Bound by the past: Serial dependence in multisensory perception across domains and modalities

Serial dependence refers to the tendency of perceptual judgments to be systematically biased toward or away from recently encountered stimuli. This mechanism is believed to promote perceptual stability in a dynamic and often noisy environment. Over the past decade, serial dependence has become a central topic in perceptual neuroscience, with extensive evidence demonstrating its influence on a wide range of visual features, from low-level attributes such as orientation and position to higher-level judgments such as attractiveness. However, comparatively little is known about how serial dependence operates in temporal perception and across sensory modalities. The primary objective of this doctoral project was to investigate whether serial dependence extends to the perception of temporal duration in vision and audition, to examine its manifestation in tactile perception, and to assess the role of visual experience in shaping multisensory serial dependence. In particular, this work aimed to determine whether serial dependence reflects a general computational principle of perception or whether it is modulated by modality-specific constraints and sensory history. To address these questions, a series of psychophysical experiments was conducted using discrimination and reproduction paradigms. Serial dependence was examined in visual and auditory temporal perception, as well as in tactile motion perception. To investigate the contribution of visual experience, experiments were performed both in sighted participants and in individuals with visual impairment. In addition, mixed reality technology was adopted to enable precise experimental control and to isolate visual contributions in visuo-haptic tasks, allowing the study of serial dependence in controlled multisensory environments. The results provide converging evidence that serial dependence extends beyond classical visual features to temporal perception in both vision and audition, although its expression depends on temporal context and task demands. In auditory temporal perception, serial dependence emerged under specific temporal conditions, suggesting a dependence on the temporal proximity of successive stimuli. In the tactile domain, serial dependence effects were more selective and appeared to depend on visual experience, as they were absent in visually impaired participants. Multisensory experiments further indicated that vision plays a critical role in shaping serial dependence in haptic judgments, highlighting the importance of cross-modal calibration processes. Overall, these findings suggest that serial dependence is a flexible perceptual mechanism whose expression varies across sensory modalities and depends on both stimulus characteristics and sensory experience. By demonstrating modality-specific constraints and a critical role of visual experience in multisensory serial dependence, the findings provide new insights into the computational principles underlying perceptual stability. Additionally, these results have direct implications for the design of human-robot interaction systems, multisensory interfaces, and rehabilitation technologies, where adaptive integration of sensory information over time is essential. Moreover, the experimental paradigms and mixed-reality methodologies developed in this project offer robust tools for studying perception in controlled multisensory environments, with potential applications in device calibration, sensory substitution, and computational models of human perception.