User-side memory in LLMs is typically scored as a single "personalization" capability: given a user's history, is the output more user-aware? We show this aggregate metric hides opposite-direction failures. Memory factorises into at least three orthogonal axes -- behavioral consistency (style, voice), factual presence (recall facts in history), and factual absence (abstain when a fact is absent) -- and no single substrate wins all three. Comparing per-user gamma-LoRA (a small LoRA adapter trained on each user's history; gamma denotes per-user, not per-task) against BGE-large dense top-K retrieval on a controlled 50-user synthetic corpus and a real-data probe (LaMP-3), we find gamma-LoRA decisively wins behavioral style while RAG decisively wins factual absence -- and the same query-projection cells in attention layers 21-35 causally load-bear both effects in opposite directions (zeroing those LoRA weights raises absence-probe TPR by +33 pp and drops presence-probe TPR by 20 pp). On the more heavily RLHF-tuned Llama-3.1-8B-Instruct the asymmetry strengthens, not heals: parametric memory's behavioral advantage collapses while its absence-calibration deficit against retrieval widens -- an alignment tax on parametric user-memory. On real-data LaMP-3, gamma-LoRA underperforms a majority baseline; a 9-condition mitigation sweep diagnoses this as instruction-following collapse, not substrate failure (a 9x2 cross-product shows the eval-time {1..5} logit mask drives main_acc to >=0.995 on every recipe), and the best training-time fix replicates bit-identically on Llama. Finally, substrate-selection routing is question-classification, not calibration: a 110M DistilBERT on the question text alone beats every logit-based router. We contribute the diagnostic framework, the diagnosed real-data negative, the alignment-tax replication, and the routing-as-classification finding.

翻译：大型语言模型（LLM）中的用户侧记忆通常被统称为单一的“个性化”能力：给定用户的历史记录，输出是否更具用户感知？我们证明这一总体指标掩盖了方向相反的故障。记忆至少可分解为三个正交轴——行为一致性（风格、语态）、事实存在性（回忆历史中的事实）与事实缺失性（无事实时弃权）——且没有任何一种底物能在所有维度上胜出。通过比较每个用户的gamma-LoRA（一种针对每个用户历史训练的微小LoRA适配器；gamma表示按用户而非按任务）与BGE-large密集top-K检索，在受控的50用户合成语料库及真实数据探针（LaMP-3）上，我们发现gamma-LoRA在行为风格上取得决定性优势，而RAG在事实缺失性上取得决定性优势——且注意力层21-35中相同的查询投影细胞以相反方向因果承载这两种效应（将LoRA权重归零可使缺失探针TPR提高33个百分点，同时使存在探针TPR降低20个百分点）。在更依赖RLHF调优的Llama-3.1-8B-Instruct上，这种不对称性不仅未缓解反而加剧：参数化记忆的行为优势崩溃，同时其缺失校准能力相对于检索的缺陷扩大——这体现了参数化用户记忆的对齐税。在真实数据LaMP-3上，gamma-LoRA性能低于多数基线；通过9种条件缓解扫描，我们诊断出这是由于指令遵循能力崩溃而非底物失效（9×2交叉实验显示，评估时的{1..5}逻辑掩码使每种方案的主准确率均≥0.995），且最佳训练时修复方案在Llama上实现了比特级复制。最后，底物选择路由本质上是问题分类而非校准：基于问题文本的110M DistilBERT优于所有基于逻辑的路由器。我们贡献了诊断框架、诊断出的真实数据负例、对齐税复现结果以及路由即分类的发现。