comparison so much as it makes far heavier demands on a brittle step-by-step execution channel, where comparison barely taxes it at all.

The constructive flip side is that you can claw some of composition's generalization back by changing its structure. Selectively augmenting natural language with symbolic scaffolding — not full formalization, just enough to pin down the structure between steps — beats both pure language and pure logic, because it gives the chain a backbone that doesn't depend on having memorized the exact instance Why does partial formalization outperform full symbolic logic?. The lesson hiding in all of this: composition isn't doomed, but its generalization is bottlenecked by chain length and execution, and the fixes are about shortening, structuring, or offloading the chain — not about thinking harder.


Sources 8 notes

Do transformers actually learn systematic compositional reasoning?

Research shows transformers succeed on in-distribution tasks by memorizing computation subgraphs from training data, not by learning systematic rules. They fail drastically on novel compositions, with errors compounding across reasoning steps.

Do language models fail at reasoning due to complexity or novelty?

LRMs don't break at complexity thresholds but at instance-novelty boundaries. Models fit instance-based patterns rather than generalizable algorithms, so any reasoning chain succeeds if trained on similar instances, regardless of length.

Why do reasoning LLMs fail at deeper problem solving?

Current reasoning models lack the three properties of systematic exploration: validity, effectiveness, and necessity. This causes success probability to drop exponentially with problem depth, making medium problems solvable but deep problems catastrophically harder.

Does chain-of-thought reasoning reveal genuine inference or pattern matching?

CoT works by constraining models to reproduce familiar reasoning patterns from training, not by enabling novel symbolic reasoning. Performance degrades predictably under distribution shifts—the signature of imitation rather than capability emergence.

Does chain-of-thought reasoning actually generalize beyond training data?

DataAlchemy experiments show CoT fails systematically under distributional shifts in task, length, and format. Models produce fluent but logically inconsistent reasoning — imitating reasoning form without valid underlying logic.

What makes chain-of-thought reasoning actually work?

CoT systems reproduce the form of reasoning through pattern matching rather than performing genuine logical inference. This explains why format effects dominate content, why structurally invalid prompts succeed, and why stronger reasoning models become less instruction-compliant.

Are reasoning model collapses really failures of reasoning?

Models confined to text-only generation cannot execute multi-step procedures at scale, even when they know the underlying algorithm. Tool-enabled models solve problems beyond the supposed reasoning cliff, suggesting the bottleneck is procedural execution bandwidth.

Why does partial formalization outperform full symbolic logic?

QuaSAR and Logic-of-Thought both achieve 4-8% accuracy gains by enriching natural language with selective symbolic elements rather than replacing it. Full formalization loses semantic information; pure language lacks structure. Augmentation preserves both.

Research prompt for your LLMexpand ↓

Copy into ChatGPT or Claude to take this line of inquiry further — it asks the model to find newer work and re-test which earlier constraints still hold.

You are a reasoning researcher auditing claims about why comparison reasoning outlasts composition reasoning under distribution shift. The question remains open: is this a fundamental property of how transformers work, or has it been partially resolved by new model capabilities, training methods, or execution infrastructure?

What a curated library found — and when (dated claims, not current truth):
Findings span 2023–2026; treat these as dated snapshots:
• Transformers reduce composition to memorizing computation subgraphs; errors compound multiplicatively across reasoning steps, causing exponential drop-off in success on deeper problems (~2025).
• Reasoning failures are driven by instance-level unfamiliarity, not task-level complexity; CoT reproduces the *form* of reasoning via pattern-matching against familiar schemata rather than performing genuine inference (~2025–2026).
• Chain-of-thought degrades predictably when task, length, or format drifts from training; a model imitating reasoning form can fake a short comparison but exposure happens at composition length (~2025–2026).
• Some apparent composition failures are *execution* ceilings, not reasoning ones; augmenting with symbolic scaffolding (partial formalization) recovers generalization without full logic (~2025).

Anchor papers (verify; mind their dates):
• arXiv:2305.18654 (2023-05) Faith and Fate: Limits of Transformers on Compositionality
• arXiv:2502.12616 (2025-02) Improving Chain-of-Thought Reasoning via Quasi-Symbolic Abstractions
• arXiv:2506.02878 (2025-06) CoT is Not True Reasoning, It Is Just a Tight Constraint to Imitate
• arXiv:2602.06176 (2026-02) Large Language Model Reasoning Failures

Your task:
(1) RE-TEST EACH CONSTRAINT by asking: Have newer models (o3, GPT-4-Turbo-level or beyond), training methods (process reward modeling, RL on multi-step scaffolding), or orchestration (tree search, caching intermediate symbolic representations, multi-agent composition) actually *relaxed* the exponential decay in composition depth or the necessity of instance-level memorization? Cite what resolves or still upholds each claim; separate durable question (Why does chaining degrade?) from perishable limitation (Can't do >5 steps).
(2) Surface the strongest *contradicting* or *superseding* work from the last 6 months (post-2510 if available). Does any paper show composition generalizing as well as comparison under held-out distribution, or demonstrate that CoT is *not* mere imitation?
(3) Propose 2 research questions assuming the regime may have shifted: (a) Under what model scale, training method, or execution mode does composition's generalization curve flatten? (b) Is the comparison-vs.-composition gap fundamentally about *chain length* vs. *reasoning capacity*, and can you decouple them?

Cite arXiv IDs; flag anything you cannot ground in a real paper.

Next inquiring lines