The Machine's Internal Clock: Do LLMs Share Human Temporal Illusions?

2026-08-15

AI Summary

This study investigates whether large language models (LLMs) exhibit human-like temporal illusion perception by constructing a narrative benchmark and employing human-AI comparative analysis alongside chain-of-thought evaluation. The findings reveal that LLMs’ temporal judgments primarily rely on literature-based knowledge retrieval rather than intrinsic cognitive simulation. Although model performance aligns with theoretical predictions from existing literature, it significantly diverges from human behavioral patterns. This research elucidates the fundamental mechanisms underlying LLM temporal cognition, demonstrating that current models lack anthropomorphic perceptual biases. These results provide critical empirical evidence for understanding the cognitive boundaries of LLMs and inform future alignment research by highlighting the distinction between statistical text generation and genuine temporal perception.

Abstract

Human perception of time is subjective. Well-documented temporal illusions show that the brain relies on context and relational cues for judging duration instead of tracking elapsed time directly. Prior studies established these effects with visual and auditory stimuli. Existing LLM evaluations of temporal perception focus on estimating event durations or multi-step temporal reasoning. In this work, we investigate whether written narratives alone can evoke human temporal illusions, using a new benchmark of 6,684 narrative pairs spanning five illusions. We find that human readers (60 participants) prefer expected scenarios in only two of the five illusions, those where the manipulation is directly visible in text rather than requiring readers to internally simulate duration. We evaluate 14 LLMs on the same benchmark. Surprisingly, we find that models pick the literature-predicted scenario across four of the five illusions, diverging from human behavior. Reasoning traces show that ~70% of responses explicitly evoke psychology research, suggesting that this alignment is consistent with retrieval of published findings rather than human-like temporal biases.

Problem

• Temporal Illusions
• Large Language Models
• Time Perception
• Narrative Understanding

Innovation

• Temporal Illusions
• Narrative Benchmark
• Large Language Models
• Knowledge Retrieval
• Human-AI Divergence