Expecting the Future: Entropy-Based Predictions in Time Perception

Abstract

Abstract What cognitive mechanisms underlie the human experience of time’s directionality, often described as the psychological arrow of time? This study explores this question by examining the role of visual entropy, a measure of disorder within a system, in shaping our perception of time. Grounded in the second law of thermodynamics and predictive processing frameworks, we hypothesised that the human cognitive system expects an increase in visual entropy over time and that violations of this expectation (prediction errors) would lead to a relative overestimation of perceived duration. Participants reproduced the duration of computer-generated animations depicting three-dimensional greyscale shapes shattering from wholes into parts (low to high entropy) or the reverse (high to low entropy) presented at different lengths (500, 1000, 1500, and 2000 milliseconds). The results revealed a significant interaction: while forward animations (low to high entropy) were perceived as longer at the shortest duration (500 ms), backward animations (high to low entropy) were overestimated at longer durations (1500 and 2000 ms). This pattern suggests an initial reliance on high-precision expectations for shorter stimuli, with a shift towards increased weighting of prediction errors for longer durations, leading to the overestimation of entropy-reversed stimuli. These results align with predictive processing frameworks, indicating that the brain dynamically reconciles incoming sensory data with prior expectations of entropy progression, thereby shaping our subjective experience of temporal directionality. This research introduces a novel paradigm for examining time perception and supports the application of predictive processing theories in understanding the cognitive and neural underpinnings of temporal experience.