ΔΦ COGNITIVE RESEARCH TOOL v5.0 Transient Reductions in Short-Term Temporal Dependence: A Cross-Task Behavioral Signature

This study introduces a behavioral analysis framework for detecting transient changes in short-term temporal dependence in human decision-making. Using multiple independent datasets spanning conflict (Stroop, Flanker), memory (Sternberg), and reward (Bandit) paradigms, we apply an autocorrelation-based detection method (Keller-ΔΦ Protocol) to reaction time sequences.Across all tasks, we identify a consistent and domain-general signal: brief reductions in lag-1 temporal predictability, indicating a temporary decrease in dependence on the immediately preceding state. These events occur independently of errors, are robust across detection thresholds, and do not involve changes in mean behavioral amplitude.Post-event analysis shows small but reliable performance costs, including slight increases in reaction time and, in memory tasks, reduced accuracy. No evidence is found for persistent behavioral change, cumulative improvement, or parameter drift, ruling out learning or consolidation interpretations.Instead, the results support a control-based interpretation: the cognitive system intermittently reduces short-term temporal coupling, temporarily disrupting performance before resuming baseline operation. This phenomenon is observable purely at the behavioral level and does not rely on physiological measurement.The ΔΦ framework (ΔΦ = ρ × v) is used strictly as an analytical lens to describe temporal structure in behavioral data, not as a mechanistic claim about neural implementation.All datasets are publicly available, and the analysis pipeline is reproducible.