[P1] Poster: Day 1

[P1-22] How ensemble temporal statistics influence duration perception of visual events

*Valeria Centanino¹, Gianfranco Fortunato¹, Domenica Bueti¹ (1. International School for Advanced Studies (SISSA) (Italy))

Keywords:

temporal context,duration tuning,7T-fMRI

The human ability to reproduce the duration of brief sensory events is shaped by the statistical distribution of recently experienced durations, referred to as the temporal context. For example, when the same physical duration is presented within different duration ranges, its reproduction tends to be systematically biased toward the mean of the respective range, leading to different reproductions across contexts. Temporal context also changes when we are exposed to a fixed set of durations that vary in their frequency of occurrence, though the effects of such context remain less well understood. At the neural level, functional MRI (fMRI) studies have shown that the processing of brief visual durations is supported by tuning mechanisms that change across the cortical hierarchy—from monotonic tuning in early visual areas to unimodal tuning in downstream regions. However, it remains unclear how and where these tuning properties adapt to contextual biases. In this study, 30 participants reproduced 8 visual durations presented under either a uniform or a positively skewed distribution. To investigate the neural underpinnings of this contextual manipulation, a separate group of 15 participants performed the same task while undergoing ultra-high-field (7T) fMRI. Behavioral data showed that, under the skewed condition, all durations were reproduced as longer, suggesting a repulsive effect of temporal statistics on behavioral responses. Representational similarity analysis further revealed a systematic forward shift in reproductions: responses under the skewed condition became more similar to those of the next longer duration in the uniform condition, indicating a fine-grained adjustment of timing performance driven by temporal statistics. For the neural data, we plan to use neuronal model-based analysis to estimate monotonic and unimodal responses to durations. This approach will be instrumental in characterizing tuning differences between statistical conditions and linking them to behavioral outcomes. Overall, this work may advance our understanding of the neural mechanisms underlying context-driven temporal distortions.