Transformers Remember First, Forget Last: Dual-Process Interference in LLMs

TL;DR

This study investigates how large language models handle conflicting information, revealing a dominant proactive interference pattern that favors early memories over recent ones, contrasting human memory dynamics.

Contribution

It adapts cognitive psychology interference paradigms to analyze LLM memory behavior, uncovering separate mechanisms for RI and PI and their dependence on model size.

Findings

Proactive interference dominates retroactive interference in LLMs.

Model size predicts resistance to retroactive interference.

RI failures are passive, PI failures are active, with minimal hallucinations.

Abstract

When large language models encounter conflicting information in context, which memories survive -- early or recent? We adapt classical interference paradigms from cognitive psychology to answer this question, testing 39 LLMs across diverse architectures and scales. Every model shows the same pattern: proactive interference (PI) dominates retroactive interference (RI) universally (Cohen's d = 1.73, p < 0.0001), meaning early encodings are protected at the cost of recent information -- the opposite of human memory, where RI typically dominates. Three findings indicate that RI and PI reflect separate memory mechanisms. RI and PI are uncorrelated (R^2 = 0.044), rejecting a unified "memory capacity." Model size predicts RI resistance (R^2 = 0.49) but not PI (R^2 = 0.06, n.s.) -- only RI is capacity-dependent. And error analysis reveals distinct failure modes: RI failures are passive retrieval failures (51%), while PI failures show active primacy intrusion (56%); both show <1% hallucination. These patterns parallel the consolidation-retrieval distinction in cognitive science, suggesting that transformer attention creates a primacy bias with direct implications for interference-heavy applications.